[ { "paper_id": "1345", "title": "Three Generative, Lexicalised Models for Statistical Parsing", "abstract": "In this paper we first propose a new statistical parsing model, which is a generative model of lexicalised context-free grammar. We then extend the model to include a probabilistic treatment of both subcategorisation and wh-movement. Results on Wall Street Journal text show that the parser performs at 88.1/87.5% constituent precision/recall, an average improvement of 2.3% over (Collins 96).", "classified_sentences": [ { "sentence": "In this paper we first propose a new statistical parsing model, which is a generative model of lexicalised context-free grammar.", "category": "method" }, { "sentence": "We then extend the model to include a probabilistic treatment of both subcategorisation and wh-movement.", "category": "method" }, { "sentence": "Results on Wall Street Journal text show that the parser performs at 88.1/87.5% constituent precision/recall, an average improvement of 2.3% over (Collins 96).", "category": "result" } ] }, { "paper_id": "14223", "title": "Grammar as a Foreign Language", "abstract": "Syntactic constituency parsing is a fundamental problem in natural language processing and has been the subject of intensive research and engineering for decades. As a result, the most accurate parsers are domain specific, complex, and inefficient. In this paper we show that the domain agnostic attention-enhanced sequence-to-sequence model achieves state-of-the-art results on the most widely used syntactic constituency parsing dataset, when trained on a large synthetic corpus that was annotated using existing parsers. It also matches the performance of standard parsers when trained only on a small human-annotated dataset, which shows that this model is highly data-efficient, in contrast to sequence-to-sequence models without the attention mechanism. Our parser is also fast, processing over a hundred sentences per second with an unoptimized CPU implementation.", "classified_sentences": [ { "sentence": "Syntactic constituency parsing is a fundamental problem in natural language processing and has been the subject of intensive research and engineering for decades.", "category": "background" }, { "sentence": "As a result, the most accurate parsers are domain specific, complex, and inefficient.", "category": "background" }, { "sentence": "In this paper we show that the domain agnostic attention-enhanced sequence-to-sequence model achieves state-of-the-art results on the most widely used syntactic constituency parsing dataset, when trained on a large synthetic corpus that was annotated using existing parsers.", "category": "result" }, { "sentence": "It also matches the performance of standard parsers when trained only on a small human-annotated dataset, which shows that this model is highly data-efficient, in contrast to sequence-to-sequence models without the attention mechanism.", "category": "result" }, { "sentence": "Our parser is also fast, processing over a hundred sentences per second with an unoptimized CPU implementation.", "category": "result" } ] }, { "paper_id": "118556", "title": "The Hearsay-II Speech-Understanding System: Integrating Knowledge to Resolve Uncertainty", "abstract": "The Hearsay-II system, developed during the DARPA-sponsored five-year speech-understanding research program, represents both a specific solution to the speech-understanding problem and a general framework for coordinating independent processes to achieve cooperative problem-solving behavior. As a computational problem, speech understanding reflects a large number of intrinsically interesting issues. Spoken sounds are achieved by a long chain of successive transformations, from intentions, through semantic and syntactic structuring, to the eventually resulting audible acoustic waves. As a consequence, interpreting speech means effectively inverting these transformations to recover the speaker's intention from the sound. At each step in the interpretive process, ambiguity and uncertainty arise. The Hearsay-II problem-solving framework reconstructs an intention from hypothetical interpretations formulated at various levels of abstraction. In addition, it allocates limited processing resources first to the most promising incremental actions. The final configuration of the Hearsay-II system comprises problem-solving components to generate and evaluate speech hypotheses, and a focus-of-control mechanism to identify potential actions of greatest value. Many of these specific procedures reveal novel approaches to speech problems. Most important, the system successfully integrates and coordinates all of these independent activities to resolve uncertainty and control combinatorics. Several adaptations of the Hearsay-II framework have already been undertaken in other problem domains, and it is anticipated that this trend will continue; many future systems necessarily will integrate diverse sources of knowledge to solve complex problems cooperatively. Discussed in this paper are the characteristics of the speech problem in particular, the special kinds of problem-solving uncertainty in that domain, the structure of the Hearsay-II system developed to cope with that uncertainty, and the relationship between Hearsay-II's structure and those of other speech-understanding systems. The paper is intended for the general computer science audience and presupposes no speech or artificial intelligence background.", "classified_sentences": [ { "sentence": "The Hearsay-II system, developed during the DARPA-sponsored five-year speech-understanding research program, represents both a specific solution to the speech-understanding problem and a general framework for coordinating independent processes to achieve cooperative problem-solving behavior.", "category": "method" }, { "sentence": "As a computational problem, speech understanding reflects a large number of intrinsically interesting issues.", "category": "background" }, { "sentence": "Spoken sounds are achieved by a long chain of successive transformations, from intentions, through semantic and syntactic structuring, to the eventually resulting audible acoustic waves.", "category": "background" }, { "sentence": "As a consequence, interpreting speech means effectively inverting these transformations to recover the speaker's intention from the sound.", "category": "background" }, { "sentence": "At each step in the interpretive process, ambiguity and uncertainty arise.", "category": "background" }, { "sentence": "The Hearsay-II problem-solving framework reconstructs an intention from hypothetical interpretations formulated at various levels of abstraction.", "category": "method" }, { "sentence": "In addition, it allocates limited processing resources first to the most promising incremental actions.", "category": "method" }, { "sentence": "The final configuration of the Hearsay-II system comprises problem-solving components to generate and evaluate speech hypotheses, and a focus-of-control mechanism to identify potential actions of greatest value.", "category": "method" }, { "sentence": "Many of these specific procedures reveal novel approaches to speech problems.", "category": "result" }, { "sentence": "Most important, the system successfully integrates and coordinates all of these independent activities to resolve uncertainty and control combinatorics.", "category": "result" }, { "sentence": "Several adaptations of the Hearsay-II framework have already been undertaken in other problem domains, and it is anticipated that this trend will continue; many future systems necessarily will integrate diverse sources of knowledge to solve complex problems cooperatively.", "category": "result" }, { "sentence": "Discussed in this paper are the characteristics of the speech problem in particular, the special kinds of problem-solving uncertainty in that domain, the structure of the Hearsay-II system developed to cope with that uncertainty, and the relationship between Hearsay-II's structure and those of other speech-understanding systems.", "category": "background" }, { "sentence": "The paper is intended for the general computer science audience and presupposes no speech or artificial intelligence background.", "category": "background" } ] }, { "paper_id": "149368", "title": "Exploiting affinity propagation for automatic acquisition of domain concept in ontology learning", "abstract": "Semantic Web uses domain ontology to bridge the gap among the members of a domain through minimization of conceptual and terminological incompatibilities. However, several barriers must be overcome before domain ontology becomes a practical and useful tool. One important issue is identification and selection of domain concepts for domain ontology learning when several hundreds or even thousands of terms are extracted and available from relevant text documents shared among the members of a domain. We present a novel domain concept acquisition and selection approach for ontology learning that uses affinity propagation algorithm, which takes as input semantic and structural similarity between pairs of extracted terms called data points. Real-valued messages are passed between data points (terms) until high quality set of exemplars (concepts) and cluster iteratively emerges. All exemplars will be considered as domain concepts for learning domain ontologies. Our empirical results show that our approach achieves high precision and recall in selection of domain concepts using less number of iterations.", "classified_sentences": [ { "sentence": "Semantic Web uses domain ontology to bridge the gap among the members of a domain through minimization of conceptual and terminological incompatibilities.", "category": "background" }, { "sentence": "However, several barriers must be overcome before domain ontology becomes a practical and useful tool.", "category": "background" }, { "sentence": "One important issue is identification and selection of domain concepts for domain ontology learning when several hundreds or even thousands of terms are extracted and available from relevant text documents shared among the members of a domain.", "category": "background" }, { "sentence": "We present a novel domain concept acquisition and selection approach for ontology learning that uses affinity propagation algorithm, which takes as input semantic and structural similarity between pairs of extracted terms called data points.", "category": "method" }, { "sentence": "Real-valued messages are passed between data points (terms) until high quality set of exemplars (concepts) and cluster iteratively emerges.", "category": "method" }, { "sentence": "All exemplars will be considered as domain concepts for learning domain ontologies.", "category": "method" }, { "sentence": "Our empirical results show that our approach achieves high precision and recall in selection of domain concepts using less number of iterations.", "category": "result" } ] }, { "paper_id": "281719", "title": "Selective Deep Convolutional Features for Image Retrieval", "abstract": "Convolutional Neural Network (CNN) is a very powerful approach to extract discriminative local descriptors for effective image search. Recent work adopts fine-tuned strategies to further improve the discriminative power of the descriptors. Taking a different approach, in this paper, we propose a novel framework to achieve competitive retrieval performance. Firstly, we propose various masking schemes, namely SIFT-mask, SUM-mask, and MAX-mask, to select a representative subset of local convolutional features and remove a large number of redundant features. We demonstrate that this can effectively address the burstiness issue and improve retrieval accuracy. Secondly, we propose to employ recent embedding and aggregating methods to further enhance feature discriminability. Extensive experiments demonstrate that our proposed framework achieves state-of-the-art retrieval accuracy.", "classified_sentences": [ { "sentence": "Convolutional Neural Network (CNN) is a very powerful approach to extract discriminative local descriptors for effective image search.", "category": "background" }, { "sentence": "Recent work adopts fine-tuned strategies to further improve the discriminative power of the descriptors.", "category": "background" }, { "sentence": "Taking a different approach, in this paper, we propose a novel framework to achieve competitive retrieval performance.", "category": "method" }, { "sentence": "Firstly, we propose various masking schemes, namely SIFT-mask, SUM-mask, and MAX-mask, to select a representative subset of local convolutional features and remove a large number of redundant features.", "category": "method" }, { "sentence": "We demonstrate that this can effectively address the burstiness issue and improve retrieval accuracy.", "category": "result" }, { "sentence": "Secondly, we propose to employ recent embedding and aggregating methods to further enhance feature discriminability.", "category": "method" }, { "sentence": "Extensive experiments demonstrate that our proposed framework achieves state-of-the-art retrieval accuracy.", "category": "result" } ] }, { "paper_id": "297308", "title": "Correlation of psychooncological distress- screening and quality of life assessment in neurosurgical patients", "abstract": "Background Cerebral tumors are associated with high rates of anxiety, depression and reduced health related quality of life. But still psychooncological screening instruments are not implemented in the daily routine of neurosurgical departments. In contrast the EORTC QLQ-C30/ EORTC QLQ- BN20 questionnaire is often used to evaluate quality of life in the framework of clinical studies. We were therefore interested, if conspicuous distress screening results are also reflected by HRQOL assessment. Patients and Methods Patients who were electively admitted for surgery of intracranial lesions were screened for their psychooncological distress using two self-assessment instruments (Hospital Anxiety and Depression Scale (HADS) and Distress Thermometer (DT)) and one external assessment questionnaire (Psychooncological base documentation (PO-Bado). Results were correlated with three subscales of the EORTC-QLQ-C30 and EORTC-QLQ-BN20 questionnaire. Results From October 2013 to March 2015, 594 patients were admitted for elective cranial neurosurgical procedure. 489 neurosurgical patients were screened for increased distress. Data from 450 patients could be correlated with the EORTC-QLQ-C30 and EORTC-QLQ-BN20. In 265 patients screening revealed increased distress. A concurrent reduced global health /higher rates of future uncertainty and conspicuous distress screening results are found in 173 patients (69.5%) compared to 30.5% of patients (n= 76) with unremarkable screening. Increased distress screening was highly significant with increased level of future uncertainty as well as decreased level of quality of life and global health (p<0.0001). Conclusion Psychooncological distress is accompanied by reduced quality of life, global heath and increased future uncertainty. Therefore HQOL assessment can be helpful identifying patients with increased distress.", "classified_sentences": [ { "sentence": "Background Cerebral tumors are associated with high rates of anxiety, depression and reduced health related quality of life.", "category": "background" }, { "sentence": "But still psychooncological screening instruments are not implemented in the daily routine of neurosurgical departments.", "category": "background" }, { "sentence": "In contrast the EORTC QLQ-C30/ EORTC QLQ- BN20 questionnaire is often used to evaluate quality of life in the framework of clinical studies.", "category": "background" }, { "sentence": "We were therefore interested, if conspicuous distress screening results are also reflected by HRQOL assessment.", "category": "method" }, { "sentence": "Patients and Methods Patients who were electively admitted for surgery of intracranial lesions were screened for their psychooncological distress using two self-assessment instruments (Hospital Anxiety and Depression Scale (HADS) and Distress Thermometer (DT)) and one external assessment questionnaire (Psychooncological base documentation (PO-Bado).", "category": "method" }, { "sentence": "Results were correlated with three subscales of the EORTC-QLQ-C30 and EORTC-QLQ-BN20 questionnaire.", "category": "method" }, { "sentence": "Results From October 2013 to March 2015, 594 patients were admitted for elective cranial neurosurgical procedure.", "category": "result" }, { "sentence": "489 neurosurgical patients were screened for increased distress.", "category": "result" }, { "sentence": "Data from 450 patients could be correlated with the EORTC-QLQ-C30 and EORTC-QLQ-BN20.", "category": "result" }, { "sentence": "In 265 patients screening revealed increased distress.", "category": "result" }, { "sentence": "A concurrent reduced global health /higher rates of future uncertainty and conspicuous distress screening results are found in 173 patients (69.5%) compared to 30.5% of patients (n= 76) with unremarkable screening.", "category": "result" }, { "sentence": "Increased distress screening was highly significant with increased level of future uncertainty as well as decreased level of quality of life and global health (p<0.0001).", "category": "result" }, { "sentence": "Conclusion Psychooncological distress is accompanied by reduced quality of life, global heath and increased future uncertainty.", "category": "result" }, { "sentence": "Therefore HQOL assessment can be helpful identifying patients with increased distress.", "category": "result" } ] }, { "paper_id": "344349", "title": "A new approach to interdomain routing based on secure multi-party computation", "abstract": "Interdomain routing involves coordination among mutually distrustful parties, leading to the requirements that BGP provide policy autonomy, flexibility, and privacy. BGP provides these properties via the distributed execution of policy-based decisions during the iterative route computation process. This approach has poor convergence properties, makes planning and failover difficult, and is extremely difficult to change. To rectify these and other problems, we propose a radically different approach to interdomain-route computation, based on secure multi-party computation (SMPC). Our approach provides stronger privacy guarantees than BGP and enables the deployment of new policy paradigms. We report on an initial exploration of this idea and outline future directions for research.", "classified_sentences": [ { "sentence": "Interdomain routing involves coordination among mutually distrustful parties, leading to the requirements that BGP provide policy autonomy, flexibility, and privacy.", "category": "background" }, { "sentence": "BGP provides these properties via the distributed execution of policy-based decisions during the iterative route computation process.", "category": "method" }, { "sentence": "This approach has poor convergence properties, makes planning and failover difficult, and is extremely difficult to change.", "category": "background" }, { "sentence": "To rectify these and other problems, we propose a radically different approach to interdomain-route computation, based on secure multi-party computation (SMPC).", "category": "method" }, { "sentence": "Our approach provides stronger privacy guarantees than BGP and enables the deployment of new policy paradigms.", "category": "result" }, { "sentence": "We report on an initial exploration of this idea and outline future directions for research.", "category": "result" } ] }, { "paper_id": "357776", "title": "Supervised hashing with kernels", "abstract": "Recent years have witnessed the growing popularity of hashing in large-scale vision problems. It has been shown that the hashing quality could be boosted by leveraging supervised information into hash function learning. However, the existing supervised methods either lack adequate performance or often incur cumbersome model training. In this paper, we propose a novel kernel-based supervised hashing model which requires a limited amount of supervised information, i.e., similar and dissimilar data pairs, and a feasible training cost in achieving high quality hashing. The idea is to map the data to compact binary codes whose Hamming distances are minimized on similar pairs and simultaneously maximized on dissimilar pairs. Our approach is distinct from prior works by utilizing the equivalence between optimizing the code inner products and the Hamming distances. This enables us to sequentially and efficiently train the hash functions one bit at a time, yielding very short yet discriminative codes. We carry out extensive experiments on two image benchmarks with up to one million samples, demonstrating that our approach significantly outperforms the state-of-the-arts in searching both metric distance neighbors and semantically similar neighbors, with accuracy gains ranging from 13% to 46%.", "classified_sentences": [ { "sentence": "Recent years have witnessed the growing popularity of hashing in large-scale vision problems.", "category": "background" }, { "sentence": "It has been shown that the hashing quality could be boosted by leveraging supervised information into hash function learning.", "category": "background" }, { "sentence": "However, the existing supervised methods either lack adequate performance or often incur cumbersome model training.", "category": "background" }, { "sentence": "In this paper, we propose a novel kernel-based supervised hashing model which requires a limited amount of supervised information, i.e., similar and dissimilar data pairs, and a feasible training cost in achieving high quality hashing.", "category": "method" }, { "sentence": "The idea is to map the data to compact binary codes whose Hamming distances are minimized on similar pairs and simultaneously maximized on dissimilar pairs.", "category": "method" }, { "sentence": "Our approach is distinct from prior works by utilizing the equivalence between optimizing the code inner products and the Hamming distances.", "category": "method" }, { "sentence": "This enables us to sequentially and efficiently train the hash functions one bit at a time, yielding very short yet discriminative codes.", "category": "method" }, { "sentence": "We carry out extensive experiments on two image benchmarks with up to one million samples, demonstrating that our approach significantly outperforms the state-of-the-arts in searching both metric distance neighbors and semantically similar neighbors, with accuracy gains ranging from 13% to 46%.", "category": "result" } ] }, { "paper_id": "641747", "title": "State-of-the-Art in Visual Attention Modeling", "abstract": "Modeling visual attention-particularly stimulus-driven, saliency-based attention-has been a very active research area over the past 25 years. Many different models of attention are now available which, aside from lending theoretical contributions to other fields, have demonstrated successful applications in computer vision, mobile robotics, and cognitive systems. Here we review, from a computational perspective, the basic concepts of attention implemented in these models. We present a taxonomy of nearly 65 models, which provides a critical comparison of approaches, their capabilities, and shortcomings. In particular, 13 criteria derived from behavioral and computational studies are formulated for qualitative comparison of attention models. Furthermore, we address several challenging issues with models, including biological plausibility of the computations, correlation with eye movement datasets, bottom-up and top-down dissociation, and constructing meaningful performance measures. Finally, we highlight current research trends in attention modeling and provide insights for future.", "classified_sentences": [ { "sentence": "Modeling visual attention-particularly stimulus-driven, saliency-based attention-has been a very active research area over the past 25 years.", "category": "background" }, { "sentence": "Many different models of attention are now available which, aside from lending theoretical contributions to other fields, have demonstrated successful applications in computer vision, mobile robotics, and cognitive systems.", "category": "background" }, { "sentence": "Here we review, from a computational perspective, the basic concepts of attention implemented in these models.", "category": "method" }, { "sentence": "We present a taxonomy of nearly 65 models, which provides a critical comparison of approaches, their capabilities, and shortcomings.", "category": "method" }, { "sentence": "In particular, 13 criteria derived from behavioral and computational studies are formulated for qualitative comparison of attention models.", "category": "method" }, { "sentence": "Furthermore, we address several challenging issues with models, including biological plausibility of the computations, correlation with eye movement datasets, bottom-up and top-down dissociation, and constructing meaningful performance measures.", "category": "method" }, { "sentence": "Finally, we highlight current research trends in attention modeling and provide insights for future.", "category": "result" } ] }, { "paper_id": "1048051", "title": "An Empirical Analysis of Phishing Blacklists", "abstract": "In this paper, we study the eectiveness of phishing blacklists. We used 191 fresh phish that were less than 30 minutes old to conduct two tests on eight anti-phishing toolbars. We found that 63% of the phishing campaigns in our dataset lasted less than two hours. Blacklists were ineective when protecting users initially, as most of them caught less than 20% of phish at hour zero. We also found that blacklists were updated at dierent speeds, and varied in coverage, as 47% - 83% of phish appeared on blacklists 12 hours from the initial test. We found that two tools using heuristics to complement blacklists caught signicantly more phish initially than those using only blacklists. However, it took a long time for phish detected by heuristics to appear on blacklists. Finally, we tested the toolbars on a set of 13,458 legitimate URLs for false positives, and did not nd any instance of mislabeling for either blacklists or heuristics. We present these ndings and discuss ways in which anti-phishing tools can be improved.", "classified_sentences": [ { "sentence": "In this paper, we study the eectiveness of phishing blacklists.", "category": "background" }, { "sentence": "We used 191 fresh phish that were less than 30 minutes old to conduct two tests on eight anti-phishing toolbars.", "category": "method" }, { "sentence": "We found that 63% of the phishing campaigns in our dataset lasted less than two hours.", "category": "result" }, { "sentence": "Blacklists were ineective when protecting users initially, as most of them caught less than 20% of phish at hour zero.", "category": "result" }, { "sentence": "We also found that blacklists were updated at dierent speeds, and varied in coverage, as 47% - 83% of phish appeared on blacklists 12 hours from the initial test.", "category": "result" }, { "sentence": "We found that two tools using heuristics to complement blacklists caught signicantly more phish initially than those using only blacklists.", "category": "result" }, { "sentence": "However, it took a long time for phish detected by heuristics to appear on blacklists.", "category": "result" }, { "sentence": "Finally, we tested the toolbars on a set of 13,458 legitimate URLs for false positives, and did not nd any instance of mislabeling for either blacklists or heuristics.", "category": "result" }, { "sentence": "We present these ndings and discuss ways in which anti-phishing tools can be improved.", "category": "result" } ] }, { "paper_id": "1119562", "title": "High Prevalence of Abnormal Glucose Metabolism and Poor Sensitivity of Fasting Plasma Glucose in the Chronic Phase of Stroke", "abstract": "Background: Cardiovascular disease is the leading cause of death in long-term stroke survivors, and whole-body glucose metabolism is strongly linked to cardiovascular disease risk. This study provides important preliminary information on the prevalence of abnormal glucose metabolism in chronic stroke patients (mean 3 years after stroke) and reports on the utility of screening for abnormalities using fasting plasma glucose (FPG) in this population. Methods: Two hundred and sixteen chronic hemiparetic stroke patients were screened for diabetes status by medical history and FPG. A subset (n = 80) was evaluated by oral glucose tolerance test to assess the utility of screening for abnormalities using FPG alone. Results: Seventy-five of the 216 (35%) had type 2 diabetes by medical history. Another 70 were either diabetic (n = 11) or had impaired fasting glucose (n = 59) based on a single blood draw at the time of screening. FPG among non-diabetic stroke patients had a sensitivity of 49% for predicting abnormalities in the 2-hour glucose level during oral glucose tolerance test. Cumulative results identify 77% as abnormal (impaired or diabetic) on the basis of medical history, fasting plasma glucose, and/or 2-hour glucose level. Conclusions: The prevalence of abnormal glucose metabolism is extremely high in chronic stroke and is underestimated on the basis of FPG.", "classified_sentences": [ { "sentence": "Background: Cardiovascular disease is the leading cause of death in long-term stroke survivors, and whole-body glucose metabolism is strongly linked to cardiovascular disease risk.", "category": "background" }, { "sentence": "This study provides important preliminary information on the prevalence of abnormal glucose metabolism in chronic stroke patients (mean 3 years after stroke) and reports on the utility of screening for abnormalities using fasting plasma glucose (FPG) in this population.", "category": "background" }, { "sentence": "Methods: Two hundred and sixteen chronic hemiparetic stroke patients were screened for diabetes status by medical history and FPG.", "category": "method" }, { "sentence": "A subset (n = 80) was evaluated by oral glucose tolerance test to assess the utility of screening for abnormalities using FPG alone.", "category": "method" }, { "sentence": "Results: Seventy-five of the 216 (35%) had type 2 diabetes by medical history.", "category": "result" }, { "sentence": "Another 70 were either diabetic (n = 11) or had impaired fasting glucose (n = 59) based on a single blood draw at the time of screening.", "category": "result" }, { "sentence": "FPG among non-diabetic stroke patients had a sensitivity of 49% for predicting abnormalities in the 2-hour glucose level during oral glucose tolerance test.", "category": "result" }, { "sentence": "Cumulative results identify 77% as abnormal (impaired or diabetic) on the basis of medical history, fasting plasma glucose, and/or 2-hour glucose level.", "category": "result" }, { "sentence": "Conclusions: The prevalence of abnormal glucose metabolism is extremely high in chronic stroke and is underestimated on the basis of FPG.", "category": "result" } ] }, { "paper_id": "1192893", "title": "Cycle Life Evaluation Based on Accelerated Aging Testing for Lithium-Ion Capacitors as Alternative to Rechargeable Batteries", "abstract": "Lithium-ion capacitors (LICs) are a hybrid energy storage device combining the energy storage mechanisms of lithium-ion batteries (LIBs) and electric double-layer capacitors (EDLCs), and are considered attractive not only in high-power applications but also as an alternative to rechargeable batteries due to their inherent long cycle life and relatively high energy density. The cycle life testing was performed for commercial-off-the-shelf (COTS) LIC cells procured from three different manufactures, and the cycle life prediction model developed for EDLCs in the previous work was applied to LICs. Based on the resultant capacitance retention trends, the activation energies of degradation ratios were calculated using an Arrhenius equation, whereupon aging acceleration factors were determined. The calculated acceleration factors varied depending on manufacturers, suggesting that a proper aging acceleration factor should be determined for each manufacture cell based on cycle life testing rather than simply applying a rule of thumb which had been accepted for LIBs and EDLCs. The resulting and predicted capacitance retention trends correlated well, verifying that the cycle life prediction model established for EDLCs in the previous work would also be usable for LICs as an alternative to rechargeable batteries.", "classified_sentences": [ { "sentence": "Lithium-ion capacitors (LICs) are a hybrid energy storage device combining the energy storage mechanisms of lithium-ion batteries (LIBs) and electric double-layer capacitors (EDLCs), and are considered attractive not only in high-power applications but also as an alternative to rechargeable batteries due to their inherent long cycle life and relatively high energy density.", "category": "background" }, { "sentence": "The cycle life testing was performed for commercial-off-the-shelf (COTS) LIC cells procured from three different manufactures, and the cycle life prediction model developed for EDLCs in the previous work was applied to LICs.", "category": "method" }, { "sentence": "Based on the resultant capacitance retention trends, the activation energies of degradation ratios were calculated using an Arrhenius equation, whereupon aging acceleration factors were determined.", "category": "method" }, { "sentence": "The calculated acceleration factors varied depending on manufacturers, suggesting that a proper aging acceleration factor should be determined for each manufacture cell based on cycle life testing rather than simply applying a rule of thumb which had been accepted for LIBs and EDLCs.", "category": "result" }, { "sentence": "The resulting and predicted capacitance retention trends correlated well, verifying that the cycle life prediction model established for EDLCs in the previous work would also be usable for LICs as an alternative to rechargeable batteries.", "category": "result" } ] }, { "paper_id": "1234761", "title": "Beyond Locality-Sensitive Hashing", "abstract": "We present a new data structure for the c-approximate near neighbor problem (ANN) in the Euclidean space. For n points in Rd, our algorithm achieves Oc(nρ + dlogn) query time and Oc(n1+ρ + dlogn) space, where ρ ≤ 7/(8c2) + O(1/c3) + oc(1). This is the first improvement over the result by Andoni and Indyk (FOCS 2006) and the first data structure that bypasses a locality-sensitive hashing lower bound proved by O'Donnell, Wu and Zhou (ICS 2011). By a standard reduction we obtain a data structure for the Hamming space and e1 norm with ρ ≤ 7/(8c)+ O(1/c3/2)+ oc(1), which is the first improvement over the result of Indyk and Motwani (STOC 1998).", "classified_sentences": [ { "sentence": "We present a new data structure for the c-approximate near neighbor problem (ANN) in the Euclidean space.", "category": "method" }, { "sentence": "For n points in Rd, our algorithm achieves Oc(nρ + dlogn) query time and Oc(n1+ρ + dlogn) space, where ρ ≤ 7/(8c2) + O(1/c3) + oc(1).", "category": "result" }, { "sentence": "This is the first improvement over the result by Andoni and Indyk (FOCS 2006) and the first data structure that bypasses a locality-sensitive hashing lower bound proved by O'Donnell, Wu and Zhou (ICS 2011).", "category": "result" }, { "sentence": "By a standard reduction we obtain a data structure for the Hamming space and e1 norm with ρ ≤ 7/(8c)+ O(1/c3/2)+ oc(1), which is the first improvement over the result of Indyk and Motwani (STOC 1998).", "category": "result" } ] }, { "paper_id": "1334960", "title": "Frequency-tuned salient region detection", "abstract": "Detection of visually salient image regions is useful for applications like object segmentation, adaptive compression, and object recognition. In this paper, we introduce a method for salient region detection that outputs full resolution saliency maps with well-defined boundaries of salient objects. These boundaries are preserved by retaining substantially more frequency content from the original image than other existing techniques. Our method exploits features of color and luminance, is simple to implement, and is computationally efficient. We compare our algorithm to five state-of-the-art salient region detection methods with a frequency domain analysis, ground truth, and a salient object segmentation application. Our method outperforms the five algorithms both on the ground-truth evaluation and on the segmentation task by achieving both higher precision and better recall.", "classified_sentences": [ { "sentence": "Detection of visually salient image regions is useful for applications like object segmentation, adaptive compression, and object recognition.", "category": "background" }, { "sentence": "In this paper, we introduce a method for salient region detection that outputs full resolution saliency maps with well-defined boundaries of salient objects.", "category": "method" }, { "sentence": "These boundaries are preserved by retaining substantially more frequency content from the original image than other existing techniques.", "category": "method" }, { "sentence": "Our method exploits features of color and luminance, is simple to implement, and is computationally efficient.", "category": "method" }, { "sentence": "We compare our algorithm to five state-of-the-art salient region detection methods with a frequency domain analysis, ground truth, and a salient object segmentation application.", "category": "method" }, { "sentence": "Our method outperforms the five algorithms both on the ground-truth evaluation and on the segmentation task by achieving both higher precision and better recall.", "category": "result" } ] }, { "paper_id": "1392065", "title": "Analysis of chain reaction between two stock indices fluctuations by statistical physics systems", "abstract": "In this paper, we consider the statistical properties of chain reaction of stock indices. The theory of interacting systems and statistical physics are applied to describe and study the fluctuations of two stock indices in a stock market, and the properties of the interacting reaction of the two indices are investigated in the present paper. In this work, stochastic analysis and the two random paths model are used to study the probability distribution for the chain reaction of stock indices, further we show the asymptotical behavior of probability measures of the fluctuations for the two stock indices model. In the last part, we discuss the convergence of the finite dimensional probability distributions for the financial model.", "classified_sentences": [ { "sentence": "In this paper, we consider the statistical properties of chain reaction of stock indices.", "category": "background" }, { "sentence": "The theory of interacting systems and statistical physics are applied to describe and study the fluctuations of two stock indices in a stock market, and the properties of the interacting reaction of the two indices are investigated in the present paper.", "category": "method" }, { "sentence": "In this work, stochastic analysis and the two random paths model are used to study the probability distribution for the chain reaction of stock indices, further we show the asymptotical behavior of probability measures of the fluctuations for the two stock indices model.", "category": "method" }, { "sentence": "In the last part, we discuss the convergence of the finite dimensional probability distributions for the financial model.", "category": "result" } ] }, { "paper_id": "1622331", "title": "Intranuclear Anchoring of Repetitive DNA Sequences", "abstract": "Centromeres, telomeres, and ribosomal gene clusters consist of repetitive DNA sequences. To assess their contributions to the spatial organization of the interphase genome, their interactions with the nucleoskeleton were examined in quiescent and activated human lymphocytes. The nucleoskeletons were prepared using “physiological” conditions. The resulting structures were probed for specific DNA sequences of centromeres, telomeres, and ribosomal genes by in situ hybridization; the electroeluted DNA fractions were examined by blot hybridization. In both nonstimulated and stimulated lymphocytes, centromeric alpha-satellite repeats were almost exclusively found in the eluted fraction, while telomeric sequences remained attached to the nucleoskeleton. Ribosomal genes showed a transcription-dependent attachment pattern: in unstimulated lymphocytes, transcriptionally inactive ribosomal genes located outside the nucleolus were eluted completely. When comparing transcription unit and intergenic spacer, significantly more of the intergenic spacer was removed. In activated lymphocytes, considerable but similar amounts of both rDNA fragments were eluted. The results demonstrate that: (a) the various repetitive DNA sequences differ significantly in their intranuclear anchoring, (b) telomeric rather than centromeric DNA sequences form stable attachments to the nucleoskeleton, and (c) different attachment mechanisms might be responsible for the interaction of ribosomal genes with the nucleoskeleton.", "classified_sentences": [ { "sentence": "Centromeres, telomeres, and ribosomal gene clusters consist of repetitive DNA sequences.", "category": "background" }, { "sentence": "To assess their contributions to the spatial organization of the interphase genome, their interactions with the nucleoskeleton were examined in quiescent and activated human lymphocytes.", "category": "method" }, { "sentence": "The nucleoskeletons were prepared using “physiological” conditions.", "category": "method" }, { "sentence": "The resulting structures were probed for specific DNA sequences of centromeres, telomeres, and ribosomal genes by in situ hybridization; the electroeluted DNA fractions were examined by blot hybridization.", "category": "method" }, { "sentence": "In both nonstimulated and stimulated lymphocytes, centromeric alpha-satellite repeats were almost exclusively found in the eluted fraction, while telomeric sequences remained attached to the nucleoskeleton.", "category": "result" }, { "sentence": "Ribosomal genes showed a transcription-dependent attachment pattern: in unstimulated lymphocytes, transcriptionally inactive ribosomal genes located outside the nucleolus were eluted completely.", "category": "result" }, { "sentence": "When comparing transcription unit and intergenic spacer, significantly more of the intergenic spacer was removed.", "category": "result" }, { "sentence": "In activated lymphocytes, considerable but similar amounts of both rDNA fragments were eluted.", "category": "result" }, { "sentence": "The results demonstrate that: (a) the various repetitive DNA sequences differ significantly in their intranuclear anchoring, (b) telomeric rather than centromeric DNA sequences form stable attachments to the nucleoskeleton, and (c) different attachment mechanisms might be responsible for the interaction of ribosomal genes with the nucleoskeleton.", "category": "result" } ] }, { "paper_id": "1683999", "title": "Effects of Ammonia and β‐Methylene‐dl‐Aspartate on the Oxidation of Glucose and Pyruvate by Neurons and Astrocytes in Primary Culture", "abstract": "Abstract: Both ammonia and β‐methylene‐dl‐aspartate (β‐MA), an irreversible inhibitor of aspartate aminotransferase activity and thus of the malate‐aspartate shuttle, were found previously to decrease oxidative metabolism in cerebral cortex slices. In the present work, the possibility that ammonia and β‐MA affect energy metabolism by a common mechanism (i.e., via inhibition of the malate‐aspartate shuttle) was investigated using primary cultures of neurons and astrocytes. Incubation of astrocytes for 30 min with 5 mMβ‐MA resulted in a decreased production of 14CO2 from [U‐14Clglucose, but did not affect 14CO2 production from [2–14C] pyruvate. Conversely, incubation of astrocytes with 3 mM ammonium chloride resulted in decreased 14CO2 production from [2–14C] pyruvate, but 14CO2 production from [U‐14C] glucose was not significantly affected. Ammonium chloride had no significant effect on 14CO2 production from either [U‐14C] glucose or [2–14]pyruvate by neurons. However, incubation of neurons with β‐MA or β‐MA plus ammonium chloride resulted in a 45% decrease of 14CO2 production from both [U‐14C] glucose and [2–14C] pyruvate. A 2‐h incubation of astrocytes with β‐MA resulted in no change in ATP levels, but a 35% decrease in phosphocreatine. Similar treatment of neurons resulted in >50% decrease in ATP, but had little effect on phosphocreatine. β‐MA also caused a decrease in glutamate and aspartate content of neurons, but not of astrocytes. The different metabolic responses of neurons and astrocytes towards β‐MA were probably not due to a differential inhibition of aspartate aminotransferase which was inhibited by ∼45% in astrocytes and by ∼55% in neurons.", "classified_sentences": [ { "sentence": "Both ammonia and β‐methylene‐dl‐aspartate (β‐MA), an irreversible inhibitor of aspartate aminotransferase activity and thus of the malate‐aspartate shuttle, were found previously to decrease oxidative metabolism in cerebral cortex slices.", "category": "background" }, { "sentence": "In the present work, the possibility that ammonia and β‐MA affect energy metabolism by a common mechanism (i.e., via inhibition of the malate‐aspartate shuttle) was investigated using primary cultures of neurons and astrocytes.", "category": "method" }, { "sentence": "Incubation of astrocytes for 30 min with 5 mMβ‐MA resulted in a decreased production of 14CO2 from [U‐14Clglucose, but did not affect 14CO2 production from [2–14C] pyruvate.", "category": "result" }, { "sentence": "Conversely, incubation of astrocytes with 3 mM ammonium chloride resulted in decreased 14CO2 production from [2–14C] pyruvate, but 14CO2 production from [U‐14C] glucose was not significantly affected.", "category": "result" }, { "sentence": "Ammonium chloride had no significant effect on 14CO2 production from either [U‐14C] glucose or [2–14]pyruvate by neurons.", "category": "result" }, { "sentence": "However, incubation of neurons with β‐MA or β‐MA plus ammonium chloride resulted in a 45% decrease of 14CO2 production from both [U‐14C] glucose and [2–14C] pyruvate.", "category": "result" }, { "sentence": "A 2‐h incubation of astrocytes with β‐MA resulted in no change in ATP levels, but a 35% decrease in phosphocreatine.", "category": "result" }, { "sentence": "Similar treatment of neurons resulted in >50% decrease in ATP, but had little effect on phosphocreatine.", "category": "result" }, { "sentence": "β‐MA also caused a decrease in glutamate and aspartate content of neurons, but not of astrocytes.", "category": "result" }, { "sentence": "The different metabolic responses of neurons and astrocytes towards β‐MA were probably not due to a differential inhibition of aspartate aminotransferase which was inhibited by ∼45% in astrocytes and by ∼55% in neurons.", "category": "result" } ] }, { "paper_id": "1916754", "title": "Simple Semi-supervised Dependency Parsing", "abstract": "We present a simple and effective semisupervised method for training dependency parsers. We focus on the problem of lexical representation, introducing features that incorporate word clusters derived from a large unannotated corpus. We demonstrate the effectiveness of the approach in a series of dependency parsing experiments on the Penn Treebank and Prague Dependency Treebank, and we show that the cluster-based features yield substantial gains in performance across a wide range of conditions. For example, in the case of English unlabeled second-order parsing, we improve from a baseline accuracy of 92.02% to 93.16%, and in the case of Czech unlabeled second-order parsing, we improve from a baseline accuracy of 86.13% to 87.13%. In addition, we demonstrate that our method also improves performance when small amounts of training data are available, and can roughly halve the amount of supervised data required to reach a desired level of performance.", "classified_sentences": [ { "sentence": "We present a simple and effective semisupervised method for training dependency parsers.", "category": "method" }, { "sentence": "We focus on the problem of lexical representation, introducing features that incorporate word clusters derived from a large unannotated corpus.", "category": "method" }, { "sentence": "We demonstrate the effectiveness of the approach in a series of dependency parsing experiments on the Penn Treebank and Prague Dependency Treebank, and we show that the cluster-based features yield substantial gains in performance across a wide range of conditions.", "category": "result" }, { "sentence": "For example, in the case of English unlabeled second-order parsing, we improve from a baseline accuracy of 92.02% to 93.16%, and in the case of Czech unlabeled second-order parsing, we improve from a baseline accuracy of 86.13% to 87.13%.", "category": "result" }, { "sentence": "In addition, we demonstrate that our method also improves performance when small amounts of training data are available, and can roughly halve the amount of supervised data required to reach a desired level of performance.", "category": "result" } ] }, { "paper_id": "1968620", "title": "Trading sex: Voluntary or coerced? The experiences of homeless youth", "abstract": "This study examined the circumstances surrounding a homeless youth's “decision “ to trade sex for food, money, shelter, or drugs. Forty homeless youth in 4 Midwestern states participated in individual, in‐depth qualitative interviews. Interviewers recruited youth through both service agencies and street outreach. The findings revealed that approximately one third of the sample had some experience with trading sex, whether it was in the form of having traded sex, having been propositioned to trade sex but having refused, or having friends or acquaintances that had traded sex. Young people's reports indicated that they had traded sex for things they deemed necessary in order to survive (i.e., food, shelter, money, or drugs) and that they did not want to trade sex, but did so because they were desperate and lacked alternatives. Additionally, others were coerced, manipulated, or forced to do so, indicating that the decision to trade sex is not always voluntary. We discuss the implications of these findings in terms of cumulative effects on youths’ later development. Directions for future research among this population are also discussed.", "classified_sentences": [ { "sentence": "This study examined the circumstances surrounding a homeless youth's “decision “ to trade sex for food, money, shelter, or drugs.", "category": "background" }, { "sentence": "Forty homeless youth in 4 Midwestern states participated in individual, in‐depth qualitative interviews.", "category": "method" }, { "sentence": "Interviewers recruited youth through both service agencies and street outreach.", "category": "method" }, { "sentence": "The findings revealed that approximately one third of the sample had some experience with trading sex, whether it was in the form of having traded sex, having been propositioned to trade sex but having refused, or having friends or acquaintances that had traded sex.", "category": "result" }, { "sentence": "Young people's reports indicated that they had traded sex for things they deemed necessary in order to survive (i.e., food, shelter, money, or drugs) and that they did not want to trade sex, but did so because they were desperate and lacked alternatives.", "category": "result" }, { "sentence": "Additionally, others were coerced, manipulated, or forced to do so, indicating that the decision to trade sex is not always voluntary.", "category": "result" }, { "sentence": "We discuss the implications of these findings in terms of cumulative effects on youths’ later development.", "category": "result" }, { "sentence": "Directions for future research among this population are also discussed.", "category": "result" } ] }, { "paper_id": "2097188", "title": "Simultaneous Feature Aggregating and Hashing for Large-Scale Image Search", "abstract": "In most state-of-the-art hashing-based visual search systems, local image descriptors of an image are first aggregated as a single feature vector. This feature vector is then subjected to a hashing function that produces a binary hash code. In previous work, the aggregating and the hashing processes are designed independently. In this paper, we propose a novel framework where feature aggregating and hashing are designed simultaneously and optimized jointly. Specifically, our joint optimization produces aggregated representations that can be better reconstructed by some binary codes. This leads to more discriminative binary hash codes and improved retrieval accuracy. In addition, we also propose a fast version of the recently-proposed Binary Autoencoder to be used in our proposed framework. We perform extensive retrieval experiments on several benchmark datasets with both SIFT and convolutional features. Our results suggest that the proposed framework achieves significant improvements over the state of the art.", "classified_sentences": [ { "sentence": "In most state-of-the-art hashing-based visual search systems, local image descriptors of an image are first aggregated as a single feature vector.", "category": "background" }, { "sentence": "This feature vector is then subjected to a hashing function that produces a binary hash code.", "category": "background" }, { "sentence": "In previous work, the aggregating and the hashing processes are designed independently.", "category": "background" }, { "sentence": "In this paper, we propose a novel framework where feature aggregating and hashing are designed simultaneously and optimized jointly.", "category": "method" }, { "sentence": "Specifically, our joint optimization produces aggregated representations that can be better reconstructed by some binary codes.", "category": "method" }, { "sentence": "This leads to more discriminative binary hash codes and improved retrieval accuracy.", "category": "result" }, { "sentence": "In addition, we also propose a fast version of the recently-proposed Binary Autoencoder to be used in our proposed framework.", "category": "method" }, { "sentence": "We perform extensive retrieval experiments on several benchmark datasets with both SIFT and convolutional features.", "category": "method" }, { "sentence": "Our results suggest that the proposed framework achieves significant improvements over the state of the art.", "category": "result" } ] }, { "paper_id": "2361503", "title": "K-Means Hashing: An Affinity-Preserving Quantization Method for Learning Binary Compact Codes", "abstract": "In computer vision there has been increasing interest in learning hashing codes whose Hamming distance approximates the data similarity. The hashing functions play roles in both quantizing the vector space and generating similarity-preserving codes. Most existing hashing methods use hyper-planes (or kernelized hyper-planes) to quantize and encode. In this paper, we present a hashing method adopting the k-means quantization. We propose a novel Affinity-Preserving K-means algorithm which simultaneously performs k-means clustering and learns the binary indices of the quantized cells. The distance between the cells is approximated by the Hamming distance of the cell indices. We further generalize our algorithm to a product space for learning longer codes. Experiments show our method, named as K-means Hashing (KMH), outperforms various state-of-the-art hashing encoding methods.", "classified_sentences": [ { "sentence": "In computer vision there has been increasing interest in learning hashing codes whose Hamming distance approximates the data similarity.", "category": "background" }, { "sentence": "The hashing functions play roles in both quantizing the vector space and generating similarity-preserving codes.", "category": "background" }, { "sentence": "Most existing hashing methods use hyper-planes (or kernelized hyper-planes) to quantize and encode.", "category": "background" }, { "sentence": "In this paper, we present a hashing method adopting the k-means quantization.", "category": "method" }, { "sentence": "We propose a novel Affinity-Preserving K-means algorithm which simultaneously performs k-means clustering and learns the binary indices of the quantized cells.", "category": "method" }, { "sentence": "The distance between the cells is approximated by the Hamming distance of the cell indices.", "category": "method" }, { "sentence": "We further generalize our algorithm to a product space for learning longer codes.", "category": "method" }, { "sentence": "Experiments show our method, named as K-means Hashing (KMH), outperforms various state-of-the-art hashing encoding methods.", "category": "result" } ] }, { "paper_id": "2467910", "title": "Robust Multispectral Image Registration Using Mutual-Information Models", "abstract": "Image registration is a vital step in the processing of multispectral imagery. The accuracy to which imagery collected at multiple wavelengths can be aligned directly affects the resolution of the spectral end products. Automated registration of the multispectral imagery can often be unreliable, particularly between visible and infrared imagery, due to the significant differences in scene reflectance at different wavelengths. This is further complicated by the thermal features that exist at longer wavelengths. We develop new mathematical and computational models for robust image registration. In particular, we develop a frequency-domain model for the mutual-information surface around the optimal parameters and use it to develop a robust gradient ascent algorithm. For a robust performance, we require that the algorithm be initialized close to the optimal registration parameters. As a measure of how close we need to be, we propose the use of the correlation length and provide an efficient algorithm for estimating it. We measure the performance of the proposed algorithm over hundreds of random initializations to demonstrate its robustness on real data. We find that the algorithm should be expected to converge, as long as the registration parameters are initialized to be within the correlation-length distance from the optimum", "classified_sentences": [ { "sentence": "Image registration is a vital step in the processing of multispectral imagery.", "category": "background" }, { "sentence": "The accuracy to which imagery collected at multiple wavelengths can be aligned directly affects the resolution of the spectral end products.", "category": "background" }, { "sentence": "Automated registration of the multispectral imagery can often be unreliable, particularly between visible and infrared imagery, due to the significant differences in scene reflectance at different wavelengths.", "category": "background" }, { "sentence": "This is further complicated by the thermal features that exist at longer wavelengths.", "category": "background" }, { "sentence": "We develop new mathematical and computational models for robust image registration.", "category": "method" }, { "sentence": "In particular, we develop a frequency-domain model for the mutual-information surface around the optimal parameters and use it to develop a robust gradient ascent algorithm.", "category": "method" }, { "sentence": "For a robust performance, we require that the algorithm be initialized close to the optimal registration parameters.", "category": "method" }, { "sentence": "As a measure of how close we need to be, we propose the use of the correlation length and provide an efficient algorithm for estimating it.", "category": "method" }, { "sentence": "We measure the performance of the proposed algorithm over hundreds of random initializations to demonstrate its robustness on real data.", "category": "result" }, { "sentence": "We find that the algorithm should be expected to converge, as long as the registration parameters are initialized to be within the correlation-length distance from the optimum", "category": "result" } ] }, { "paper_id": "2519230", "title": "Deep semantic ranking based hashing for multi-label image retrieval", "abstract": "With the rapid growth of web images, hashing has received increasing interests in large scale image retrieval. Research efforts have been devoted to learning compact binary codes that preserve semantic similarity based on labels. However, most of these hashing methods are designed to handle simple binary similarity. The complex multi-level semantic structure of images associated with multiple labels have not yet been well explored. Here we propose a deep semantic ranking based method for learning hash functions that preserve multilevel semantic similarity between multi-label images. In our approach, deep convolutional neural network is incorporated into hash functions to jointly learn feature representations and mappings from them to hash codes, which avoids the limitation of semantic representation power of hand-crafted features. Meanwhile, a ranking list that encodes the multilevel similarity information is employed to guide the learning of such deep hash functions. An effective scheme based on surrogate loss is used to solve the intractable optimization problem of nonsmooth and multivariate ranking measures involved in the learning procedure. Experimental results show the superiority of our proposed approach over several state-of-the-art hashing methods in term of ranking evaluation metrics when tested on multi-label image datasets.", "classified_sentences": [ { "sentence": "With the rapid growth of web images, hashing has received increasing interests in large scale image retrieval.", "category": "background" }, { "sentence": "Research efforts have been devoted to learning compact binary codes that preserve semantic similarity based on labels.", "category": "background" }, { "sentence": "However, most of these hashing methods are designed to handle simple binary similarity.", "category": "background" }, { "sentence": "The complex multi-level semantic structure of images associated with multiple labels have not yet been well explored.", "category": "background" }, { "sentence": "Here we propose a deep semantic ranking based method for learning hash functions that preserve multilevel semantic similarity between multi-label images.", "category": "method" }, { "sentence": "In our approach, deep convolutional neural network is incorporated into hash functions to jointly learn feature representations and mappings from them to hash codes, which avoids the limitation of semantic representation power of hand-crafted features.", "category": "method" }, { "sentence": "Meanwhile, a ranking list that encodes the multilevel similarity information is employed to guide the learning of such deep hash functions.", "category": "method" }, { "sentence": "An effective scheme based on surrogate loss is used to solve the intractable optimization problem of nonsmooth and multivariate ranking measures involved in the learning procedure.", "category": "method" }, { "sentence": "Experimental results show the superiority of our proposed approach over several state-of-the-art hashing methods in term of ranking evaluation metrics when tested on multi-label image datasets.", "category": "result" } ] }, { "paper_id": "2605321", "title": "Iterative Quantization: A Procrustean Approach to Learning Binary Codes for Large-Scale Image Retrieval", "abstract": "This paper addresses the problem of learning similarity-preserving binary codes for efficient similarity search in large-scale image collections. We formulate this problem in terms of finding a rotation of zero-centered data so as to minimize the quantization error of mapping this data to the vertices of a zero-centered binary hypercube, and propose a simple and efficient alternating minimization algorithm to accomplish this task. This algorithm, dubbed iterative quantization (ITQ), has connections to multiclass spectral clustering and to the orthogonal Procrustes problem, and it can be used both with unsupervised data embeddings such as PCA and supervised embeddings such as canonical correlation analysis (CCA). The resulting binary codes significantly outperform several other state-of-the-art methods. We also show that further performance improvements can result from transforming the data with a nonlinear kernel mapping prior to PCA or CCA. Finally, we demonstrate an application of ITQ to learning binary attributes or \"classemes\" on the ImageNet data set.", "classified_sentences": [ { "sentence": "This paper addresses the problem of learning similarity-preserving binary codes for efficient similarity search in large-scale image collections.", "category": "background" }, { "sentence": "We formulate this problem in terms of finding a rotation of zero-centered data so as to minimize the quantization error of mapping this data to the vertices of a zero-centered binary hypercube, and propose a simple and efficient alternating minimization algorithm to accomplish this task.", "category": "method" }, { "sentence": "This algorithm, dubbed iterative quantization (ITQ), has connections to multiclass spectral clustering and to the orthogonal Procrustes problem, and it can be used both with unsupervised data embeddings such as PCA and supervised embeddings such as canonical correlation analysis (CCA).", "category": "method" }, { "sentence": "The resulting binary codes significantly outperform several other state-of-the-art methods.", "category": "result" }, { "sentence": "We also show that further performance improvements can result from transforming the data with a nonlinear kernel mapping prior to PCA or CCA.", "category": "method" }, { "sentence": "Finally, we demonstrate an application of ITQ to learning binary attributes or \"classemes\" on the ImageNet data set.", "category": "result" } ] }, { "paper_id": "2621465", "title": "Exploring Compositional Architectures and Word Vector Representations for Prepositional Phrase Attachment", "abstract": "Prepositional phrase (PP) attachment disambiguation is a known challenge in syntactic parsing. The lexical sparsity associated with PP attachments motivates research in word representations that can capture pertinent syntactic and semantic features of the word. One promising solution is to use word vectors induced from large amounts of raw text. However, state-of-the-art systems that employ such representations yield modest gains in PP attachment accuracy. In this paper, we show that word vector representations can yield significant PP attachment performance gains. This is achieved via a non-linear architecture that is discriminatively trained to maximize PP attachment accuracy. The architecture is initialized with word vectors trained from unlabeled data, and relearns those to maximize attachment accuracy. We obtain additional performance gains with alternative representations such as dependency-based word vectors. When tested on both English and Arabic datasets, our method outperforms both a strong SVM classifier and state-of-the-art parsers. For instance, we achieve 82.6% PP attachment accuracy on Arabic, while the Turbo and Charniak self-trained parsers obtain 76.7% and 80.8% respectively.", "classified_sentences": [ { "sentence": "Prepositional phrase (PP) attachment disambiguation is a known challenge in syntactic parsing.", "category": "background" }, { "sentence": "The lexical sparsity associated with PP attachments motivates research in word representations that can capture pertinent syntactic and semantic features of the word.", "category": "background" }, { "sentence": "One promising solution is to use word vectors induced from large amounts of raw text.", "category": "method" }, { "sentence": "However, state-of-the-art systems that employ such representations yield modest gains in PP attachment accuracy.", "category": "background" }, { "sentence": "In this paper, we show that word vector representations can yield significant PP attachment performance gains.", "category": "method" }, { "sentence": "This is achieved via a non-linear architecture that is discriminatively trained to maximize PP attachment accuracy.", "category": "method" }, { "sentence": "The architecture is initialized with word vectors trained from unlabeled data, and relearns those to maximize attachment accuracy.", "category": "method" }, { "sentence": "We obtain additional performance gains with alternative representations such as dependency-based word vectors.", "category": "method" }, { "sentence": "When tested on both English and Arabic datasets, our method outperforms both a strong SVM classifier and state-of-the-art parsers.", "category": "result" }, { "sentence": "For instance, we achieve 82.6% PP attachment accuracy on Arabic, while the Turbo and Charniak self-trained parsers obtain 76.7% and 80.8% respectively.", "category": "result" } ] }, { "paper_id": "2839336", "title": "Effects of Alcohol and Other Drugs on Driver Performance", "abstract": "In the past century we have learned that driving performance is impaired by alcohol even in low dosage, and that many other drugs are also linked to impairment. This paper is a summary of some of the more relevant studies in the past fifty years – an overview of our knowledge and unanswered questions. There is no evidence of a threshold blood alcohol (BAC) below which impairment does not occur, and there is no defined category of drivers who will not be impaired by alcohol. Alcohol increases not only the probability of collision, but also the probability of poor clinical outcome for injuries sustained when impaired by alcohol. This review samples the results of the myriad studies that have been performed during the last half century as experiments have moved from examination of simple sensory, perceptual and motor behaviours to more complex measures of cognitive functioning such as divided attention and mental workload. These more sophisticated studies show that significant impairment occurs at very low BACs (<0.02 gm/100 ml). However, much remains to be determined regarding the more emotional aspects of behaviour, such as judgment, aggression and risk taking. Considering that the majority of alcohol related accidents occur at night, there is a need for increased examination on the role of fatigue, circadian cycles and sleep loss. The study of the effects of drugs other than alcohol is more complex because of the number of substances of potential interest, the difficulties estimating drug levels and the complexity of the drug/subject interactions. The drugs of current concern are marijuana, the benzodiazepines, other psychoactive medications, the stimulants and the narcotics. No one test or group of tests currently meets the need for detecting and documenting impairment, either in the laboratory or at the roadside.", "classified_sentences": [ { "sentence": "In the past century we have learned that driving performance is impaired by alcohol even in low dosage, and that many other drugs are also linked to impairment.", "category": "background" }, { "sentence": "This paper is a summary of some of the more relevant studies in the past fifty years – an overview of our knowledge and unanswered questions.", "category": "background" }, { "sentence": "There is no evidence of a threshold blood alcohol (BAC) below which impairment does not occur, and there is no defined category of drivers who will not be impaired by alcohol.", "category": "background" }, { "sentence": "Alcohol increases not only the probability of collision, but also the probability of poor clinical outcome for injuries sustained when impaired by alcohol.", "category": "background" }, { "sentence": "This review samples the results of the myriad studies that have been performed during the last half century as experiments have moved from examination of simple sensory, perceptual and motor behaviours to more complex measures of cognitive functioning such as divided attention and mental workload.", "category": "method" }, { "sentence": "These more sophisticated studies show that significant impairment occurs at very low BACs (<0.02 gm/100 ml).", "category": "result" }, { "sentence": "However, much remains to be determined regarding the more emotional aspects of behaviour, such as judgment, aggression and risk taking.", "category": "background" }, { "sentence": "Considering that the majority of alcohol related accidents occur at night, there is a need for increased examination on the role of fatigue, circadian cycles and sleep loss.", "category": "background" }, { "sentence": "The study of the effects of drugs other than alcohol is more complex because of the number of substances of potential interest, the difficulties estimating drug levels and the complexity of the drug/subject interactions.", "category": "background" }, { "sentence": "The drugs of current concern are marijuana, the benzodiazepines, other psychoactive medications, the stimulants and the narcotics.", "category": "background" }, { "sentence": "No one test or group of tests currently meets the need for detecting and documenting impairment, either in the laboratory or at the roadside.", "category": "result" } ] }, { "paper_id": "2880288", "title": "Online Comments on Smoking Bans in Psychiatric Hospitals Units", "abstract": "Objective: Individuals with mental health concerns are disproportionately affected by and suffer the negative consequences of tobacco use disorder, perhaps because smoking has historically been part of psychiatry's culture. In the early 1990s, psychiatric inpatient facilities were exempted from U.S. hospital smoking bans, in response to public outcry with national media attention. Almost 2 decades later, the current study characterizes online conversation about psychiatric hospital smoking bans. Previous commenting studies have demonstrated commenting's negativity, documenting the “nasty effect” wherein negative comments color perceptions of neutral articles. Thus, we focused particular attention on cited barriers to implementing health-positive smoke-free policies. Methods: We collected online comments (N = 261) responding to popular media articles on smoking bans in inpatient psychiatry between 2013 and 2014 and conducted an inductive and exploratory qualitative content analysis. Results: Verifying previous studies documenting the prevalence of negative commenting, of the comments explicitly supporting or refuting psychiatry smoking bans, there were over twice as many con comments (n = 44) than pro (n = 18). Many commenters argued for access to outdoor smoking areas and warned of patient agitation and risk posed to care workers. Identified content themes included psychiatric medication and negative side effects, broken mental health systems and institutions, denigration of the health risks of tobacco in the context of mental illness, typical pro-smoking arguments about “smokers’ rights” and alternatives (including e-cigarettes), addiction, and stigma. Conclusions: The current findings provide a platform to begin to understand how people talk about mental health issues and smoking. Our analysis also raised complex issues concerning forces that impact U.S. patients with serious mental illness but over which they have little control, including medication, the U.S. health system, stigma, perceptions that life with chronic serious mental illness is not worth living, and psychological and physical pain of coping with mental illness. In consideration of identified barriers raised in opposition to smoking bans in inpatient psychiatry, efforts should emphasize patient stakeholder involvement; patient, visitor, and staff protection from smoke exposure; the effectiveness of nicotine replacement for managing withdrawal; and the lack of evidence that cigarettes are therapeutic.", "classified_sentences": [ { "sentence": "Objective: Individuals with mental health concerns are disproportionately affected by and suffer the negative consequences of tobacco use disorder, perhaps because smoking has historically been part of psychiatry's culture.", "category": "background" }, { "sentence": "In the early 1990s, psychiatric inpatient facilities were exempted from U.S. hospital smoking bans, in response to public outcry with national media attention.", "category": "background" }, { "sentence": "Almost 2 decades later, the current study characterizes online conversation about psychiatric hospital smoking bans.", "category": "background" }, { "sentence": "Previous commenting studies have demonstrated commenting's negativity, documenting the “nasty effect” wherein negative comments color perceptions of neutral articles.", "category": "background" }, { "sentence": "Thus, we focused particular attention on cited barriers to implementing health-positive smoke-free policies.", "category": "method" }, { "sentence": "Methods: We collected online comments (N = 261) responding to popular media articles on smoking bans in inpatient psychiatry between 2013 and 2014 and conducted an inductive and exploratory qualitative content analysis.", "category": "method" }, { "sentence": "Results: Verifying previous studies documenting the prevalence of negative commenting, of the comments explicitly supporting or refuting psychiatry smoking bans, there were over twice as many con comments (n = 44) than pro (n = 18).", "category": "result" }, { "sentence": "Many commenters argued for access to outdoor smoking areas and warned of patient agitation and risk posed to care workers.", "category": "result" }, { "sentence": "Identified content themes included psychiatric medication and negative side effects, broken mental health systems and institutions, denigration of the health risks of tobacco in the context of mental illness, typical pro-smoking arguments about “smokers’ rights” and alternatives (including e-cigarettes), addiction, and stigma.", "category": "result" }, { "sentence": "Conclusions: The current findings provide a platform to begin to understand how people talk about mental health issues and smoking.", "category": "result" }, { "sentence": "Our analysis also raised complex issues concerning forces that impact U.S. patients with serious mental illness but over which they have little control, including medication, the U.S. health system, stigma, perceptions that life with chronic serious mental illness is not worth living, and psychological and physical pain of coping with mental illness.", "category": "result" }, { "sentence": "In consideration of identified barriers raised in opposition to smoking bans in inpatient psychiatry, efforts should emphasize patient stakeholder involvement; patient, visitor, and staff protection from smoke exposure; the effectiveness of nicotine replacement for managing withdrawal; and the lack of evidence that cigarettes are therapeutic.", "category": "result" } ] }, { "paper_id": "2915436", "title": "End-to-end Congestion Control for Flows with Variable Packet Size", "abstract": "Current TCP-friendly congestion control mechanisms such as those used in TFRC adjust the packet rate in order to adapt to network conditions and obtain a throughput not exceeding that of a TCP connection operating under the same conditions. In an environment where the bottleneck resource is packet processing, this is the correct behavior. However, if the bottleneck resource is bandwidth, and flows may use packets of different sizes, resource sharing then depends on packet size and is no longer fair. Now for some applications, such as Internet telephony, it is more natural to adjust the packet size, while keeping the packet rate as constant as possible. In this paper we study the impact of variations in packet size on equation-based congestion control and propose methods to remove the throughput bias resulting from the use of different packet sizes. We investigate in detail the design space of the approaches by means of mathematical analysis and propose a number of possible designs. We evaluate these designs through simulation and conclude with some concrete proposals. Our findings can be used to design a TCP-friendly congestion control mechanism for applications that adjust packet size rather than rate, or that are forced to use a small packet size. We base our analysis on the TFRC protocol, but similar considerations also hold for other congestion control mechanisms.", "classified_sentences": [ { "sentence": "Current TCP-friendly congestion control mechanisms such as those used in TFRC adjust the packet rate in order to adapt to network conditions and obtain a throughput not exceeding that of a TCP connection operating under the same conditions.", "category": "background" }, { "sentence": "In an environment where the bottleneck resource is packet processing, this is the correct behavior.", "category": "background" }, { "sentence": "However, if the bottleneck resource is bandwidth, and flows may use packets of different sizes, resource sharing then depends on packet size and is no longer fair.", "category": "background" }, { "sentence": "Now for some applications, such as Internet telephony, it is more natural to adjust the packet size, while keeping the packet rate as constant as possible.", "category": "background" }, { "sentence": "In this paper we study the impact of variations in packet size on equation-based congestion control and propose methods to remove the throughput bias resulting from the use of different packet sizes.", "category": "method" }, { "sentence": "We investigate in detail the design space of the approaches by means of mathematical analysis and propose a number of possible designs.", "category": "method" }, { "sentence": "We evaluate these designs through simulation and conclude with some concrete proposals.", "category": "method" }, { "sentence": "Our findings can be used to design a TCP-friendly congestion control mechanism for applications that adjust packet size rather than rate, or that are forced to use a small packet size.", "category": "result" }, { "sentence": "We base our analysis on the TFRC protocol, but similar considerations also hold for other congestion control mechanisms.", "category": "background" } ] }, { "paper_id": "3108956", "title": "A Model of Saliency-Based Visual Attention for Rapid Scene Analysis", "abstract": "A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented. Multiscale image features are combined into a single topographical saliency map. A dynamical neural network then selects attended locations in order of decreasing saliency. The system breaks down the complex problem of scene understanding by rapidly selecting, in a computationally efficient manner, conspicuous locations to be analyzed in detail.", "classified_sentences": [ { "sentence": "A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented.", "category": "method" }, { "sentence": "Multiscale image features are combined into a single topographical saliency map.", "category": "method" }, { "sentence": "A dynamical neural network then selects attended locations in order of decreasing saliency.", "category": "method" }, { "sentence": "The system breaks down the complex problem of scene understanding by rapidly selecting, in a computationally efficient manner, conspicuous locations to be analyzed in detail.", "category": "result" } ] }, { "paper_id": "3146611", "title": "An Efficient Algorithm for Easy-First Non-Directional Dependency Parsing", "abstract": "We present a novel deterministic dependency parsing algorithm that attempts to create the easiest arcs in the dependency structure first in a non-directional manner. Traditional deterministic parsing algorithms are based on a shift-reduce framework: they traverse the sentence from left-to-right and, at each step, perform one of a possible set of actions, until a complete tree is built. A drawback of this approach is that it is extremely local: while decisions can be based on complex structures on the left, they can look only at a few words to the right. In contrast, our algorithm builds a dependency tree by iteratively selecting the best pair of neighbours to connect at each parsing step. This allows incorporation of features from already built structures both to the left and to the right of the attachment point. The parser learns both the attachment preferences and the order in which they should be performed. The result is a deterministic, best-first, O(nlogn) parser, which is significantly more accurate than best-first transition based parsers, and nears the performance of globally optimized parsing models.", "classified_sentences": [ { "sentence": "We present a novel deterministic dependency parsing algorithm that attempts to create the easiest arcs in the dependency structure first in a non-directional manner.", "category": "method" }, { "sentence": "Traditional deterministic parsing algorithms are based on a shift-reduce framework: they traverse the sentence from left-to-right and, at each step, perform one of a possible set of actions, until a complete tree is built.", "category": "background" }, { "sentence": "A drawback of this approach is that it is extremely local: while decisions can be based on complex structures on the left, they can look only at a few words to the right.", "category": "background" }, { "sentence": "In contrast, our algorithm builds a dependency tree by iteratively selecting the best pair of neighbours to connect at each parsing step.", "category": "method" }, { "sentence": "This allows incorporation of features from already built structures both to the left and to the right of the attachment point.", "category": "method" }, { "sentence": "The parser learns both the attachment preferences and the order in which they should be performed.", "category": "method" }, { "sentence": "The result is a deterministic, best-first, O(nlogn) parser, which is significantly more accurate than best-first transition based parsers, and nears the performance of globally optimized parsing models.", "category": "result" } ] }, { "paper_id": "3569355", "title": "Percutaneous cryoablation of breast tumours in patients with stable metastatic breast cancer: safety, feasibility and efficacy.", "abstract": "PURPOSE To evaluate safety, feasibility and efficacy of percutaneous cryoablation of breast tumours in patients with clinically stable metastatic breast cancer, and to compare the findings with reports on alternative procedures, namely surgery and local radiotherapy. METHODS 17 female patients (average age of 54.8 years ± 10.8; range 37-72) with primary breast tumour not surgically treated because of metastatic disease were included. Patients were treated for their primary lesion by percutaneous cryotherapy in period of stable disease. This minimally intervention was performed using ultrasound or CT scan guidance. All patients had clinical and breast-MRI evaluation before and at 1, 3, 6 and 12 months after the procedure. RESULTS All procedures were performed under local anaesthesia and technically successful. The mean largest diameter of the primary lesions was 16 ± 12 mm (size range 5-45 mm). In 15 patients, we obtained a complete regression of the primary breast lesion without recurrence during the follow-up period. Two patients with lesions measured at 40 and 45 mm had recurrence in follow up period. A second session of cryotherapy was performed for these 2 patients, not included in this study. Five patients had painful masses before cryotherapy. All were immediately relieved after the intervention and durably during all follow-up. CONCLUSION These results show that the cryoablation of primary breast lesions seems to be well suited to the palliative care of metastatic patients, particularly because of its good tolerance, low complication rate and ability to provide local or analgesic control. Advances in knowledge: Therapies are limited for these symptomatic patients at metastatic state of primary breast tumour. This study shows that cryoablation in palliative care is achievable in common practice, is effective in local control of the tumour and can provide immediate and long-term analgesic control.", "classified_sentences": [ { "sentence": "PURPOSE To evaluate safety, feasibility and efficacy of percutaneous cryoablation of breast tumours in patients with clinically stable metastatic breast cancer, and to compare the findings with reports on alternative procedures, namely surgery and local radiotherapy.", "category": "background" }, { "sentence": "METHODS 17 female patients (average age of 54.8 years ± 10.8; range 37-72) with primary breast tumour not surgically treated because of metastatic disease were included.", "category": "method" }, { "sentence": "Patients were treated for their primary lesion by percutaneous cryotherapy in period of stable disease.", "category": "method" }, { "sentence": "This minimally intervention was performed using ultrasound or CT scan guidance.", "category": "method" }, { "sentence": "All patients had clinical and breast-MRI evaluation before and at 1, 3, 6 and 12 months after the procedure.", "category": "method" }, { "sentence": "RESULTS All procedures were performed under local anaesthesia and technically successful.", "category": "result" }, { "sentence": "The mean largest diameter of the primary lesions was 16 ± 12 mm (size range 5-45 mm).", "category": "result" }, { "sentence": "In 15 patients, we obtained a complete regression of the primary breast lesion without recurrence during the follow-up period.", "category": "result" }, { "sentence": "Two patients with lesions measured at 40 and 45 mm had recurrence in follow up period.", "category": "result" }, { "sentence": "A second session of cryotherapy was performed for these 2 patients, not included in this study.", "category": "result" }, { "sentence": "Five patients had painful masses before cryotherapy.", "category": "result" }, { "sentence": "All were immediately relieved after the intervention and durably during all follow-up.", "category": "result" }, { "sentence": "CONCLUSION These results show that the cryoablation of primary breast lesions seems to be well suited to the palliative care of metastatic patients, particularly because of its good tolerance, low complication rate and ability to provide local or analgesic control.", "category": "result" }, { "sentence": "Advances in knowledge: Therapies are limited for these symptomatic patients at metastatic state of primary breast tumour.", "category": "background" }, { "sentence": "This study shows that cryoablation in palliative care is achievable in common practice, is effective in local control of the tumour and can provide immediate and long-term analgesic control.", "category": "result" } ] }, { "paper_id": "3670018", "title": "URLNet: Learning a URL Representation with Deep Learning for Malicious URL Detection", "abstract": "Malicious URLs host unsolicited content and are used to perpetrate cybercrimes. It is imperative to detect them in a timely manner. Traditionally, this is done through the usage of blacklists, which cannot be exhaustive, and cannot detect newly generated malicious URLs. To address this, recent years have witnessed several efforts to perform Malicious URL Detection using Machine Learning. The most popular and scalable approaches use lexical properties of the URL string by extracting Bag-of-words like features, followed by applying machine learning models such as SVMs. There are also other features designed by experts to improve the prediction performance of the model. These approaches suffer from several limitations: (i) Inability to effectively capture semantic meaning and sequential patterns in URL strings; (ii) Requiring substantial manual feature engineering; and (iii) Inability to handle unseen features and generalize to test data. To address these challenges, we propose URLNet, an end-to-end deep learning framework to learn a nonlinear URL embedding for Malicious URL Detection directly from the URL. Specifically, we apply Convolutional Neural Networks to both characters and words of the URL String to learn the URL embedding in a jointly optimized framework. This approach allows the model to capture several types of semantic information, which was not possible by the existing models. We also propose advanced word-embeddings to solve the problem of too many rare words observed in this task. We conduct extensive experiments on a large-scale dataset and show a significant performance gain over existing methods. We also conduct ablation studies to evaluate the performance of various components of URLNet.", "classified_sentences": [ { "sentence": "Malicious URLs host unsolicited content and are used to perpetrate cybercrimes.", "category": "background" }, { "sentence": "It is imperative to detect them in a timely manner.", "category": "background" }, { "sentence": "Traditionally, this is done through the usage of blacklists, which cannot be exhaustive, and cannot detect newly generated malicious URLs.", "category": "background" }, { "sentence": "To address this, recent years have witnessed several efforts to perform Malicious URL Detection using Machine Learning.", "category": "background" }, { "sentence": "The most popular and scalable approaches use lexical properties of the URL string by extracting Bag-of-words like features, followed by applying machine learning models such as SVMs.", "category": "method" }, { "sentence": "There are also other features designed by experts to improve the prediction performance of the model.", "category": "method" }, { "sentence": "These approaches suffer from several limitations: (i) Inability to effectively capture semantic meaning and sequential patterns in URL strings; (ii) Requiring substantial manual feature engineering; and (iii) Inability to handle unseen features and generalize to test data.", "category": "background" }, { "sentence": "To address these challenges, we propose URLNet, an end-to-end deep learning framework to learn a nonlinear URL embedding for Malicious URL Detection directly from the URL.", "category": "method" }, { "sentence": "Specifically, we apply Convolutional Neural Networks to both characters and words of the URL String to learn the URL embedding in a jointly optimized framework.", "category": "method" }, { "sentence": "This approach allows the model to capture several types of semantic information, which was not possible by the existing models.", "category": "method" }, { "sentence": "We also propose advanced word-embeddings to solve the problem of too many rare words observed in this task.", "category": "method" }, { "sentence": "We conduct extensive experiments on a large-scale dataset and show a significant performance gain over existing methods.", "category": "result" }, { "sentence": "We also conduct ablation studies to evaluate the performance of various components of URLNet.", "category": "result" } ] }, { "paper_id": "3693305", "title": "A Survey on Learning to Hash", "abstract": "Nearest neighbor search is a problem of finding the data points from the database such that the distances from them to the query point are the smallest. Learning to hash is one of the major solutions to this problem and has been widely studied recently. In this paper, we present a comprehensive survey of the learning to hash algorithms, categorize them according to the manners of preserving the similarities into: pairwise similarity preserving, multiwise similarity preserving, implicit similarity preserving, as well as quantization, and discuss their relations. We separate quantization from pairwise similarity preserving as the objective function is very different though quantization, as we show, can be derived from preserving the pairwise similarities. In addition, we present the evaluation protocols, and the general performance analysis, and point out that the quantization algorithms perform superiorly in terms of search accuracy, search time cost, and space cost. Finally, we introduce a few emerging topics.", "classified_sentences": [ { "sentence": "Nearest neighbor search is a problem of finding the data points from the database such that the distances from them to the query point are the smallest.", "category": "background" }, { "sentence": "Learning to hash is one of the major solutions to this problem and has been widely studied recently.", "category": "background" }, { "sentence": "In this paper, we present a comprehensive survey of the learning to hash algorithms, categorize them according to the manners of preserving the similarities into: pairwise similarity preserving, multiwise similarity preserving, implicit similarity preserving, as well as quantization, and discuss their relations.", "category": "method" }, { "sentence": "We separate quantization from pairwise similarity preserving as the objective function is very different though quantization, as we show, can be derived from preserving the pairwise similarities.", "category": "method" }, { "sentence": "In addition, we present the evaluation protocols, and the general performance analysis, and point out that the quantization algorithms perform superiorly in terms of search accuracy, search time cost, and space cost.", "category": "result" }, { "sentence": "Finally, we introduce a few emerging topics.", "category": "result" } ] }, { "paper_id": "4600117", "title": "A Fiber Optic Interferometric Sensor Platform for Determining Gas Diffusivity in Zeolite Films", "abstract": "Fiber optic interferometer (FOI) sensors have been fabricated by directly growing pure-silica MFI-type zeolite (i.e., silicalite) films on straight-cut endfaces of single-mode communication optical fibers. The FOI sensor has been demonstrated for determining molecular diffusivity in the zeolite by monitoring the temporal response of light interference from the zeolite film during the dynamic process of gas adsorption. The optical thickness of the zeolite film depends on the amount of gas adsorption that causes the light interference to shift upon loading molecules into the zeolitic channels. Thus, the time-dependence of the optical signal reflected from the coated zeolite film can represent the adsorption uptake curve, which allows computation of the diffusivity using models derived from the Fick’s Law equations. In this study, the diffusivity of isobutane in silicalite has been determined by the new FOI sensing method, and the results are in good agreement with literature values obtained by various conventional macroscopic techniques. The FOI sensor platform, because of its robustness and small size, could be useful for studying molecular diffusion in zeolitic materials under conditions that are inaccessible to the existing techniques.", "classified_sentences": [ { "sentence": "Fiber optic interferometer (FOI) sensors have been fabricated by directly growing pure-silica MFI-type zeolite (i.e., silicalite) films on straight-cut endfaces of single-mode communication optical fibers.", "category": "method" }, { "sentence": "The FOI sensor has been demonstrated for determining molecular diffusivity in the zeolite by monitoring the temporal response of light interference from the zeolite film during the dynamic process of gas adsorption.", "category": "method" }, { "sentence": "The optical thickness of the zeolite film depends on the amount of gas adsorption that causes the light interference to shift upon loading molecules into the zeolitic channels.", "category": "method" }, { "sentence": "Thus, the time-dependence of the optical signal reflected from the coated zeolite film can represent the adsorption uptake curve, which allows computation of the diffusivity using models derived from the Fick’s Law equations.", "category": "method" }, { "sentence": "In this study, the diffusivity of isobutane in silicalite has been determined by the new FOI sensing method, and the results are in good agreement with literature values obtained by various conventional macroscopic techniques.", "category": "result" }, { "sentence": "The FOI sensor platform, because of its robustness and small size, could be useful for studying molecular diffusion in zeolitic materials under conditions that are inaccessible to the existing techniques.", "category": "background" } ] }, { "paper_id": "4808651", "title": "Renormalization for a Scalar Field in an External Scalar Potential", "abstract": "The Pauli--Villars regularization procedure confirms and sharpens the conclusions reached previously by covariant point splitting. The divergences in the stress tensor of a quantized scalar field interacting with a static scalar potential are isolated into a three-parameter local, covariant functional of the background potential. These divergences can be naturally absorbed into coupling constants of the potential, regarded as a dynamical object in its own right; here this is demonstrated in detail for two different models of the field-potential coupling. here is a residual dependence on the logarithm of the potential, reminiscent of the renormalization group in fully interacting quantum field theories; these terms are finite but numerically dependent on an arbitrary mass or length parameter, which is purely a matter of convention. This work is one step in a program to elucidate boundary divergences by replacing a sharp boundary by a steeply rising smooth potential.", "classified_sentences": [ { "sentence": "The Pauli--Villars regularization procedure confirms and sharpens the conclusions reached previously by covariant point splitting.", "category": "method" }, { "sentence": "The divergences in the stress tensor of a quantized scalar field interacting with a static scalar potential are isolated into a three-parameter local, covariant functional of the background potential.", "category": "result" }, { "sentence": "These divergences can be naturally absorbed into coupling constants of the potential, regarded as a dynamical object in its own right; here this is demonstrated in detail for two different models of the field-potential coupling.", "category": "result" }, { "sentence": "here is a residual dependence on the logarithm of the potential, reminiscent of the renormalization group in fully interacting quantum field theories; these terms are finite but numerically dependent on an arbitrary mass or length parameter, which is purely a matter of convention.", "category": "background" }, { "sentence": "This work is one step in a program to elucidate boundary divergences by replacing a sharp boundary by a steeply rising smooth potential.", "category": "background" } ] }, { "paper_id": "5767210", "title": "The Use of Ensembles to Identify Forecasts with Small and Large Uncertainty", "abstract": "Abstract In the past decade ensemble forecasting has developed into an integral part of numerical weather prediction. Flow-dependent forecast probability distributions can be readily generated from an ensemble, allowing for the identification of forecast cases with high and low uncertainty. The ability of the NCEP ensemble to distinguish between high and low uncertainty forecast cases is studied here quantitatively. Ensemble mode forecasts, along with traditional higher-resolution control forecasts, are verified in terms of predicting the probability of the true state being in 1 of 10 climatologically equally likely 500-hPa height intervals. A stratification of the forecast cases by the degree of overall agreement among the ensemble members reveals great differences in forecast performance between the cases identified by the ensemble as the least and most uncertain. A new ensemble-based forecast product, the “relative measure of predictability,” is introduced to identify forecasts with below and above ave.", "classified_sentences": [ { "sentence": "Abstract In the past decade ensemble forecasting has developed into an integral part of numerical weather prediction.", "category": "background" }, { "sentence": "Flow-dependent forecast probability distributions can be readily generated from an ensemble, allowing for the identification of forecast cases with high and low uncertainty.", "category": "method" }, { "sentence": "The ability of the NCEP ensemble to distinguish between high and low uncertainty forecast cases is studied here quantitatively.", "category": "method" }, { "sentence": "Ensemble mode forecasts, along with traditional higher-resolution control forecasts, are verified in terms of predicting the probability of the true state being in 1 of 10 climatologically equally likely 500-hPa height intervals.", "category": "method" }, { "sentence": "A stratification of the forecast cases by the degree of overall agreement among the ensemble members reveals great differences in forecast performance between the cases identified by the ensemble as the least and most uncertain.", "category": "result" }, { "sentence": "A new ensemble-based forecast product, the “relative measure of predictability,” is introduced to identify forecasts with below and above ave.", "category": "method" } ] }, { "paper_id": "5981544", "title": "Saliency detection using maximum symmetric surround", "abstract": "Detection of visually salient image regions is useful for applications like object segmentation, adaptive compression, and object recognition. Recently, full-resolution salient maps that retain well-defined boundaries have attracted attention. In these maps, boundaries are preserved by retaining substantially more frequency content from the original image than older techniques. However, if the salient regions comprise more than half the pixels of the image, or if the background is complex, the background gets highlighted instead of the salient object. In this paper, we introduce a method for salient region detection that retains the advantages of such saliency maps while overcoming their shortcomings. Our method exploits features of color and luminance, is simple to implement and is computationally efficient. We compare our algorithm to six state-of-the-art salient region detection methods using publicly available ground truth. Our method outperforms the six algorithms by achieving both higher precision and better recall. We also show application of our saliency maps in an automatic salient object segmentation scheme using graph-cuts.", "classified_sentences": [ { "sentence": "Detection of visually salient image regions is useful for applications like object segmentation, adaptive compression, and object recognition.", "category": "background" }, { "sentence": "Recently, full-resolution salient maps that retain well-defined boundaries have attracted attention.", "category": "background" }, { "sentence": "In these maps, boundaries are preserved by retaining substantially more frequency content from the original image than older techniques.", "category": "background" }, { "sentence": "However, if the salient regions comprise more than half the pixels of the image, or if the background is complex, the background gets highlighted instead of the salient object.", "category": "background" }, { "sentence": "In this paper, we introduce a method for salient region detection that retains the advantages of such saliency maps while overcoming their shortcomings.", "category": "method" }, { "sentence": "Our method exploits features of color and luminance, is simple to implement and is computationally efficient.", "category": "method" }, { "sentence": "We compare our algorithm to six state-of-the-art salient region detection methods using publicly available ground truth.", "category": "method" }, { "sentence": "Our method outperforms the six algorithms by achieving both higher precision and better recall.", "category": "result" }, { "sentence": "We also show application of our saliency maps in an automatic salient object segmentation scheme using graph-cuts.", "category": "result" } ] }, { "paper_id": "6061457", "title": "Contribution of stratospheric ozone to the interannual variability of tropospheric ozone in the northern extratropics", "abstract": "[1] We examined the role of variability in the input of stratospheric ozone on the interannual variability of tropospheric ozone in the northern extratropics using correlations of monthly ozone anomalies for the lower stratosphere and the troposphere. We used output from a multiyear simulation of the NASA Goddard Space Flight Center (GSFC) Chemistry and Transport Model (CTM), and evaluated model results using ozonesonde data. The GSFC CTM explicitly calculates stratospheric ozone and simulates separate tracers of stratospheric and tropospheric ozone (O3-strat and O3-trop, respectively). The climatological seasonal cycle of ozone shows that O3-strat contributes significantly to the spring maximum of ozone at 500 hPa, ∼40% at high latitudes and ∼30% at midlatitudes. We find large regional differences in the correlation of ozone in the lower stratosphere and troposphere in the model that are supported by the ozonesonde data. Highest correlations are found from the eastern Atlantic to Europe, from the eastern Pacific to the western United States, and over the polar regions, in winter-spring. This spatial pattern is due to the input of O3-strat into the troposphere. The distribution and time lag of the correlations (highest with no lag for midlatitudes and a 1–2 month lag for polar regions) are consistent with the dynamical indicators of stratosphere-troposphere exchange (STE), such as storm tracks in the midlatitudes and slow descending motion in the polar region. Our simple approach can be widely applied to diagnose the effect of STE on tropospheric ozone.", "classified_sentences": [ { "sentence": "We examined the role of variability in the input of stratospheric ozone on the interannual variability of tropospheric ozone in the northern extratropics using correlations of monthly ozone anomalies for the lower stratosphere and the troposphere.", "category": "method" }, { "sentence": "We used output from a multiyear simulation of the NASA Goddard Space Flight Center (GSFC) Chemistry and Transport Model (CTM), and evaluated model results using ozonesonde data.", "category": "method" }, { "sentence": "The GSFC CTM explicitly calculates stratospheric ozone and simulates separate tracers of stratospheric and tropospheric ozone (O3-strat and O3-trop, respectively).", "category": "method" }, { "sentence": "The climatological seasonal cycle of ozone shows that O3-strat contributes significantly to the spring maximum of ozone at 500 hPa, ∼40% at high latitudes and ∼30% at midlatitudes.", "category": "background" }, { "sentence": "We find large regional differences in the correlation of ozone in the lower stratosphere and troposphere in the model that are supported by the ozonesonde data.", "category": "result" }, { "sentence": "Highest correlations are found from the eastern Atlantic to Europe, from the eastern Pacific to the western United States, and over the polar regions, in winter-spring.", "category": "result" }, { "sentence": "This spatial pattern is due to the input of O3-strat into the troposphere.", "category": "result" }, { "sentence": "The distribution and time lag of the correlations (highest with no lag for midlatitudes and a 1–2 month lag for polar regions) are consistent with the dynamical indicators of stratosphere-troposphere exchange (STE), such as storm tracks in the midlatitudes and slow descending motion in the polar region.", "category": "result" }, { "sentence": "Our simple approach can be widely applied to diagnose the effect of STE on tropospheric ozone.", "category": "result" } ] }, { "paper_id": "6069002", "title": "Modelling of sensory and instrumental texture parameters in processed cheese by near infrared reflectance spectroscopy", "abstract": "This study investigated the application of near infrared (NIR) reflectance spectroscopy to the measurement of texture (sensory and instrumental) in experimental processed cheese samples. Spectra (750 to 2498 nm) of cheeses were recorded after 2 and 4 weeks storage at 4 °C. Trained assessors evaluated 9 sensory properties, a texture profile analyser (TPA) was used to record 5 instrumental parameters and cheese ‘meltability’ was measured by computer vision. Predictive models for sensory and instrumental texture parameters were developed using partial least squares regression on raw or pre-treated spectral data. Sensory attributes and instrumental texture measurements were modelled with sufficient accuracy to recommend the use of NIR reflectance spectroscopy for routine quality assessment of processed cheese.", "classified_sentences": [ { "sentence": "This study investigated the application of near infrared (NIR) reflectance spectroscopy to the measurement of texture (sensory and instrumental) in experimental processed cheese samples.", "category": "background" }, { "sentence": "Spectra (750 to 2498 nm) of cheeses were recorded after 2 and 4 weeks storage at 4 °C.", "category": "method" }, { "sentence": "Trained assessors evaluated 9 sensory properties, a texture profile analyser (TPA) was used to record 5 instrumental parameters and cheese ‘meltability’ was measured by computer vision.", "category": "method" }, { "sentence": "Predictive models for sensory and instrumental texture parameters were developed using partial least squares regression on raw or pre-treated spectral data.", "category": "method" }, { "sentence": "Sensory attributes and instrumental texture measurements were modelled with sufficient accuracy to recommend the use of NIR reflectance spectroscopy for routine quality assessment of processed cheese.", "category": "result" } ] }, { "paper_id": "6235360", "title": "Improving Coreference Resolution by Learning Entity-Level Distributed Representations", "abstract": "A long-standing challenge in coreference resolution has been the incorporation of entity-level information - features defined over clusters of mentions instead of mention pairs. We present a neural network based coreference system that produces high-dimensional vector representations for pairs of coreference clusters. Using these representations, our system learns when combining clusters is desirable. We train the system with a learning-to-search algorithm that teaches it which local decisions (cluster merges) will lead to a high-scoring final coreference partition. The system substantially outperforms the current state-of-the-art on the English and Chinese portions of the CoNLL 2012 Shared Task dataset despite using few hand-engineered features.", "classified_sentences": [ { "sentence": "A long-standing challenge in coreference resolution has been the incorporation of entity-level information - features defined over clusters of mentions instead of mention pairs.", "category": "background" }, { "sentence": "We present a neural network based coreference system that produces high-dimensional vector representations for pairs of coreference clusters.", "category": "method" }, { "sentence": "Using these representations, our system learns when combining clusters is desirable.", "category": "method" }, { "sentence": "We train the system with a learning-to-search algorithm that teaches it which local decisions (cluster merges) will lead to a high-scoring final coreference partition.", "category": "method" }, { "sentence": "The system substantially outperforms the current state-of-the-art on the English and Chinese portions of the CoNLL 2012 Shared Task dataset despite using few hand-engineered features.", "category": "result" } ] }, { "paper_id": "6455915", "title": "Deeper syntax for better semantic parsing", "abstract": "Syntax plays an important role in the task of predicting the semantic structure of a sentence. But syntactic phenomena such as alternations, control and raising tend to obfuscate the relation between syntax and semantics. In this paper we predict the semantic structure of a sentence using a deeper syntax than what is usually done. This deep syntactic representation abstracts away from purely syntactic phenomena and proposes a structural organization of the sentence that is closer to the semantic representation. Experiments conducted on a French corpus annotated with semantic frames showed that a semantic parser reaches better performances with such a deep syntactic input.", "classified_sentences": [ { "sentence": "Syntax plays an important role in the task of predicting the semantic structure of a sentence.", "category": "background" }, { "sentence": "But syntactic phenomena such as alternations, control and raising tend to obfuscate the relation between syntax and semantics.", "category": "background" }, { "sentence": "In this paper we predict the semantic structure of a sentence using a deeper syntax than what is usually done.", "category": "method" }, { "sentence": "This deep syntactic representation abstracts away from purely syntactic phenomena and proposes a structural organization of the sentence that is closer to the semantic representation.", "category": "method" }, { "sentence": "Experiments conducted on a French corpus annotated with semantic frames showed that a semantic parser reaches better performances with such a deep syntactic input.", "category": "result" } ] }, { "paper_id": "6493988", "title": "Corpus Based PP Attachment Ambiguity Resolution with a Semantic Dictionary", "abstract": "This paper deals with two important ambiguities of natural language: prepositional phrase attachment and word sense ambiguity. We propose a new supervised learning method for PPattachment based on a semantically tagged corpus. Because any sufficiently big sense-tagged corpus does not exist, we also propose a new unsupervised context based word sense disambiguation algorithm which amends the training corpus for the PP attachment by word sense tags. We present the results of our approach and evaluate the achieved PP attachment accuracy in comparison with other methods.", "classified_sentences": [ { "sentence": "This paper deals with two important ambiguities of natural language: prepositional phrase attachment and word sense ambiguity.", "category": "background" }, { "sentence": "We propose a new supervised learning method for PPattachment based on a semantically tagged corpus.", "category": "method" }, { "sentence": "Because any sufficiently big sense-tagged corpus does not exist, we also propose a new unsupervised context based word sense disambiguation algorithm which amends the training corpus for the PP attachment by word sense tags.", "category": "method" }, { "sentence": "We present the results of our approach and evaluate the achieved PP attachment accuracy in comparison with other methods.", "category": "result" } ] }, { "paper_id": "6946103", "title": "Embedding Lexical Features via Low-Rank Tensors", "abstract": "Modern NLP models rely heavily on engineered features, which often combine word and contextual information into complex lexical features. Such combination results in large numbers of features, which can lead to over-fitting. We present a new model that represents complex lexical features---comprised of parts for words, contextual information and labels---in a tensor that captures conjunction information among these parts. We apply low-rank tensor approximations to the corresponding parameter tensors to reduce the parameter space and improve prediction speed. Furthermore, we investigate two methods for handling features that include $n$-grams of mixed lengths. Our model achieves state-of-the-art results on tasks in relation extraction, PP-attachment, and preposition disambiguation.", "classified_sentences": [ { "sentence": "Modern NLP models rely heavily on engineered features, which often combine word and contextual information into complex lexical features.", "category": "background" }, { "sentence": "Such combination results in large numbers of features, which can lead to over-fitting.", "category": "background" }, { "sentence": "We present a new model that represents complex lexical features---comprised of parts for words, contextual information and labels---in a tensor that captures conjunction information among these parts.", "category": "method" }, { "sentence": "We apply low-rank tensor approximations to the corresponding parameter tensors to reduce the parameter space and improve prediction speed.", "category": "method" }, { "sentence": "Furthermore, we investigate two methods for handling features that include $n$-grams of mixed lengths.", "category": "method" }, { "sentence": "Our model achieves state-of-the-art results on tasks in relation extraction, PP-attachment, and preposition disambiguation.", "category": "result" } ] }, { "paper_id": "7019971", "title": "CD28 expression is required after T cell priming for helper T cell responses and protective immunity to infection", "abstract": "The co-stimulatory molecule CD28 is essential for activation of helper T cells. Despite this critical role, it is not known whether CD28 has functions in maintaining T cell responses following activation. To determine the role for CD28 after T cell priming, we generated a strain of mice where CD28 is removed from CD4+ T cells after priming. We show that continued CD28 expression is important for effector CD4+ T cells following infection; maintained CD28 is required for the expansion of T helper type 1 cells, and for the differentiation and maintenance of T follicular helper cells during viral infection. Persistent CD28 is also required for clearance of the bacterium Citrobacter rodentium from the gastrointestinal tract. Together, this study demonstrates that CD28 persistence is required for helper T cell polarization in response to infection, describing a novel function for CD28 that is distinct from its role in T cell priming. DOI: http://dx.doi. org/10.7554/eLife.03180. 001", "classified_sentences": [ { "sentence": "The co-stimulatory molecule CD28 is essential for activation of helper T cells.", "category": "background" }, { "sentence": "Despite this critical role, it is not known whether CD28 has functions in maintaining T cell responses following activation.", "category": "background" }, { "sentence": "To determine the role for CD28 after T cell priming, we generated a strain of mice where CD28 is removed from CD4+ T cells after priming.", "category": "method" }, { "sentence": "We show that continued CD28 expression is important for effector CD4+ T cells following infection; maintained CD28 is required for the expansion of T helper type 1 cells, and for the differentiation and maintenance of T follicular helper cells during viral infection.", "category": "result" }, { "sentence": "Persistent CD28 is also required for clearance of the bacterium Citrobacter rodentium from the gastrointestinal tract.", "category": "result" }, { "sentence": "Together, this study demonstrates that CD28 persistence is required for helper T cell polarization in response to infection, describing a novel function for CD28 that is distinct from its role in T cell priming.", "category": "result" } ] }, { "paper_id": "7054314", "title": "Molecular mimicry between the human immunodeficiency virus type 1 gp120 V3 loop and human brain proteins", "abstract": "Immunologically cross-reactive proteins in the human brain that resemble the V3 loop of human immunodeficiency virus type 1 (HIV-1) gp120 have been identified. When several homogenized tissues from normal brains were used, a monoclonal antibody raised against amino acids 308 to 320 of the V3 loop reacted with three prominent human brain proteins (HBP) of 35, 55, and 110 kDa. Among the three, the 55-kDa HBP appears to be specific to the central nervous system. These results indicate that the V3 loop of HIV-1 gp120 shares an epitope with HBP. An immune response to the V3 loop that generates cross-reactive antibodies to cellular proteins may be an autoimmune mechanism by which HIV-1 can damage the central nervous system.", "classified_sentences": [ { "sentence": "Immunologically cross-reactive proteins in the human brain that resemble the V3 loop of human immunodeficiency virus type 1 (HIV-1) gp120 have been identified.", "category": "background" }, { "sentence": "When several homogenized tissues from normal brains were used, a monoclonal antibody raised against amino acids 308 to 320 of the V3 loop reacted with three prominent human brain proteins (HBP) of 35, 55, and 110 kDa.", "category": "method" }, { "sentence": "Among the three, the 55-kDa HBP appears to be specific to the central nervous system.", "category": "result" }, { "sentence": "These results indicate that the V3 loop of HIV-1 gp120 shares an epitope with HBP.", "category": "result" }, { "sentence": "An immune response to the V3 loop that generates cross-reactive antibodies to cellular proteins may be an autoimmune mechanism by which HIV-1 can damage the central nervous system.", "category": "result" } ] }, { "paper_id": "7302715", "title": "A Methodology for Exposing Process Risk in Emergent System Properties", "abstract": "Determining whether software and systems achieve desired emergent properties, such as safety, reliability, or security, requires an analysis of the system as a whole. This requires the system to be in the latter stages of development, when changes are difficult and costly to implement. In this paper, we propose the Process Risk Assessment (PRA) methodology for analyzing and evaluating such emergent properties earlier in the development cycle. Properties such as safety and reliability result from one or more development processes put in place to help achieve those properties. The PRA method analyzes artifacts from these processes (e.g., designs pertaining to reliability, or safety analysis reports) to determine: 1) whether the process itself is appropriate for achieving the desired property; and 2) whether the process is being followed appropriately. From PRA analysis, process risk can be quantified to indicate whether the system will have the desired properties. We applied this method to evaluate one emergent property, software safety, during the early stages of the development lifecycle for a network-centric, Department of Defense system-of-systems and several NASA spaceflight projects. We analyzed the safety processes implemented on these projects and their resulting artifacts. The PRA methodology identified potential risks in the software safety process and provided feedback to the projects for reducing these risks.", "classified_sentences": [ { "sentence": "Determining whether software and systems achieve desired emergent properties, such as safety, reliability, or security, requires an analysis of the system as a whole.", "category": "background" }, { "sentence": "This requires the system to be in the latter stages of development, when changes are difficult and costly to implement.", "category": "background" }, { "sentence": "In this paper, we propose the Process Risk Assessment (PRA) methodology for analyzing and evaluating such emergent properties earlier in the development cycle.", "category": "method" }, { "sentence": "Properties such as safety and reliability result from one or more development processes put in place to help achieve those properties.", "category": "background" }, { "sentence": "The PRA method analyzes artifacts from these processes (e.g., designs pertaining to reliability, or safety analysis reports) to determine: 1) whether the process itself is appropriate for achieving the desired property; and 2) whether the process is being followed appropriately.", "category": "method" }, { "sentence": "From PRA analysis, process risk can be quantified to indicate whether the system will have the desired properties.", "category": "method" }, { "sentence": "We applied this method to evaluate one emergent property, software safety, during the early stages of the development lifecycle for a network-centric, Department of Defense system-of-systems and several NASA spaceflight projects.", "category": "result" }, { "sentence": "We analyzed the safety processes implemented on these projects and their resulting artifacts.", "category": "result" }, { "sentence": "The PRA methodology identified potential risks in the software safety process and provided feedback to the projects for reducing these risks.", "category": "result" } ] }, { "paper_id": "8017095", "title": "Full-frame video stabilization", "abstract": "Video stabilization is an important video enhancement technology which aims at removing annoying shaky motion from videos. We propose a practical and robust approach of video stabilization that produces full-frame stabilized videos with good visual quality. While most previous methods end up with producing low resolution stabilized videos, our completion method can produce full-frame videos by naturally filling in missing image parts by locally aligning image data of neighboring frames. To achieve this, motion inpainting is proposed to enforce spatial and temporal consistency of the completion in both static and dynamic image areas. In addition, image quality in the stabilized video is enhanced with a new practical deblurring algorithm. Instead of estimating point spread functions, our method transfers and interpolates sharper image pixels of neighbouring frames to increase the sharpness of the frame. The proposed video completion and deblurring methods enabled us to develop a complete video stabilizer which can naturally keep the original image quality in the stabilized videos. The effectiveness of our method is confirmed by extensive experiments over a wide variety of videos.", "classified_sentences": [ { "sentence": "Video stabilization is an important video enhancement technology which aims at removing annoying shaky motion from videos.", "category": "background" }, { "sentence": "We propose a practical and robust approach of video stabilization that produces full-frame stabilized videos with good visual quality.", "category": "method" }, { "sentence": "While most previous methods end up with producing low resolution stabilized videos, our completion method can produce full-frame videos by naturally filling in missing image parts by locally aligning image data of neighboring frames.", "category": "method" }, { "sentence": "To achieve this, motion inpainting is proposed to enforce spatial and temporal consistency of the completion in both static and dynamic image areas.", "category": "method" }, { "sentence": "In addition, image quality in the stabilized video is enhanced with a new practical deblurring algorithm.", "category": "method" }, { "sentence": "Instead of estimating point spread functions, our method transfers and interpolates sharper image pixels of neighbouring frames to increase the sharpness of the frame.", "category": "method" }, { "sentence": "The proposed video completion and deblurring methods enabled us to develop a complete video stabilizer which can naturally keep the original image quality in the stabilized videos.", "category": "method" }, { "sentence": "The effectiveness of our method is confirmed by extensive experiments over a wide variety of videos.", "category": "result" } ] }, { "paper_id": "8629021", "title": "Detecting Malicious Web Links and Identifying Their Attack Types", "abstract": "Malicious URLs have been widely used to mount various cyber attacks including spamming, phishing and malware. Detection of malicious URLs and identification of threat types are critical to thwart these attacks. Knowing the type of a threat enables estimation of severity of the attack and helps adopt an effective countermeasure. Existing methods typically detect malicious URLs of a single attack type. In this paper, we propose method using machine learning to detect malicious URLs of all the popular attack types and identify the nature of attack a malicious URL attempts to launch. Our method uses a variety of discriminative features including textual properties, link structures, webpage contents, DNS information, and network traffic. Many of these features are novel and highly effective. Our experimental studies with 40,000 benign URLs and 32,000 malicious URLs obtained from real-life Internet sources show that our method delivers a superior performance: the accuracy was over 98% in detecting malicious URLs and over 93% in identifying attack types. We also report our studies on the effectiveness of each group of discriminative features, and discuss their evadability.", "classified_sentences": [ { "sentence": "Malicious URLs have been widely used to mount various cyber attacks including spamming, phishing and malware.", "category": "background" }, { "sentence": "Detection of malicious URLs and identification of threat types are critical to thwart these attacks.", "category": "background" }, { "sentence": "Knowing the type of a threat enables estimation of severity of the attack and helps adopt an effective countermeasure.", "category": "background" }, { "sentence": "Existing methods typically detect malicious URLs of a single attack type.", "category": "background" }, { "sentence": "In this paper, we propose method using machine learning to detect malicious URLs of all the popular attack types and identify the nature of attack a malicious URL attempts to launch.", "category": "method" }, { "sentence": "Our method uses a variety of discriminative features including textual properties, link structures, webpage contents, DNS information, and network traffic.", "category": "method" }, { "sentence": "Many of these features are novel and highly effective.", "category": "method" }, { "sentence": "Our experimental studies with 40,000 benign URLs and 32,000 malicious URLs obtained from real-life Internet sources show that our method delivers a superior performance: the accuracy was over 98% in detecting malicious URLs and over 93% in identifying attack types.", "category": "result" }, { "sentence": "We also report our studies on the effectiveness of each group of discriminative features, and discuss their evadability.", "category": "result" } ] }, { "paper_id": "9496819", "title": "All About VLAD", "abstract": "The objective of this paper is large scale object instance retrieval, given a query image. A starting point of such systems is feature detection and description, for example using SIFT. The focus of this paper, however, is towards very large scale retrieval where, due to storage requirements, very compact image descriptors are required and no information about the original SIFT descriptors can be accessed directly at run time. We start from VLAD, the state-of-the art compact descriptor introduced by Jegou et al. for this purpose, and make three novel contributions: first, we show that a simple change to the normalization method significantly improves retrieval performance, second, we show that vocabulary adaptation can substantially alleviate problems caused when images are added to the dataset after initial vocabulary learning. These two methods set a new state-of-the-art over all benchmarks investigated here for both mid-dimensional (20k-D to 30k-D) and small (128-D) descriptors. Our third contribution is a multiple spatial VLAD representation, MultiVLAD, that allows the retrieval and localization of objects that only extend over a small part of an image (again without requiring use of the original image SIFT descriptors).", "classified_sentences": [ { "sentence": "The objective of this paper is large scale object instance retrieval, given a query image.", "category": "background" }, { "sentence": "A starting point of such systems is feature detection and description, for example using SIFT.", "category": "background" }, { "sentence": "The focus of this paper, however, is towards very large scale retrieval where, due to storage requirements, very compact image descriptors are required and no information about the original SIFT descriptors can be accessed directly at run time.", "category": "background" }, { "sentence": "We start from VLAD, the state-of-the-art compact descriptor introduced by Jegou et al. for this purpose, and make three novel contributions: first, we show that a simple change to the normalization method significantly improves retrieval performance, second, we show that vocabulary adaptation can substantially alleviate problems caused when images are added to the dataset after initial vocabulary learning.", "category": "method" }, { "sentence": "These two methods set a new state-of-the-art over all benchmarks investigated here for both mid-dimensional (20k-D to 30k-D) and small (128-D) descriptors.", "category": "result" }, { "sentence": "Our third contribution is a multiple spatial VLAD representation, MultiVLAD, that allows the retrieval and localization of objects that only extend over a small part of an image (again without requiring use of the original image SIFT descriptors).", "category": "method" } ] }, { "paper_id": "10426543", "title": "Using Self-Organizing Maps for Fraud Prediction at Online Auction Sites", "abstract": "Online auction sites have to deal with a enormous amount of product listings, of which a fraction is fraudulent. Although small in proportion, fraudulent listings are costly for site operators, buyers and legitimate sellers. Fraud prediction in this scenario can benefit significantly from machine learning techniques, although interpretability of model predictions is a concern. In this work we extend an unsupervised learning technique -- Self-Organizing Maps -- to use labeled data for binary classification under a constraint on the proportion of false positives. The resulting technique was applied to the prediction of non-delivery fraud, achieving good results while being easier to interpret.", "classified_sentences": [ { "sentence": "Online auction sites have to deal with a enormous amount of product listings, of which a fraction is fraudulent.", "category": "background" }, { "sentence": "Although small in proportion, fraudulent listings are costly for site operators, buyers and legitimate sellers.", "category": "background" }, { "sentence": "Fraud prediction in this scenario can benefit significantly from machine learning techniques, although interpretability of model predictions is a concern.", "category": "background" }, { "sentence": "In this work we extend an unsupervised learning technique -- Self-Organizing Maps -- to use labeled data for binary classification under a constraint on the proportion of false positives.", "category": "method" }, { "sentence": "The resulting technique was applied to the prediction of non-delivery fraud, achieving good results while being easier to interpret.", "category": "result" } ] }, { "paper_id": "10486135", "title": "eXpose: A Character-Level Convolutional Neural Network with Embeddings For Detecting Malicious URLs, File Paths and Registry Keys", "abstract": "For years security machine learning research has promised to obviate the need for signature based detection by automatically learning to detect indicators of attack. Unfortunately, this vision hasn't come to fruition: in fact, developing and maintaining today's security machine learning systems can require engineering resources that are comparable to that of signature-based detection systems, due in part to the need to develop and continuously tune the \"features\" these machine learning systems look at as attacks evolve. Deep learning, a subfield of machine learning, promises to change this by operating on raw input signals and automating the process of feature design and extraction. In this paper we propose the eXpose neural network, which uses a deep learning approach we have developed to take generic, raw short character strings as input (a common case for security inputs, which include artifacts like potentially malicious URLs, file paths, named pipes, named mutexes, and registry keys), and learns to simultaneously extract features and classify using character-level embeddings and convolutional neural network. In addition to completely automating the feature design and extraction process, eXpose outperforms manual feature extraction based baselines on all of the intrusion detection problems we tested it on, yielding a 5%-10% detection rate gain at 0.1% false positive rate compared to these baselines.", "classified_sentences": [ { "sentence": "For years security machine learning research has promised to obviate the need for signature based detection by automatically learning to detect indicators of attack.", "category": "background" }, { "sentence": "Unfortunately, this vision hasn't come to fruition: in fact, developing and maintaining today's security machine learning systems can require engineering resources that are comparable to that of signature-based detection systems, due in part to the need to develop and continuously tune the \"features\" these machine learning systems look at as attacks evolve.", "category": "background" }, { "sentence": "Deep learning, a subfield of machine learning, promises to change this by operating on raw input signals and automating the process of feature design and extraction.", "category": "background" }, { "sentence": "In this paper we propose the eXpose neural network, which uses a deep learning approach we have developed to take generic, raw short character strings as input (a common case for security inputs, which include artifacts like potentially malicious URLs, file paths, named pipes, named mutexes, and registry keys), and learns to simultaneously extract features and classify using character-level embeddings and convolutional neural network.", "category": "method" }, { "sentence": "In addition to completely automating the feature design and extraction process, eXpose outperforms manual feature extraction based baselines on all of the intrusion detection problems we tested it on, yielding a 5%-10% detection rate gain at 0.1% false positive rate compared to these baselines.", "category": "result" } ] }, { "paper_id": "10913456", "title": "Event Detection and Domain Adaptation with Convolutional Neural Networks", "abstract": "We study the event detection problem using convolutional neural networks (CNNs) that overcome the two fundamental limitations of the traditional feature-based approaches to this task: complicated feature engineering for rich feature sets and error propagation from the preceding stages which generate these features. The experimental results show that the CNNs outperform the best reported feature-based systems in the general setting as well as the domain adaptation setting without resorting to extensive external resources.", "classified_sentences": [ { "sentence": "We study the event detection problem using convolutional neural networks (CNNs) that overcome the two fundamental limitations of the traditional feature-based approaches to this task: complicated feature engineering for rich feature sets and error propagation from the preceding stages which generate these features.", "category": "method" }, { "sentence": "The experimental results show that the CNNs outperform the best reported feature-based systems in the general setting as well as the domain adaptation setting without resorting to extensive external resources.", "category": "result" } ] }, { "paper_id": "11010048", "title": "Saliency estimation using a non-parametric low-level vision model", "abstract": "Many successful models for predicting attention in a scene involve three main steps: convolution with a set of filters, a center-surround mechanism and spatial pooling to construct a saliency map. However, integrating spatial information and justifying the choice of various parameter values remain open problems. In this paper we show that an efficient model of color appearance in human vision, which contains a principled selection of parameters as well as an innate spatial pooling mechanism, can be generalized to obtain a saliency model that outperforms state-of-the-art models. Scale integration is achieved by an inverse wavelet transform over the set of scale-weighted center-surround responses. The scale-weighting function (termed ECSF) has been optimized to better replicate psychophysical data on color appearance, and the appropriate sizes of the center-surround inhibition windows have been determined by training a Gaussian Mixture Model on eye-fixation data, thus avoiding ad-hoc parameter selection. Additionally, we conclude that the extension of a color appearance model to saliency estimation adds to the evidence for a common low-level visual front-end for different visual tasks.", "classified_sentences": [ { "sentence": "Many successful models for predicting attention in a scene involve three main steps: convolution with a set of filters, a center-surround mechanism and spatial pooling to construct a saliency map.", "category": "background" }, { "sentence": "However, integrating spatial information and justifying the choice of various parameter values remain open problems.", "category": "background" }, { "sentence": "In this paper we show that an efficient model of color appearance in human vision, which contains a principled selection of parameters as well as an innate spatial pooling mechanism, can be generalized to obtain a saliency model that outperforms state-of-the-art models.", "category": "method" }, { "sentence": "Scale integration is achieved by an inverse wavelet transform over the set of scale-weighted center-surround responses.", "category": "method" }, { "sentence": "The scale-weighting function (termed ECSF) has been optimized to better replicate psychophysical data on color appearance, and the appropriate sizes of the center-surround inhibition windows have been determined by training a Gaussian Mixture Model on eye-fixation data, thus avoiding ad-hoc parameter selection.", "category": "method" }, { "sentence": "Additionally, we conclude that the extension of a color appearance model to saliency estimation adds to the evidence for a common low-level visual front-end for different visual tasks.", "category": "result" } ] }, { "paper_id": "11106781", "title": "FAemb: A function approximation-based embedding method for image retrieval", "abstract": "The objective of this paper is to design an embedding method mapping local features describing image (e.g. SIFT) to a higher dimensional representation used for image retrieval problem. By investigating the relationship between the linear approximation of a nonlinear function in high dimensional space and state-of-the-art feature representation used in image retrieval, i.e., VLAD, we first introduce a new approach for the approximation. The embedded vectors resulted by the function approximation process are then aggregated to form a single representation used in the image retrieval framework. The evaluation shows that our embedding method gives a performance boost over the state of the art in image retrieval, as demonstrated by our experiments on the standard public image retrieval benchmarks.", "classified_sentences": [ { "sentence": "The objective of this paper is to design an embedding method mapping local features describing image (e.g. SIFT) to a higher dimensional representation used for image retrieval problem.", "category": "method" }, { "sentence": "By investigating the relationship between the linear approximation of a nonlinear function in high dimensional space and state-of-the-art feature representation used in image retrieval, i.e., VLAD, we first introduce a new approach for the approximation.", "category": "method" }, { "sentence": "The embedded vectors resulted by the function approximation process are then aggregated to form a single representation used in the image retrieval framework.", "category": "method" }, { "sentence": "The evaluation shows that our embedding method gives a performance boost over the state of the art in image retrieval, as demonstrated by our experiments on the standard public image retrieval benchmarks.", "category": "result" } ] }, { "paper_id": "11403097", "title": "Adaptive Multi-Compositionality for Recursive Neural Models with Applications to Sentiment Analysis", "abstract": "Recursive neural models have achieved promising results in many natural language processing tasks. The main difference among these models lies in the composition function, i.e., how to obtain the vector representation for a phrase or sentence using the representations of words it contains. This paper introduces a novel Adaptive Multi-Compositionality (AdaMC) layer to recursive neural models. The basic idea is to use more than one composition functions and adaptively select them depending on the input vectors. We present a general framework to model each semantic composition as a distribution over these composition functions. The composition functions and parameters used for adaptive selection are learned jointly from data. We integrate AdaMC into existing recursive neural models and conduct extensive experiments on the Stanford Sentiment Treebank. The results illustrate that AdaMC significantly outperforms state-of-the-art sentiment classification methods. It helps push the best accuracy of sentence-level negative/positive classification from 85.4% up to 88.5%.", "classified_sentences": [ { "sentence": "Recursive neural models have achieved promising results in many natural language processing tasks.", "category": "background" }, { "sentence": "The main difference among these models lies in the composition function, i.e., how to obtain the vector representation for a phrase or sentence using the representations of words it contains.", "category": "background" }, { "sentence": "This paper introduces a novel Adaptive Multi-Compositionality (AdaMC) layer to recursive neural models.", "category": "method" }, { "sentence": "The basic idea is to use more than one composition functions and adaptively select them depending on the input vectors.", "category": "method" }, { "sentence": "We present a general framework to model each semantic composition as a distribution over these composition functions.", "category": "method" }, { "sentence": "The composition functions and parameters used for adaptive selection are learned jointly from data.", "category": "method" }, { "sentence": "We integrate AdaMC into existing recursive neural models and conduct extensive experiments on the Stanford Sentiment Treebank.", "category": "method" }, { "sentence": "The results illustrate that AdaMC significantly outperforms state-of-the-art sentiment classification methods.", "category": "result" }, { "sentence": "It helps push the best accuracy of sentence-level negative/positive classification from 85.4% up to 88.5%.", "category": "result" } ] }, { "paper_id": "11505317", "title": "Indonesian Twitter text authority classification for government in Bandung", "abstract": "Nowadays, social media based complaint management systems have been deployed in several countries and cities including Bandung. We proposed an automatic authority classification for Twitter text in Indonesian as part of the complaint management system. Our analysis showed that there are several Twitter message types raised in official account Twitter of the city government. The classification employed a statistical based multi-label text classification. Here, we compared several techniques in the classification such as the features, the algorithms and the classification schemes. In the features comparison, we examined several features such as the complaint word feature, n-gram feature, and the @username feature. In the algorithms comparison, we employed Decision Tree algorithm, Naïve Bayes algorithm, and Support Vector Machine algorithm with multi-label classification techniques of Binary Relevance and Label Power Set. In the complaint classification schemes, we compared the direct classification and two steps classification. Using 2244 twitter texts from twitter of Bandung city government and 5-fold cross validation, the best experimental result of 70.90% accuracy was achieved by the feature combination of 1-gram and complaint word, with Support Vector Machine and Label Power Set as the algorithm, in the direct scheme of text classification.", "classified_sentences": [ { "sentence": "Nowadays, social media based complaint management systems have been deployed in several countries and cities including Bandung.", "category": "background" }, { "sentence": "We proposed an automatic authority classification for Twitter text in Indonesian as part of the complaint management system.", "category": "method" }, { "sentence": "Our analysis showed that there are several Twitter message types raised in official account Twitter of the city government.", "category": "result" }, { "sentence": "The classification employed a statistical based multi-label text classification.", "category": "method" }, { "sentence": "Here, we compared several techniques in the classification such as the features, the algorithms and the classification schemes.", "category": "method" }, { "sentence": "In the features comparison, we examined several features such as the complaint word feature, n-gram feature, and the @username feature.", "category": "method" }, { "sentence": "In the algorithms comparison, we employed Decision Tree algorithm, Naïve Bayes algorithm, and Support Vector Machine algorithm with multi-label classification techniques of Binary Relevance and Label Power Set.", "category": "method" }, { "sentence": "In the complaint classification schemes, we compared the direct classification and two steps classification.", "category": "method" }, { "sentence": "Using 2244 twitter texts from twitter of Bandung city government and 5-fold cross validation, the best experimental result of 70.90% accuracy was achieved by the feature combination of 1-gram and complaint word, with Support Vector Machine and Label Power Set as the algorithm, in the direct scheme of text classification.", "category": "result" } ] }, { "paper_id": "11599080", "title": "Coarse-to-Fine n-Best Parsing and MaxEnt Discriminative Reranking", "abstract": "Discriminative reranking is one method for constructing high-performance statistical parsers (Collins, 2000). A discriminative reranker requires a source of candidate parses for each sentence. This paper describes a simple yet novel method for constructing sets of 50-best parses based on a coarse-to-fine generative parser (Charniak, 2000). This method generates 50-best lists that are of substantially higher quality than previously obtainable. We used these parses as the input to a MaxEnt reranker (Johnson et al., 1999; Riezler et al., 2002) that selects the best parse from the set of parses for each sentence, obtaining an f-score of 91.0% on sentences of length 100 or less.", "classified_sentences": [ { "sentence": "Discriminative reranking is one method for constructing high-performance statistical parsers (Collins, 2000).", "category": "background" }, { "sentence": "A discriminative reranker requires a source of candidate parses for each sentence.", "category": "background" }, { "sentence": "This paper describes a simple yet novel method for constructing sets of 50-best parses based on a coarse-to-fine generative parser (Charniak, 2000).", "category": "method" }, { "sentence": "This method generates 50-best lists that are of substantially higher quality than previously obtainable.", "category": "result" }, { "sentence": "We used these parses as the input to a MaxEnt reranker (Johnson et al., 1999; Riezler et al., 2002) that selects the best parse from the set of parses for each sentence, obtaining an f-score of 91.0% on sentences of length 100 or less.", "category": "result" } ] }, { "paper_id": "13218809", "title": "Trypanosoma brucei FLA1 Is Required for Flagellum Attachment and Cytokinesis*", "abstract": "The single flagellum of the protozoan parasiteTrypanosoma brucei is attached along the length of the cell body by a complex structure that requires the FLA1 protein. We show here that inhibition of FLA1 expression by RNA interference in procyclic trypanosomes causes flagellar detachment and prevents cytokinesis. Despite being unable to divide, these cells undergo mitosis and develop a multinucleated phenotype. The Trypanosoma cruzi FLA1 homolog, GP72, is unable to complement either the flagellar detachment or cytokinesis defects in procyclic T. brucei that have been depleted of FLA1 by RNA interference. Instead, GP72 itself caused flagellar detachment when expressed in T. brucei. In contrast to T. brucei cells depleted of FLA1, procyclic T. bruceiexpressing GP72 continued to divide despite having detached flagella, demonstrating that flagellar attachment is not absolutely necessary for cytokinesis. We have also identified a FLA1-related gene (FLA2) whose sequence is similar but not identical toFLA1. Inhibition of FLA1 and FLA2expression in bloodstream T. brucei caused flagellar detachment and blocked cytokinesis but did not inhibit mitosis. These experiments demonstrate that the FLA proteins are essential and suggest that in procyclic T. brucei, the FLA1 protein has separable functions in flagellar attachment and cytokinesis.", "classified_sentences": [ { "sentence": "The single flagellum of the protozoan parasite Trypanosoma brucei is attached along the length of the cell body by a complex structure that requires the FLA1 protein.", "category": "background" }, { "sentence": "We show here that inhibition of FLA1 expression by RNA interference in procyclic trypanosomes causes flagellar detachment and prevents cytokinesis.", "category": "result" }, { "sentence": "Despite being unable to divide, these cells undergo mitosis and develop a multinucleated phenotype.", "category": "result" }, { "sentence": "The Trypanosoma cruzi FLA1 homolog, GP72, is unable to complement either the flagellar detachment or cytokinesis defects in procyclic T. brucei that have been depleted of FLA1 by RNA interference.", "category": "result" }, { "sentence": "Instead, GP72 itself caused flagellar detachment when expressed in T. brucei.", "category": "result" }, { "sentence": "In contrast to T. brucei cells depleted of FLA1, procyclic T. bruceiexpressing GP72 continued to divide despite having detached flagella, demonstrating that flagellar attachment is not absolutely necessary for cytokinesis.", "category": "result" }, { "sentence": "We have also identified a FLA1-related gene (FLA2) whose sequence is similar but not identical toFLA1.", "category": "method" }, { "sentence": "Inhibition of FLA1 and FLA2expression in bloodstream T. brucei caused flagellar detachment and blocked cytokinesis but did not inhibit mitosis.", "category": "result" }, { "sentence": "These experiments demonstrate that the FLA proteins are essential and suggest that in procyclic T. brucei, the FLA1 protein has separable functions in flagellar attachment and cytokinesis.", "category": "result" } ] }, { "paper_id": "14305261", "title": "On the importance of pair-wise feature correlations for image classification", "abstract": "We show that simple linear classification of pairwise products of convolutional features achieves near state-of-the-art performance on some standard labelled image databases. Specifically, we found test classification error rates on the MNIST handwritten digits image database of under 0.5%, and achieved under 19% and under 44% error rates on the CIFAR-10 and CIFAR-100 RGB image databases. Since the number of weights in such a classifier grows with the square of the number of features, we discuss how implementation of such a pair-wise products classifier can be achieved in an SLFN architecture where the hidden unit function is the simple quadratic nonlinearity: we can this a Quadratic Neural Network (QNN). We compare this method to setting the input weights in a QNN randomly, and find optimal performance can be achieved provided the hidden layer is sufficiently large. This analysis provides insight on why `extreme-learning machines' can achieve classification performance equal to or better than the use of backpropagation training.", "classified_sentences": [ { "sentence": "We show that simple linear classification of pairwise products of convolutional features achieves near state-of-the-art performance on some standard labelled image databases.", "category": "result" }, { "sentence": "Specifically, we found test classification error rates on the MNIST handwritten digits image database of under 0.5%, and achieved under 19% and under 44% error rates on the CIFAR-10 and CIFAR-100 RGB image databases.", "category": "result" }, { "sentence": "Since the number of weights in such a classifier grows with the square of the number of features, we discuss how implementation of such a pair-wise products classifier can be achieved in an SLFN architecture where the hidden unit function is the simple quadratic nonlinearity: we can this a Quadratic Neural Network (QNN).", "category": "method" }, { "sentence": "We compare this method to setting the input weights in a QNN randomly, and find optimal performance can be achieved provided the hidden layer is sufficiently large.", "category": "method" }, { "sentence": "This analysis provides insight on why `extreme-learning machines' can achieve classification performance equal to or better than the use of backpropagation training.", "category": "background" } ] }, { "paper_id": "14327909", "title": "Dynamic visual attention: searching for coding length increments", "abstract": "A visual attention system should respond placidly when common stimuli are presented, while at the same time keep alert to anomalous visual inputs. In this paper, a dynamic visual attention model based on the rarity of features is proposed. We introduce the Incremental Coding Length (ICL) to measure the perspective entropy gain of each feature. The objective of our model is to maximize the entropy of the sampled visual features. In order to optimize energy consumption, the limit amount of energy of the system is re-distributed amongst features according to their Incremental Coding Length. By selecting features with large coding length increments, the computational system can achieve attention selectivity in both static and dynamic scenes. We demonstrate that the proposed model achieves superior accuracy in comparison to mainstream approaches in static saliency map generation. Moreover, we also show that our model captures several less-reported dynamic visual search behaviors, such as attentional swing and inhibition of return.", "classified_sentences": [ { "sentence": "A visual attention system should respond placidly when common stimuli are presented, while at the same time keep alert to anomalous visual inputs.", "category": "background" }, { "sentence": "In this paper, a dynamic visual attention model based on the rarity of features is proposed.", "category": "method" }, { "sentence": "We introduce the Incremental Coding Length (ICL) to measure the perspective entropy gain of each feature.", "category": "method" }, { "sentence": "The objective of our model is to maximize the entropy of the sampled visual features.", "category": "method" }, { "sentence": "In order to optimize energy consumption, the limit amount of energy of the system is re-distributed amongst features according to their Incremental Coding Length.", "category": "method" }, { "sentence": "By selecting features with large coding length increments, the computational system can achieve attention selectivity in both static and dynamic scenes.", "category": "method" }, { "sentence": "We demonstrate that the proposed model achieves superior accuracy in comparison to mainstream approaches in static saliency map generation.", "category": "result" }, { "sentence": "Moreover, we also show that our model captures several less-reported dynamic visual search behaviors, such as attentional swing and inhibition of return.", "category": "result" } ] }, { "paper_id": "14615983", "title": "Detecting Malicious Websites by Learning IP Address Features", "abstract": "Web-based malware attacks have become one of the most serious threats that need to be addressed urgently. Several approaches that have attracted attention as promising ways of detecting such malware include employing various blacklists. However, these conventional approaches often fail to detect new attacks owing to the versatility of malicious websites. Thus, it is difficult to maintain up-to-date blacklists with information regarding new malicious websites. To tackle this problem, we propose a new method for detecting malicious websites using the characteristics of IP addresses. Our approach leverages the empirical observation that IP addresses are more stable than other metrics such as URL and DNS. While the strings that form URLs or domain names are highly variable, IP addresses are less variable, i.e., IPv4 address space is mapped onto 4-bytes strings. We develop a lightweight and scalable detection scheme based on the machine learning technique. The aim of this study is not to provide a single solution that effectively detects web-based malware but to develop a technique that compensates the drawbacks of existing approaches. We validate the effectiveness of our approach by using real IP address data from existing blacklists and real traffic data on a campus network. The results demonstrate that our method can expand the coverage/accuracy of existing blacklists and also detect unknown malicious websites that are not covered by conventional approaches.", "classified_sentences": [ { "sentence": "Web-based malware attacks have become one of the most serious threats that need to be addressed urgently.", "category": "background" }, { "sentence": "Several approaches that have attracted attention as promising ways of detecting such malware include employing various blacklists.", "category": "background" }, { "sentence": "However, these conventional approaches often fail to detect new attacks owing to the versatility of malicious websites.", "category": "background" }, { "sentence": "Thus, it is difficult to maintain up-to-date blacklists with information regarding new malicious websites.", "category": "background" }, { "sentence": "To tackle this problem, we propose a new method for detecting malicious websites using the characteristics of IP addresses.", "category": "method" }, { "sentence": "Our approach leverages the empirical observation that IP addresses are more stable than other metrics such as URL and DNS.", "category": "method" }, { "sentence": "While the strings that form URLs or domain names are highly variable, IP addresses are less variable, i.e., IPv4 address space is mapped onto 4-bytes strings.", "category": "method" }, { "sentence": "We develop a lightweight and scalable detection scheme based on the machine learning technique.", "category": "method" }, { "sentence": "The aim of this study is not to provide a single solution that effectively detects web-based malware but to develop a technique that compensates the drawbacks of existing approaches.", "category": "method" }, { "sentence": "We validate the effectiveness of our approach by using real IP address data from existing blacklists and real traffic data on a campus network.", "category": "method" }, { "sentence": "The results demonstrate that our method can expand the coverage/accuracy of existing blacklists and also detect unknown malicious websites that are not covered by conventional approaches.", "category": "result" } ] }, { "paper_id": "14747729", "title": "(Re)ranking Meets Morphosyntax: State-of-the-art Results from the SPMRL 2013 Shared Task", "abstract": "This paper describes the IMS-SZEGED-CIS contribution to the SPMRL 2013 Shared Task. We participate in both the constituency and dependency tracks, and achieve state-of-theart for all languages. For both tracks we make significant improvements through high quality preprocessing and (re)ranking on top of strong baselines. Our system came out first for both tracks.", "classified_sentences": [ { "sentence": "This paper describes the IMS-SZEGED-CIS contribution to the SPMRL 2013 Shared Task.", "category": "background" }, { "sentence": "We participate in both the constituency and dependency tracks, and achieve state-of-theart for all languages.", "category": "method" }, { "sentence": "For both tracks we make significant improvements through high quality preprocessing and (re)ranking on top of strong baselines.", "category": "method" }, { "sentence": "Our system came out first for both tracks.", "category": "result" } ] }, { "paper_id": "14833979", "title": "Learning to Detect A Salient Object", "abstract": "We study visual attention by detecting a salient object in an input image. We formulate salient object detection as an image segmentation problem, where we separate the salient object from the image background. We propose a set of novel features including multi-scale contrast, center-surround histogram, and color spatial distribution to describe a salient object locally, regionally, and globally. A conditional random field is learned to effectively combine these features for salient object detection. We also constructed a large image database containing tens of thousands of carefully labeled images by multiple users. To our knowledge, it is the first large image database for quantitative evaluation of visual attention algorithms. We validate our approach on this image database, which is public available with this paper.", "classified_sentences": [ { "sentence": "We study visual attention by detecting a salient object in an input image.", "category": "background" }, { "sentence": "We formulate salient object detection as an image segmentation problem, where we separate the salient object from the image background.", "category": "method" }, { "sentence": "We propose a set of novel features including multi-scale contrast, center-surround histogram, and color spatial distribution to describe a salient object locally, regionally, and globally.", "category": "method" }, { "sentence": "A conditional random field is learned to effectively combine these features for salient object detection.", "category": "method" }, { "sentence": "We also constructed a large image database containing tens of thousands of carefully labeled images by multiple users.", "category": "method" }, { "sentence": "To our knowledge, it is the first large image database for quantitative evaluation of visual attention algorithms.", "category": "background" }, { "sentence": "We validate our approach on this image database, which is public available with this paper.", "category": "result" } ] }, { "paper_id": "15218192", "title": "A simple but potentially powerful approach for multilingual parsing", "abstract": "All approaches today for multilingual dependency parsing don’t use any support of sister languages. This paper present a very different approach to deal with data sparsity, language transfer, etc. Our approach really use sister languages theory, combining resources from that type of languages, we want a model who can really parse many different languages. In this paper, we present a new architecture who not use any type of oracles to make good parsing decisions. We hope that type of approach can really make dynamical parsing decisions.", "classified_sentences": [ { "sentence": "All approaches today for multilingual dependency parsing don’t use any support of sister languages.", "category": "background" }, { "sentence": "This paper present a very different approach to deal with data sparsity, language transfer, etc. Our approach really use sister languages theory, combining resources from that type of languages, we want a model who can really parse many different languages.", "category": "method" }, { "sentence": "In this paper, we present a new architecture who not use any type of oracles to make good parsing decisions.", "category": "method" }, { "sentence": "We hope that type of approach can really make dynamical parsing decisions.", "category": "result" } ] }, { "paper_id": "15261417", "title": "PhishNet: Predictive Blacklisting to Detect Phishing Attacks", "abstract": "Phishing has been easy and effective way for trickery and deception on the Internet. While solutions such as URL blacklisting have been effective to some degree, their reliance on exact match with the blacklisted entries makes it easy for attackers to evade. We start with the observation that attackers often employ simple modifications (e.g., changing top level domain) to URLs. Our system, PhishNet, exploits this observation using two components. In the first component, we propose five heuristics to enumerate simple combinations of known phishing sites to discover new phishing URLs. The second component consists of an approximate matching algorithm that dissects a URL into multiple components that are matched individually against entries in the blacklist. In our evaluation with real-time blacklist feeds, we discovered around 18,000 new phishing URLs from a set of 6,000 new blacklist entries. We also show that our approximate matching algorithm leads to very few false positives (3%) and negatives (5%).", "classified_sentences": [ { "sentence": "Phishing has been easy and effective way for trickery and deception on the Internet.", "category": "background" }, { "sentence": "While solutions such as URL blacklisting have been effective to some degree, their reliance on exact match with the blacklisted entries makes it easy for attackers to evade.", "category": "background" }, { "sentence": "We start with the observation that attackers often employ simple modifications (e.g., changing top level domain) to URLs.", "category": "background" }, { "sentence": "Our system, PhishNet, exploits this observation using two components.", "category": "method" }, { "sentence": "In the first component, we propose five heuristics to enumerate simple combinations of known phishing sites to discover new phishing URLs.", "category": "method" }, { "sentence": "The second component consists of an approximate matching algorithm that dissects a URL into multiple components that are matched individually against entries in the blacklist.", "category": "method" }, { "sentence": "In our evaluation with real-time blacklist feeds, we discovered around 18,000 new phishing URLs from a set of 6,000 new blacklist entries.", "category": "result" }, { "sentence": "We also show that our approximate matching algorithm leads to very few false positives (3%) and negatives (5%).", "category": "result" } ] }, { "paper_id": "15307333", "title": "Low-Rank Tensors for Scoring Dependency Structures", "abstract": "Accurate scoring of syntactic structures such as head-modifier arcs in dependency parsing typically requires rich, highdimensional feature representations. A small subset of such features is often selected manually. This is problematic when features lack clear linguistic meaning as in embeddings or when the information is blended across features. In this paper, we use tensors to map high-dimensional feature vectors into low dimensional representations. We explicitly maintain the parameters as a low-rank tensor to obtain low dimensional representations of words in their syntactic roles, and to leverage modularity in the tensor for easy training with online algorithms. Our parser consistently outperforms the Turbo and MST parsers across 14 different languages. We also obtain the best published UAS results on 5 languages. 1", "classified_sentences": [ { "sentence": "Accurate scoring of syntactic structures such as head-modifier arcs in dependency parsing typically requires rich, highdimensional feature representations.", "category": "background" }, { "sentence": "A small subset of such features is often selected manually.", "category": "background" }, { "sentence": "This is problematic when features lack clear linguistic meaning as in embeddings or when the information is blended across features.", "category": "background" }, { "sentence": "In this paper, we use tensors to map high-dimensional feature vectors into low dimensional representations.", "category": "method" }, { "sentence": "We explicitly maintain the parameters as a low-rank tensor to obtain low dimensional representations of words in their syntactic roles, and to leverage modularity in the tensor for easy training with online algorithms.", "category": "method" }, { "sentence": "Our parser consistently outperforms the Turbo and MST parsers across 14 different languages.", "category": "result" }, { "sentence": "We also obtain the best published UAS results on 5 languages.", "category": "result" } ] }, { "paper_id": "15407650", "title": "Span-Based Constituency Parsing with a Structure-Label System and Provably Optimal Dynamic Oracles", "abstract": "Parsing accuracy using efficient greedy transition systems has improved dramatically in recent years thanks to neural networks. Despite striking results in dependency parsing, however, neural models have not surpassed state-of-the-art approaches in constituency parsing. To remedy this, we introduce a new shift-reduce system whose stack contains merely sentence spans, represented by a bare minimum of LSTM features. We also design the first provably optimal dynamic oracle for constituency parsing, which runs in amortized O(1) time, compared to O(n^3) oracles for standard dependency parsing. Training with this oracle, we achieve the best F1 scores on both English and French of any parser that does not use reranking or external data.", "classified_sentences": [ { "sentence": "Parsing accuracy using efficient greedy transition systems has improved dramatically in recent years thanks to neural networks.", "category": "background" }, { "sentence": "Despite striking results in dependency parsing, however, neural models have not surpassed state-of-the-art approaches in constituency parsing.", "category": "background" }, { "sentence": "To remedy this, we introduce a new shift-reduce system whose stack contains merely sentence spans, represented by a bare minimum of LSTM features.", "category": "method" }, { "sentence": "We also design the first provably optimal dynamic oracle for constituency parsing, which runs in amortized O(1) time, compared to O(n^3) oracles for standard dependency parsing.", "category": "method" }, { "sentence": "Training with this oracle, we achieve the best F1 scores on both English and French of any parser that does not use reranking or external data.", "category": "result" } ] }, { "paper_id": "15664890", "title": "IRST-BP: Preposition Disambiguation based on Chain Clarifying Relationships Contexts", "abstract": "We are going to present a technique of preposition disambiguation based on sense discriminative patterns, which are acquired using a variant of Angluin's algorithm. They represent the essential information extracted from a particular type of local contexts we call Chain Clarifying Relationship contexts. The data set and the results we present are from the Semeval task, WSD of Preposition (Litkowski 2007).", "classified_sentences": [ { "sentence": "We are going to present a technique of preposition disambiguation based on sense discriminative patterns, which are acquired using a variant of Angluin's algorithm.", "category": "method" }, { "sentence": "They represent the essential information extracted from a particular type of local contexts we call Chain Clarifying Relationship contexts.", "category": "method" }, { "sentence": "The data set and the results we present are from the Semeval task, WSD of Preposition (Litkowski 2007).", "category": "result" } ] }, { "paper_id": "16445820", "title": "Learning to predict where humans look", "abstract": "For many applications in graphics, design, and human computer interaction, it is essential to understand where humans look in a scene. Where eye tracking devices are not a viable option, models of saliency can be used to predict fixation locations. Most saliency approaches are based on bottom-up computation that does not consider top-down image semantics and often does not match actual eye movements. To address this problem, we collected eye tracking data of 15 viewers on 1003 images and use this database as training and testing examples to learn a model of saliency based on low, middle and high-level image features. This large database of eye tracking data is publicly available with this paper.", "classified_sentences": [ { "sentence": "For many applications in graphics, design, and human computer interaction, it is essential to understand where humans look in a scene.", "category": "background" }, { "sentence": "Where eye tracking devices are not a viable option, models of saliency can be used to predict fixation locations.", "category": "background" }, { "sentence": "Most saliency approaches are based on bottom-up computation that does not consider top-down image semantics and often does not match actual eye movements.", "category": "background" }, { "sentence": "To address this problem, we collected eye tracking data of 15 viewers on 1003 images and use this database as training and testing examples to learn a model of saliency based on low, middle and high-level image features.", "category": "method" }, { "sentence": "This large database of eye tracking data is publicly available with this paper.", "category": "result" } ] }, { "paper_id": "18198301", "title": "Learning Unified Features from Natural and Programming Languages for Locating Buggy Source Code", "abstract": "Bug reports provide an effective way for end-users to disclose potential bugs hidden in a software system, while automatically locating the potential buggy source code according to a bug report remains a great challenge in software maintenance. Many previous studies treated the source code as natural language by representing both the bug report and source code based on bag-of-words feature representations, and correlate the bug report and source code by measuring similarity in the same lexical feature space. However, these approaches fail to consider the structure information of source code which carries additional semantics beyond the lexical terms. Such information is important in modeling program functionality. In this paper, we propose a novel convolutional neural network NP-CNN, which leverages both lexical and program structure information to learn unified features from natural language and source code in programming language for automatically locating the potential buggy source code according to bug report. Experimental results on widely-used software projects indicate that NP-CNN significantly outperforms the state-of-the-art methods in locating the buggy source files.", "classified_sentences": [ { "sentence": "Bug reports provide an effective way for end-users to disclose potential bugs hidden in a software system, while automatically locating the potential buggy source code according to a bug report remains a great challenge in software maintenance.", "category": "background" }, { "sentence": "Many previous studies treated the source code as natural language by representing both the bug report and source code based on bag-of-words feature representations, and correlate the bug report and source code by measuring similarity in the same lexical feature space.", "category": "background" }, { "sentence": "However, these approaches fail to consider the structure information of source code which carries additional semantics beyond the lexical terms.", "category": "background" }, { "sentence": "Such information is important in modeling program functionality.", "category": "background" }, { "sentence": "In this paper, we propose a novel convolutional neural network NP-CNN, which leverages both lexical and program structure information to learn unified features from natural language and source code in programming language for automatically locating the potential buggy source code according to bug report.", "category": "method" }, { "sentence": "Experimental results on widely-used software projects indicate that NP-CNN significantly outperforms the state-of-the-art methods in locating the buggy source files.", "category": "result" } ] }, { "paper_id": "152768189", "title": "A Documentary History of Modern Iraq", "abstract": "Previously published histories and primary source collections on the Iraqi experience tend to be topically focused or dedicated to presenting a top-down approach. By contrast, Stacy Holden's A Documentary History of Modern Iraq gives voice to ordinary Iraqis, clarifying the experience of the Shiites, Sunnis, Kurds, Jews, and women over the past century. Through varied documents ranging from short stories to treaties, political speeches to memoirs, and newspaper articles to book excerpts, the work synthesises previously marginalised perspectives of minorities and women with the voices of the political elite to provide an integrated picture of political change from the Ottoman Empire in 1903 to the end of the second Bush administration in 2008. Covering a broad range of topics, this bottom-up approach allows readers to fully immerse themselves in the lives of everyday Iraqis as they navigate regime shifts from the British to the Hashemite monarchy, the political upheaval of the Persian Gulf wars, and beyond. Brief introductions to each excerpt provide context and suggest questions for classroom discussion. This collection offers raw history, untainted and unfiltered by modern political framework and thought, representing a refreshing new approach to the study of Iraq.", "classified_sentences": [ { "sentence": "Previously published histories and primary source collections on the Iraqi experience tend to be topically focused or dedicated to presenting a top-down approach.", "category": "background" }, { "sentence": "By contrast, Stacy Holden's A Documentary History of Modern Iraq gives voice to ordinary Iraqis, clarifying the experience of the Shiites, Sunnis, Kurds, Jews, and women over the past century.", "category": "method" }, { "sentence": "Through varied documents ranging from short stories to treaties, political speeches to memoirs, and newspaper articles to book excerpts, the work synthesises previously marginalised perspectives of minorities and women with the voices of the political elite to provide an integrated picture of political change from the Ottoman Empire in 1903 to the end of the second Bush administration in 2008.", "category": "method" }, { "sentence": "Covering a broad range of topics, this bottom-up approach allows readers to fully immerse themselves in the lives of everyday Iraqis as they navigate regime shifts from the British to the Hashemite monarchy, the political upheaval of the Persian Gulf wars, and beyond.", "category": "method" }, { "sentence": "Brief introductions to each excerpt provide context and suggest questions for classroom discussion.", "category": "method" }, { "sentence": "This collection offers raw history, untainted and unfiltered by modern political framework and thought, representing a refreshing new approach to the study of Iraq.", "category": "background" } ] }, { "paper_id": "20173528", "title": "Control-oriented modelling and adaptive control of a single-phase quasi-Z-source inverter", "abstract": "One key feature of Z-source and quasi-Z-source converters is the inclusion of a fully reactive network (the Z network) between input voltage source and inverter bridge. The inclusion of the Z network enables the possibility of utilizing the so-called shoot-through mode, i.e. the simultaneous activation of both switches in a single inverter bridge leg. Typical control strategies for these converters involve two separate objectives: (a) the control of Z-network variables (capacitor voltages) aimed at regulating the Z-network-to-inverter-bridge voltage and (b) standard inverter bridge output control. Modelling and control of the Z-network behavior, aiming at objective (a), is complicated by the fact that the current drawn from the Z-network by the inverter bridge during non shoot-through may be discontinuous and cannot be considered constant, especially in the case of a single-phase inverter. Our main contribution is to derive an adequately simple and sufficiently useful control-oriented model for the Z-network variables, taking full account of the variability of the aforementioned current and hence avoiding restrictive assumptions on the load or the PWM modulation strategy. A second contribution is to derive, based on the Z-network model, an adaptive control law able to compensate for large-signal variations. Successful operation is illustrated by means of a simulation example.", "classified_sentences": [ { "sentence": "One key feature of Z-source and quasi-Z-source converters is the inclusion of a fully reactive network (the Z network) between input voltage source and inverter bridge.", "category": "background" }, { "sentence": "The inclusion of the Z network enables the possibility of utilizing the so-called shoot-through mode, i.e. the simultaneous activation of both switches in a single inverter bridge leg.", "category": "background" }, { "sentence": "Typical control strategies for these converters involve two separate objectives: (a) the control of Z-network variables (capacitor voltages) aimed at regulating the Z-network-to-inverter-bridge voltage and (b) standard inverter bridge output control.", "category": "method" }, { "sentence": "Modelling and control of the Z-network behavior, aiming at objective (a), is complicated by the fact that the current drawn from the Z-network by the inverter bridge during non shoot-through may be discontinuous and cannot be considered constant, especially in the case of a single-phase inverter.", "category": "background" }, { "sentence": "Our main contribution is to derive an adequately simple and sufficiently useful control-oriented model for the Z-network variables, taking full account of the variability of the aforementioned current and hence avoiding restrictive assumptions on the load or the PWM modulation strategy.", "category": "method" }, { "sentence": "A second contribution is to derive, based on the Z-network model, an adaptive control law able to compensate for large-signal variations.", "category": "method" }, { "sentence": "Successful operation is illustrated by means of a simulation example.", "category": "result" } ] }, { "paper_id": "154431099", "title": "Faculty Salary Inequality in U.S. Business Schools: A Mixed Methods Analysis", "abstract": "Through a mixed methods approach, this study provides a greater understanding of salary inequality in U.S. business schools and how it changed between 1998 to 2004. The quantitative research examines full-time faculty using individual-level salary data from both a constant sample of 307 institutions and a larger 2004 sample of 464 schools, allowing for in-depth examination of inequality including within institutions. The qualitative research used interviews with business school deans to uncover decisions that, in the aggregate, can impact faculty salary inequality. Quantitative analysis of faculty salary utilized descriptive statistics as well as several inequality measures, along with regression analyses, to reveal the level and structure of inequality and the contributions of within-institution and between-institution inequality. Salary inequality increased between 1998 and 2004. However, contrary to previous research, salary inequality isn’t attributed to superstar salaries; the growth in salary inequality is attributable to negative real growth in the lower tail of the salary distribution. Analysis between institutions reveals that the highest paying 10% of institutions are pulling away, increasing stratification between the most prestigious institutions and the others. Although private school faculty earn more than their public counterparts, salary inequality among faculty at public institutions increased more rapidly. Institutional characteristics including Carnegie classification, MBA ranking, degrees offered, accreditation, faculty size, tuition and fees, state appropriations per student and endowment per student contribute to differences in salary inequality between", "classified_sentences": [ { "sentence": "Through a mixed methods approach, this study provides a greater understanding of salary inequality in U.S. business schools and how it changed between 1998 to 2004.", "category": "background" }, { "sentence": "The quantitative research examines full-time faculty using individual-level salary data from both a constant sample of 307 institutions and a larger 2004 sample of 464 schools, allowing for in-depth examination of inequality including within institutions.", "category": "method" }, { "sentence": "The qualitative research used interviews with business school deans to uncover decisions that, in the aggregate, can impact faculty salary inequality.", "category": "method" }, { "sentence": "Quantitative analysis of faculty salary utilized descriptive statistics as well as several inequality measures, along with regression analyses, to reveal the level and structure of inequality and the contributions of within-institution and between-institution inequality.", "category": "method" }, { "sentence": "Salary inequality increased between 1998 and 2004.", "category": "result" }, { "sentence": "However, contrary to previous research, salary inequality isn’t attributed to superstar salaries; the growth in salary inequality is attributable to negative real growth in the lower tail of the salary distribution.", "category": "result" }, { "sentence": "Analysis between institutions reveals that the highest paying 10% of institutions are pulling away, increasing stratification between the most prestigious institutions and the others.", "category": "result" }, { "sentence": "Although private school faculty earn more than their public counterparts, salary inequality among faculty at public institutions increased more rapidly.", "category": "result" }, { "sentence": "Institutional characteristics including Carnegie classification, MBA ranking, degrees offered, accreditation, faculty size, tuition and fees, state appropriations per student and endowment per student contribute to differences in salary inequality between", "category": "result" } ] }, { "paper_id": "20254176", "title": "Dietary Supplementation with Virgin Coconut Oil Improves Lipid Profile and Hepatic Antioxidant Status and Has Potential Benefits on Cardiovascular Risk Indices in Normal Rats", "abstract": "ABSTRACT Research findings that suggest beneficial health effects of dietary supplementation with virgin coconut oil (VCO) are limited in the published literature. This study investigated the in vivo effects of a 5-week VCO-supplemented diet on lipid profile, hepatic antioxidant status, hepatorenal function, and cardiovascular risk indices in normal rats. Rats were randomly divided into 3 groups: 1 control and 2 treatment groups (10% and 15% VCO-supplemented diets) for 5 weeks. Serum and homogenate samples were used to analyze lipid profile, hepatorenal function markers, hepatic activities of antioxidant enzymes, and malondialdehyde level. Lipid profile of animals fed VCO diets showed significant reduction in total cholesterol (TC), triglyceride (TG), and low-density lipoprotein (LDL) levels; high-density lipoprotein (HDL) level increased significantly (p < .05) compared to control; and there were beneficial effects on cardiovascular risk indices. The level of malondialdehyde (MDA), a lipid peroxidation marker, remarkably reduced and activities of hepatic antioxidant enzymes—superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GPx)—were markedly increased in VCO diet–fed rats. The VCO diet significantly modulated creatinine, sodium (Na+), potassium (K+), chloride (Cl−), alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) compared to control. The findings suggest a beneficial effect of VCO on lipid profile, renal status, hepatic antioxidant defense system, and cardiovascular risk indices in rats.", "classified_sentences": [ { "sentence": "Research findings that suggest beneficial health effects of dietary supplementation with virgin coconut oil (VCO) are limited in the published literature.", "category": "background" }, { "sentence": "This study investigated the in vivo effects of a 5-week VCO-supplemented diet on lipid profile, hepatic antioxidant status, hepatorenal function, and cardiovascular risk indices in normal rats.", "category": "method" }, { "sentence": "Rats were randomly divided into 3 groups: 1 control and 2 treatment groups (10% and 15% VCO-supplemented diets) for 5 weeks.", "category": "method" }, { "sentence": "Serum and homogenate samples were used to analyze lipid profile, hepatorenal function markers, hepatic activities of antioxidant enzymes, and malondialdehyde level.", "category": "method" }, { "sentence": "Lipid profile of animals fed VCO diets showed significant reduction in total cholesterol (TC), triglyceride (TG), and low-density lipoprotein (LDL) levels; high-density lipoprotein (HDL) level increased significantly (p < .05) compared to control; and there were beneficial effects on cardiovascular risk indices.", "category": "result" }, { "sentence": "The level of malondialdehyde (MDA), a lipid peroxidation marker, remarkably reduced and activities of hepatic antioxidant enzymes—superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GPx)—were markedly increased in VCO diet–fed rats.", "category": "result" }, { "sentence": "The VCO diet significantly modulated creatinine, sodium (Na+), potassium (K+), chloride (Cl−), alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) compared to control.", "category": "result" }, { "sentence": "The findings suggest a beneficial effect of VCO on lipid profile, renal status, hepatic antioxidant defense system, and cardiovascular risk indices in rats.", "category": "result" } ] }, { "paper_id": "155340938", "title": "Diversity and inheritance in cowpea (Vigna unguiculata) on protein and yield components characters", "abstract": "Abstract. Purnamasari I, Sobir, Syukur M. 2019. Diversity and inheritance in cowpea (Vigna unguiculata) on protein and yield components characters. Biodiversitas 20: 1294-1298. Diallel was a method that can be used to see inheritance and diversity character of plants. The objective of this study has evaluated the inheritance and diversity of the yield component characters and seed protein content. This study used genetic material from five homozygous cowpeas (Vigna unguiculata (L.) Walp) genotypes (KM1, KM2, KM4, KM5, and TG2) and 20 F1. Genotypes were planted using a randomized complete block design with three replications. The character observed were the yield components (weight of 100 seed, number of seeds per pod, number of pods per bunch, number of pods per plant and yield) and seeds protein content. Characters that were influenced by general combining ability (GCA) consist of weight of 100 seeds, number of pods per bunch, number of seeds per pod and seed protein content. The variance of GCA has a greater value than the variance of specific combining ability (SCA). Narrow sense and broad sense heritability values were high category consist of weight 100 seed, number seeds per pod, number pods per bunch and seed protein content. That characters were inherited additively. Characters were influenced by additive gene action and have high diversity make it possible to select in F2 generation using the pedigree selection method.", "classified_sentences": [ { "sentence": "Abstract.", "category": "background" }, { "sentence": "Purnamasari I, Sobir, Syukur M.", "category": "background" }, { "sentence": "2019.", "category": "background" }, { "sentence": "Diversity and inheritance in cowpea (Vigna unguiculata) on protein and yield components characters.", "category": "background" }, { "sentence": "Biodiversitas 20: 1294-1298.", "category": "background" }, { "sentence": "Diallel was a method that can be used to see inheritance and diversity character of plants.", "category": "method" }, { "sentence": "The objective of this study has evaluated the inheritance and diversity of the yield component characters and seed protein content.", "category": "method" }, { "sentence": "This study used genetic material from five homozygous cowpeas (Vigna unguiculata (L.) Walp) genotypes (KM1, KM2, KM4, KM5, and TG2) and 20 F1.", "category": "method" }, { "sentence": "Genotypes were planted using a randomized complete block design with three replications.", "category": "method" }, { "sentence": "The character observed were the yield components (weight of 100 seed, number of seeds per pod, number of pods per bunch, number of pods per plant and yield) and seeds protein content.", "category": "method" }, { "sentence": "Characters that were influenced by general combining ability (GCA) consist of weight of 100 seeds, number of pods per bunch, number of seeds per pod and seed protein content.", "category": "result" }, { "sentence": "The variance of GCA has a greater value than the variance of specific combining ability (SCA).", "category": "result" }, { "sentence": "Narrow sense and broad sense heritability values were high category consist of weight 100 seed, number seeds per pod, number pods per bunch and seed protein content.", "category": "result" }, { "sentence": "That characters were inherited additively.", "category": "result" }, { "sentence": "Characters were influenced by additive gene action and have high diversity make it possible to select in F2 generation using the pedigree selection method.", "category": "result" } ] }, { "paper_id": "22886304", "title": "The cell biology of lignification in higher plants.", "abstract": "BACKGROUND Lignin is a polyphenolic polymer that strengthens and waterproofs the cell wall of specialized plant cell types. Lignification is part of the normal differentiation programme and functioning of specific cell types, but can also be triggered as a response to various biotic and abiotic stresses in cells that would not otherwise be lignifying. SCOPE Cell wall lignification exhibits specific characteristics depending on the cell type being considered. These characteristics include the timing of lignification during cell differentiation, the palette of associated enzymes and substrates, the sub-cellular deposition sites, the monomeric composition and the cellular autonomy for lignin monomer production. This review provides an overview of the current understanding of lignin biosynthesis and polymerization at the cell biology level. CONCLUSIONS The lignification process ranges from full autonomy to complete co-operation depending on the cell type. The different roles of lignin for the function of each specific plant cell type are clearly illustrated by the multiple phenotypic defects exhibited by knock-out mutants in lignin synthesis, which may explain why no general mechanism for lignification has yet been defined. The range of phenotypic effects observed include altered xylem sap transport, loss of mechanical support, reduced seed protection and dispersion, and/or increased pest and disease susceptibility.", "classified_sentences": [ { "sentence": "BACKGROUND Lignin is a polyphenolic polymer that strengthens and waterproofs the cell wall of specialized plant cell types.", "category": "background" }, { "sentence": "Lignification is part of the normal differentiation programme and functioning of specific cell types, but can also be triggered as a response to various biotic and abiotic stresses in cells that would not otherwise be lignifying.", "category": "background" }, { "sentence": "SCOPE Cell wall lignification exhibits specific characteristics depending on the cell type being considered.", "category": "background" }, { "sentence": "These characteristics include the timing of lignification during cell differentiation, the palette of associated enzymes and substrates, the sub-cellular deposition sites, the monomeric composition and the cellular autonomy for lignin monomer production.", "category": "background" }, { "sentence": "This review provides an overview of the current understanding of lignin biosynthesis and polymerization at the cell biology level.", "category": "method" }, { "sentence": "CONCLUSIONS The lignification process ranges from full autonomy to complete co-operation depending on the cell type.", "category": "result" }, { "sentence": "The different roles of lignin for the function of each specific plant cell type are clearly illustrated by the multiple phenotypic defects exhibited by knock-out mutants in lignin synthesis, which may explain why no general mechanism for lignification has yet been defined.", "category": "result" }, { "sentence": "The range of phenotypic effects observed include altered xylem sap transport, loss of mechanical support, reduced seed protection and dispersion, and/or increased pest and disease susceptibility.", "category": "result" } ] }, { "paper_id": "23100864", "title": "Morphologically rich Urdu grammar parsing using Earley algorithm", "abstract": "Abstract This work presents the development and evaluation of an extended Urdu parser. It further focuses on issues related to this parser and describes the changes made in the Earley algorithm to get accurate and relevant results from the Urdu parser. The parser makes use of a morphologically rich context free grammar extracted from a linguistically-rich Urdu treebank. This grammar with sufficient encoded information is comparable with the state-of-the-art parsing requirements for the morphologically rich Urdu language. The extended parsing model and the linguistically rich extracted-grammar both provide us better evaluation results in Urdu/Hindi parsing domain. The parser gives 87% of f-score, which outperforms the existing parsing work of Urdu/Hindi based on the tree-banking approach.", "classified_sentences": [ { "sentence": "Abstract This work presents the development and evaluation of an extended Urdu parser.", "category": "background" }, { "sentence": "It further focuses on issues related to this parser and describes the changes made in the Earley algorithm to get accurate and relevant results from the Urdu parser.", "category": "method" }, { "sentence": "The parser makes use of a morphologically rich context free grammar extracted from a linguistically-rich Urdu treebank.", "category": "method" }, { "sentence": "This grammar with sufficient encoded information is comparable with the state-of-the-art parsing requirements for the morphologically rich Urdu language.", "category": "method" }, { "sentence": "The extended parsing model and the linguistically rich extracted-grammar both provide us better evaluation results in Urdu/Hindi parsing domain.", "category": "result" }, { "sentence": "The parser gives 87% of f-score, which outperforms the existing parsing work of Urdu/Hindi based on the tree-banking approach.", "category": "result" } ] }, { "paper_id": "24077639", "title": "High-Osmolar and Low-Osmolar Contrast Media", "abstract": "During the past 3 years a great number of papers about adverse drug reactions to intravascular injection of high-osmolar and low-osmolar iodinated contrast media (CM) have been published. They include observational studies, randomized trials, meta-analyses and committee reports. Thorough analysis of this material substantiates an improvement in safety of at least 6-fold using nonionic low-osmolar CM compared with ionic high-osmolar CM. The point where only a small minority is continuing to argue effectively that low-osmolar CM are not better than conventional high-osmolar CM has now been reached. High-osmolar CM are used less and less for intravascular purposes, and, in fact, have been totally replaced by low-osmolar CM in 4 countries.", "classified_sentences": [ { "sentence": "During the past 3 years a great number of papers about adverse drug reactions to intravascular injection of high-osmolar and low-osmolar iodinated contrast media (CM) have been published.", "category": "background" }, { "sentence": "They include observational studies, randomized trials, meta-analyses and committee reports.", "category": "background" }, { "sentence": "Thorough analysis of this material substantiates an improvement in safety of at least 6-fold using nonionic low-osmolar CM compared with ionic high-osmolar CM.", "category": "result" }, { "sentence": "The point where only a small minority is continuing to argue effectively that low-osmolar CM are not better than conventional high-osmolar CM has now been reached.", "category": "result" }, { "sentence": "High-osmolar CM are used less and less for intravascular purposes, and, in fact, have been totally replaced by low-osmolar CM in 4 countries.", "category": "result" } ] }, { "paper_id": "25021216", "title": "Natural Hazards Preparedness in Taiwan: A Comparison Between Households With and Without Disabled Members.", "abstract": "People with disabilities are one of the most vulnerable groups to natural hazards. Preparedness is critical to protect life and reduce disaster impact. This article discusses the knowledge of the disaster preparedness behaviors of people with disabilities using updated, representative data from Taiwan (2013 Taiwan Social Change Survey), with a comparison to households without disabled members. The adoption of 6 preparedness activities-relocating vehicles or valuable things to a safer place, purchasing insurance, securing furniture, preparing an emergency kit, planning evacuation, and participating in drills-are used separately as dependent variables. The unadjusted results from Logit regression models show that the households with disabled members are less likely to prepare emergency kits and to plan evacuation. But with the adjustment of risk perception (probability, consequence, worrisome) and other factors-experience of earthquake and typhoon hazards, home ownership status, whether there are children in the home, perceived social status, family income, gender, age, education attainment, and religious status-the differences in adopting all 6 preparedness activities between households with disabled members and households without disabled members become nonsignificant. Finally, the contribution, limitations, and practice implications of this article are discussed.", "classified_sentences": [ { "sentence": "People with disabilities are one of the most vulnerable groups to natural hazards.", "category": "background" }, { "sentence": "Preparedness is critical to protect life and reduce disaster impact.", "category": "background" }, { "sentence": "This article discusses the knowledge of the disaster preparedness behaviors of people with disabilities using updated, representative data from Taiwan (2013 Taiwan Social Change Survey), with a comparison to households without disabled members.", "category": "method" }, { "sentence": "The adoption of 6 preparedness activities-relocating vehicles or valuable things to a safer place, purchasing insurance, securing furniture, preparing an emergency kit, planning evacuation, and participating in drills-are used separately as dependent variables.", "category": "method" }, { "sentence": "The unadjusted results from Logit regression models show that the households with disabled members are less likely to prepare emergency kits and to plan evacuation.", "category": "result" }, { "sentence": "But with the adjustment of risk perception (probability, consequence, worrisome) and other factors-experience of earthquake and typhoon hazards, home ownership status, whether there are children in the home, perceived social status, family income, gender, age, education attainment, and religious status-the differences in adopting all 6 preparedness activities between households with disabled members and households without disabled members become nonsignificant.", "category": "result" }, { "sentence": "Finally, the contribution, limitations, and practice implications of this article are discussed.", "category": "result" } ] }, { "paper_id": "25218728", "title": "Synthesis of phenanthrenes and polycyclic heteroarenes by transition-metal catalyzed cycloisomerization reactions.", "abstract": "Readily available biphenyl derivatives containing an alkyne unit at one of their ortho-positions are converted into substituted phenanthrenes on exposure to catalytic amounts of either PtCl2, AuCl, AuCl3, GaCl3 or InCl3 in toluene. This 6-endo-dig cyclization likely proceeds through initial pi-complexation of the alkyne unit followed by interception of the resulting eta2-metal species by the adjacent arene ring. The reaction is inherently modular, allowing for substantial structural variations and for the incorporation of substituents at any site of the phenanthrene product. Moreover, it is readily extended to the heterocyclic series as exemplified by the preparation of benzoindoles, benzocarbazoles, naphthothiophenes, as well as bridgehead nitrogen heterocycles such as pyrrolo[1,2-a]quinolines. Depending on the chosen catalyst, biaryls bearing halo-alkyne units can either be converted into the corresponding 10-halo-phenanthrenes or into the isomeric 9-halo-phenanthrenes; in the latter case, the concomitant 1,2-halide shift is best explained by assuming a metal vinylidene species as the reactive intermediate. The scope of this novel method for the preparation of polycyclic arenes is illustrated by the total synthesis of a series of polyoxygenated phenanthrenes that are close relatives of the anticancer agent combretastatin A-4, as well as by the total synthesis of the aporphine alkaloid O-methyl-dehydroisopiline and its naturally occurring symmetrical dimer.", "classified_sentences": [ { "sentence": "Readily available biphenyl derivatives containing an alkyne unit at one of their ortho-positions are converted into substituted phenanthrenes on exposure to catalytic amounts of either PtCl2, AuCl, AuCl3, GaCl3 or InCl3 in toluene.", "category": "method" }, { "sentence": "This 6-endo-dig cyclization likely proceeds through initial pi-complexation of the alkyne unit followed by interception of the resulting eta2-metal species by the adjacent arene ring.", "category": "method" }, { "sentence": "The reaction is inherently modular, allowing for substantial structural variations and for the incorporation of substituents at any site of the phenanthrene product.", "category": "method" }, { "sentence": "Moreover, it is readily extended to the heterocyclic series as exemplified by the preparation of benzoindoles, benzocarbazoles, naphthothiophenes, as well as bridgehead nitrogen heterocycles such as pyrrolo[1,2-a]quinolines.", "category": "method" }, { "sentence": "Depending on the chosen catalyst, biaryls bearing halo-alkyne units can either be converted into the corresponding 10-halo-phenanthrenes or into the isomeric 9-halo-phenanthrenes; in the latter case, the concomitant 1,2-halide shift is best explained by assuming a metal vinylidene species as the reactive intermediate.", "category": "method" }, { "sentence": "The scope of this novel method for the preparation of polycyclic arenes is illustrated by the total synthesis of a series of polyoxygenated phenanthrenes that are close relatives of the anticancer agent combretastatin A-4, as well as by the total synthesis of the aporphine alkaloid O-methyl-dehydroisopiline and its naturally occurring symmetrical dimer.", "category": "result" } ] }, { "paper_id": "25508712", "title": "High-resolution R-banding at the 1250-band level. 1. Technical considerations on cell synchronization and R-banding (RHG and RBG).", "abstract": "A comparative analysis of variables for cell synchronization was made and led to the description of optimal conditions capable of yielding, in over 95% of cases, a large number of excellent quality cells in prometaphase and late prophase. Thymidine presents advantages over amethopterin as the synchronizing agent. The block was released with either thymidine or 5-bromo-2'-deoxyuridine (BrdU) for which various concentrations were tested. The presence of colcemid was also evaluated. Without colcemid, the optimal length of the release period was precisely determined so that the wave of synchronized cells could be harvested while going through the early stages of mitoses. Subsequently, GTG, RHG and RBG banding were produced on these elongated chromosomes. A comprehensive approach to RHG banding ensured an easier and well reproducible banding technique. Nine interrelated factors which influence banding quality were studied. The analysis of their effects on this banding pattern revealed new data for the understanding of its mechanism. A high-resolution R-banding technique after BrdU incorporation and Giemsa staining is presented; it is simple, reliable and reproducible. Different conditions for the FPG (Fluorochrome-Photolysis-Giemsa) technique were studied in order to obtain sharper borders and higher contrast between positive and negative bands. Optimal conditions for the incorporation of BrdU and the FPG method produced excellent band separation and band contrast even in very elongated prophase chromosomes. They did not decrease the mitotic index, did not increase chromosomal damage significantly, and did not greatly vary from subject to subject nor with the age of the slide (between 1 day and 36 months). Homologue discordance correlated to band range was compared for GTG, RHG and RBG banding.", "classified_sentences": [ { "sentence": "A comparative analysis of variables for cell synchronization was made and led to the description of optimal conditions capable of yielding, in over 95% of cases, a large number of excellent quality cells in prometaphase and late prophase.", "category": "method" }, { "sentence": "Thymidine presents advantages over amethopterin as the synchronizing agent.", "category": "method" }, { "sentence": "The block was released with either thymidine or 5-bromo-2'-deoxyuridine (BrdU) for which various concentrations were tested.", "category": "method" }, { "sentence": "The presence of colcemid was also evaluated.", "category": "method" }, { "sentence": "Without colcemid, the optimal length of the release period was precisely determined so that the wave of synchronized cells could be harvested while going through the early stages of mitoses.", "category": "method" }, { "sentence": "Subsequently, GTG, RHG and RBG banding were produced on these elongated chromosomes.", "category": "method" }, { "sentence": "A comprehensive approach to RHG banding ensured an easier and well reproducible banding technique.", "category": "method" }, { "sentence": "Nine interrelated factors which influence banding quality were studied.", "category": "method" }, { "sentence": "The analysis of their effects on this banding pattern revealed new data for the understanding of its mechanism.", "category": "result" }, { "sentence": "A high-resolution R-banding technique after BrdU incorporation and Giemsa staining is presented; it is simple, reliable and reproducible.", "category": "method" }, { "sentence": "Different conditions for the FPG (Fluorochrome-Photolysis-Giemsa) technique were studied in order to obtain sharper borders and higher contrast between positive and negative bands.", "category": "method" }, { "sentence": "Optimal conditions for the incorporation of BrdU and the FPG method produced excellent band separation and band contrast even in very elongated prophase chromosomes.", "category": "result" }, { "sentence": "They did not decrease the mitotic index, did not increase chromosomal damage significantly, and did not greatly vary from subject to subject nor with the age of the slide (between 1 day and 36 months).", "category": "result" }, { "sentence": "Homologue discordance correlated to band range was compared for GTG, RHG and RBG banding.", "category": "result" } ] }, { "paper_id": "25664266", "title": "Frequency of anaemia in patients with systemic lupus erythematosus at tertiary care hospitals.", "abstract": "OBJECTIVE To analyze the frequency and causes of anaemia in systemic lupus erythematosus (SLE) patients attending in department of medicine at tertiary care hospitals. METHODS This retrospective, descriptive and analytical study was planned to analyze the frequency and causes of anaemia in SLE patients attending the department of medicine at (MMC) and (LUMHS) hospitals during the period of Jan 2006 to Nov 2008. The criteria used in this study were from the American College of Rheumatology. Investigations recorded were blood complete picture, absolute values, peripheral smear, and reticulocyte count in all patients of anaemia. These investigations were necessary to analyse the cases of anaemia in SLE. All investigations were not done in all cases. Patients with hypochromic microcytic anaemia were advised to have serum iron and ferritin levels, seven patients with macrocytic anaemia were advised to have direct and indirect coomb's test, LFTs, serum LDH, serum B12 and folate levels. Patients with normochromic and normocytic anaemia were considered to have anaemia of chronic disease. Bone marrow aspiration and Hb electrophoresis were done in two patients with anaemia of chronic disease. Thirty adult patients were included in this study. Special proforma were prepared to record the information from case sheets of patients including basic information, symptomatology and laboratory investigations. Severity and various types of anaemias were recorded. Anaemia was graded according to severity, as mild (Hb 10-12 G/dl), Moderate (Hb 8-10 G/dl) and severe (Hb < 8 G/dl). Haemoglobinopathies and other types of anaemias were excluded from study. RESULTS Thirty adult diagnosed patients of SLE, were included. Their ages ranged from twenty years to fifty years at time of presentation. The mean age +/- SD (range) was 28 +/- 6.22 (20-50) years and median age was 31 years. Out of thirty patients, twenty seven (90%) were females and three (10%) were males. Twenty eight (93.33%) patients presented with anaemia, 14 (46.66%) patients were of mild anaemia, 8 (26.66%) patients were of moderate grade anaemia and 6 (20%) patients had severe anaemia. Iron deficiency anaemia was found in 9 (30%) patients, 12 (40%) patients had anaemia of chronic disease and 7 (23.33%) patients had haemolytic anaemia, out of theses 7 patients, 5 (16.66%) patients had Coomb's positive haemolytic anaemia. All thirty patients had ANA positive titres > 1:80; and nineteen (63.33%) patients had anti ds DNA positive, titres > 1:10. CONCLUSION Haematologic abnormalities are common manifestations in patients with SLE. Most patients exhibit anaemia at some point during their disease course.", "classified_sentences": [ { "sentence": "To analyze the frequency and causes of anaemia in systemic lupus erythematosus (SLE) patients attending in department of medicine at tertiary care hospitals.", "category": "background" }, { "sentence": "This retrospective, descriptive and analytical study was planned to analyze the frequency and causes of anaemia in SLE patients attending the department of medicine at (MMC) and (LUMHS) hospitals during the period of Jan 2006 to Nov 2008.", "category": "method" }, { "sentence": "The criteria used in this study were from the American College of Rheumatology.", "category": "method" }, { "sentence": "Investigations recorded were blood complete picture, absolute values, peripheral smear, and reticulocyte count in all patients of anaemia.", "category": "method" }, { "sentence": "These investigations were necessary to analyse the cases of anaemia in SLE.", "category": "method" }, { "sentence": "All investigations were not done in all cases.", "category": "method" }, { "sentence": "Patients with hypochromic microcytic anaemia were advised to have serum iron and ferritin levels, seven patients with macrocytic anaemia were advised to have direct and indirect coomb's test, LFTs, serum LDH, serum B12 and folate levels.", "category": "method" }, { "sentence": "Patients with normochromic and normocytic anaemia were considered to have anaemia of chronic disease.", "category": "method" }, { "sentence": "Bone marrow aspiration and Hb electrophoresis were done in two patients with anaemia of chronic disease.", "category": "method" }, { "sentence": "Thirty adult patients were included in this study.", "category": "method" }, { "sentence": "Special proforma were prepared to record the information from case sheets of patients including basic information, symptomatology and laboratory investigations.", "category": "method" }, { "sentence": "Severity and various types of anaemias were recorded.", "category": "method" }, { "sentence": "Anaemia was graded according to severity, as mild (Hb 10-12 G/dl), Moderate (Hb 8-10 G/dl) and severe (Hb < 8 G/dl).", "category": "method" }, { "sentence": "Haemoglobinopathies and other types of anaemias were excluded from study.", "category": "method" }, { "sentence": "RESULTS Thirty adult diagnosed patients of SLE, were included.", "category": "result" }, { "sentence": "Their ages ranged from twenty years to fifty years at time of presentation.", "category": "result" }, { "sentence": "The mean age +/- SD (range) was 28 +/- 6.22 (20-50) years and median age was 31 years.", "category": "result" }, { "sentence": "Out of thirty patients, twenty seven (90%) were females and three (10%) were males.", "category": "result" }, { "sentence": "Twenty eight (93.33%) patients presented with anaemia, 14 (46.66%) patients were of mild anaemia, 8 (26.66%) patients were of moderate grade anaemia and 6 (20%) patients had severe anaemia.", "category": "result" }, { "sentence": "Iron deficiency anaemia was found in 9 (30%) patients, 12 (40%) patients had anaemia of chronic disease and 7 (23.33%) patients had haemolytic anaemia, out of these 7 patients, 5 (16.66%) patients had Coomb's positive haemolytic anaemia.", "category": "result" }, { "sentence": "All thirty patients had ANA positive titres > 1:80; and nineteen (63.33%) patients had anti ds DNA positive, titres > 1:10.", "category": "result" }, { "sentence": "CONCLUSION Haematologic abnormalities are common manifestations in patients with SLE.", "category": "result" }, { "sentence": "Most patients exhibit anaemia at some point during their disease course.", "category": "result" } ] }, { "paper_id": "25872811", "title": "Racial differences in the use of cardiac catheterization after acute myocardial infarction.", "abstract": "BACKGROUND Several studies have reported that black patients are less likely than white patients to undergo cardiac catheterization after acute myocardial infarction. The role of the race of the physician in this pattern is unknown. METHODS We analyzed data from the Cooperative Cardiovascular Project, a study of Medicare beneficiaries hospitalized for acute myocardial infarction in 1994 and 1995, to evaluate whether differences between black patients and white patients in the use of cardiac catheterization within 60 days after acute myocardial infarction varied according to the race of their attending physician. RESULTS Our study cohort consisted of 35,676 white and 4039 black patients with acute myocardial infarction who were treated by 17,550 white and 588 black physicians. Black patients had lower rates of cardiac catheterization than white patients, regardless of whether their attending physician was white (rate of catheterization, 38.4 percent vs. 45.7 percent; P< 0.001) or black (38.2 percent vs. 49.6 percent, P<0.001). We did not find a significant interaction between the race of the patients and the race of the physicians in the use of cardiac catheterization. The adjusted mortality rate among black patients was lower than or similar to that among white patients for up to three years after the infarction. CONCLUSIONS Racial differences in the use of cardiac catheterization are similar among patients treated by white physicians and those treated by black physicians, suggesting that this pattern of care is independent of the race of the physician.", "classified_sentences": [ { "sentence": "BACKGROUND Several studies have reported that black patients are less likely than white patients to undergo cardiac catheterization after acute myocardial infarction.", "category": "background" }, { "sentence": "The role of the race of the physician in this pattern is unknown.", "category": "background" }, { "sentence": "METHODS We analyzed data from the Cooperative Cardiovascular Project, a study of Medicare beneficiaries hospitalized for acute myocardial infarction in 1994 and 1995, to evaluate whether differences between black patients and white patients in the use of cardiac catheterization within 60 days after acute myocardial infarction varied according to the race of their attending physician.", "category": "method" }, { "sentence": "RESULTS Our study cohort consisted of 35,676 white and 4039 black patients with acute myocardial infarction who were treated by 17,550 white and 588 black physicians.", "category": "result" }, { "sentence": "Black patients had lower rates of cardiac catheterization than white patients, regardless of whether their attending physician was white (rate of catheterization, 38.4 percent vs. 45.7 percent; P< 0.001) or black (38.2 percent vs. 49.6 percent, P<0.001).", "category": "result" }, { "sentence": "We did not find a significant interaction between the race of the patients and the race of the physicians in the use of cardiac catheterization.", "category": "result" }, { "sentence": "The adjusted mortality rate among black patients was lower than or similar to that among white patients for up to three years after the infarction.", "category": "result" }, { "sentence": "CONCLUSIONS Racial differences in the use of cardiac catheterization are similar among patients treated by white physicians and those treated by black physicians, suggesting that this pattern of care is independent of the race of the physician.", "category": "result" } ] }, { "paper_id": "26041994", "title": "Angiotensin II receptor antagonist CV‐11974 and cerebral blood flow autoregulation", "abstract": "Objective: To investigate whether the angiotensin II (Ang II) subtype 1 receptor (AT1) antagonist CV-11974 had a similar effect to angiotensin converting enzyme inhibitors on cerebral blood flow autoregulation in normotensive Wistar-Kyoto (WKY) rats and spontaneously hypertensive rats (SHR). Methods: Sixteen WKY r ats and 16 SHR were given CV-11974 0.1 mg/kg intravenously and compared with two control groups (n = 16). Their cerebral blood flow was measured with the intracarotid xenon-133 injection method and blood pressure was raised by noradrenaline infusion and lowered by controlled haemorrhage in separate groups of rats. The limits of autoregulation were determined by computed least-sum-of-squares analysis. Results: The dose of CV-11974 given lowered blood pressure but did not influence baseline cerebral blood flow. In WKY rats the lower limit of autoregulation in control rats was 60$3 mmHg, whereas after CV-11974 administration it was 48$2 mmHg, respectively (P<0.01). In SHR the corresponding values were 85$2 and 78$2 mmHg, respectively (P<0.05). In WKY rats the upper limit of autoregulation in control rats was 144$5 mmHg, whereas after CV-11974 administration it was 126$7 mmHg (P<0.05). In SHR the corresponding figures were 174$8 and 144$6 mmHG, respectively (P<0.01). Conclusion: Thus, the AT1 receptor antagonist, although it did not influence baseline cerebral blood flow, shifted the autoregulation curve towards lower blood pressure. This effect is similar to that of angiotensin converting enzyme inhibitors, and might be due to release of Ang II-dependent tone in the larger cerebral resistance vessels.", "classified_sentences": [ { "sentence": "Objective: To investigate whether the angiotensin II (Ang II) subtype 1 receptor (AT1) antagonist CV-11974 had a similar effect to angiotensin converting enzyme inhibitors on cerebral blood flow autoregulation in normotensive Wistar-Kyoto (WKY) rats and spontaneously hypertensive rats (SHR).", "category": "background" }, { "sentence": "Methods: Sixteen WKY r ats and 16 SHR were given CV-11974 0.1 mg/kg intravenously and compared with two control groups (n = 16).", "category": "method" }, { "sentence": "Their cerebral blood flow was measured with the intracarotid xenon-133 injection method and blood pressure was raised by noradrenaline infusion and lowered by controlled haemorrhage in separate groups of rats.", "category": "method" }, { "sentence": "The limits of autoregulation were determined by computed least-sum-of-squares analysis.", "category": "method" }, { "sentence": "Results: The dose of CV-11974 given lowered blood pressure but did not influence baseline cerebral blood flow.", "category": "result" }, { "sentence": "In WKY rats the lower limit of autoregulation in control rats was 60$3 mmHg, whereas after CV-11974 administration it was 48$2 mmHg, respectively (P<0.01).", "category": "result" }, { "sentence": "In SHR the corresponding values were 85$2 and 78$2 mmHg, respectively (P<0.05).", "category": "result" }, { "sentence": "In WKY rats the upper limit of autoregulation in control rats was 144$5 mmHg, whereas after CV-11974 administration it was 126$7 mmHg (P<0.05).", "category": "result" }, { "sentence": "In SHR the corresponding figures were 174$8 and 144$6 mmHG, respectively (P<0.01).", "category": "result" }, { "sentence": "Conclusion: Thus, the AT1 receptor antagonist, although it did not influence baseline cerebral blood flow, shifted the autoregulation curve towards lower blood pressure.", "category": "result" }, { "sentence": "This effect is similar to that of angiotensin converting enzyme inhibitors, and might be due to release of Ang II-dependent tone in the larger cerebral resistance vessels.", "category": "result" } ] }, { "paper_id": "28585385", "title": "The activities of noradrenergic and dopaminergic neuron systems in experimental hydrocephalus.", "abstract": "Experimental hydrocephalus was induced in rabbits by intracisternal injection of kaolin suspension, and the concentration of noradrenaline (NA), dopamine (DA), and their metabolites was determined in several brain regions. The NA concentration had decreased in the cerebellum, hypothalamus, and pons plus medulla oblongata, and increased in the caudate nucleus at 2 days after kaolin injection (the stage of early intracranial hypertension). At 1 week (the stage of progressive hydrocephalus), the NA content had returned to control levels in all brain regions studied, and it decreased again at 4 weeks (the stage of chronic hydrocephalus) in the pons plus medulla oblongata. The DA level was unchanged throughout the 4-week period after kaolin injection. The concentration of 3-methoxy-4-hydroxyphenylethyleneglycol sulfate (MOPEG-SO4), the major metabolite of NA, was elevated in all brian regions except the caudate nucleus at all stages after kaolin injection. An increase in MOPEG-SO4 in the caudate nucleus was also observed 1 week after kaolin injection. The content of homovanillic acid (HVA), the major metabolite of DA in the rabbit brain, was decreased in the cerebral cortex at 2 days and at 1 week after kaolin injection, and in the caudate nucleus at 2 days and at 1 week, and 4 weeks. The level of HVA was increased in the hypothalamus at 2 days, in the cerebellum at 2 days and at 1 week, in the pons plus medulla oblongata at 2 days, 1 week, and 4 weeks, and in the midbrain at 4 weeks. These data suggest that, in experimental hydrocephalus in the rabbit, NA release is increased throughout the brain, while DA release is decreased in the cerebral cortex and caudate nucleus, and increased in the cerebellum, hypothalamus, midbrain, and pons plus medulla oblongata.", "classified_sentences": [ { "sentence": "Experimental hydrocephalus was induced in rabbits by intracisternal injection of kaolin suspension, and the concentration of noradrenaline (NA), dopamine (DA), and their metabolites was determined in several brain regions.", "category": "method" }, { "sentence": "The NA concentration had decreased in the cerebellum, hypothalamus, and pons plus medulla oblongata, and increased in the caudate nucleus at 2 days after kaolin injection (the stage of early intracranial hypertension).", "category": "result" }, { "sentence": "At 1 week (the stage of progressive hydrocephalus), the NA content had returned to control levels in all brain regions studied, and it decreased again at 4 weeks (the stage of chronic hydrocephalus) in the pons plus medulla oblongata.", "category": "result" }, { "sentence": "The DA level was unchanged throughout the 4-week period after kaolin injection.", "category": "result" }, { "sentence": "The concentration of 3-methoxy-4-hydroxyphenylethyleneglycol sulfate (MOPEG-SO4), the major metabolite of NA, was elevated in all brian regions except the caudate nucleus at all stages after kaolin injection.", "category": "result" }, { "sentence": "An increase in MOPEG-SO4 in the caudate nucleus was also observed 1 week after kaolin injection.", "category": "result" }, { "sentence": "The content of homovanillic acid (HVA), the major metabolite of DA in the rabbit brain, was decreased in the cerebral cortex at 2 days and at 1 week after kaolin injection, and in the caudate nucleus at 2 days and at 1 week, and 4 weeks.", "category": "result" }, { "sentence": "The level of HVA was increased in the hypothalamus at 2 days, in the cerebellum at 2 days and at 1 week, in the pons plus medulla oblongata at 2 days, 1 week, and 4 weeks, and in the midbrain at 4 weeks.", "category": "result" }, { "sentence": "These data suggest that, in experimental hydrocephalus in the rabbit, NA release is increased throughout the brain, while DA release is decreased in the cerebral cortex and caudate nucleus, and increased in the cerebellum, hypothalamus, midbrain, and pons plus medulla oblongata.", "category": "result" } ] }, { "paper_id": "28782860", "title": "Causes of school drop-out among ordinary level learners in a resettlement area in Masvingo, Zimbabwe", "abstract": "This article examined the causes of school drop-out among Ordinary level learners at a resettlement secondary school in Zimbabwe, with the aim of suggesting sound measures and solutions thus promoting learner retention in schools. The study was done in response to high dropout rates and poor academic performance of children in Zimbabwean resettlement areas. The study is informed by the attribution theory. In this study, a qualitative phenomenological case study design was used with focus group discussions, interviews and observations as data collection instruments to twelve (12) O' level learners and four (4) teachers at a resettlement secondary school in Masvingo province. Data analysis was done through the process of thematic coding. Findings revealed that poverty in households, child labour/household chores, broken families, poor supervision by parents, involvement in bad company/peer pressure, drug abuse, malnutrition and health related issues, low self-motivation and lack of interest in education among ordinary level learners were identified by teachers and learners as being the major factors influencing school dropouts. The study recommends that the government should increase the allocation of funds to resettlement areas to provide more amenities that promote learning. Further, parents should provide adequate school learning materials to their children such as school uniforms, school fees and stationery to encourage school retention. There should also be need for the government to introduce poverty alleviation strategies targeting resettled farmers since they tend to experience the impact of poverty in more adverse ways.", "classified_sentences": [ { "sentence": "This article examined the causes of school drop-out among Ordinary level learners at a resettlement secondary school in Zimbabwe, with the aim of suggesting sound measures and solutions thus promoting learner retention in schools.", "category": "background" }, { "sentence": "The study was done in response to high dropout rates and poor academic performance of children in Zimbabwean resettlement areas.", "category": "background" }, { "sentence": "The study is informed by the attribution theory.", "category": "background" }, { "sentence": "In this study, a qualitative phenomenological case study design was used with focus group discussions, interviews and observations as data collection instruments to twelve (12) O' level learners and four (4) teachers at a resettlement secondary school in Masvingo province.", "category": "method" }, { "sentence": "Data analysis was done through the process of thematic coding.", "category": "method" }, { "sentence": "Findings revealed that poverty in households, child labour/household chores, broken families, poor supervision by parents, involvement in bad company/peer pressure, drug abuse, malnutrition and health related issues, low self-motivation and lack of interest in education among ordinary level learners were identified by teachers and learners as being the major factors influencing school dropouts.", "category": "result" }, { "sentence": "The study recommends that the government should increase the allocation of funds to resettlement areas to provide more amenities that promote learning.", "category": "result" }, { "sentence": "Further, parents should provide adequate school learning materials to their children such as school uniforms, school fees and stationery to encourage school retention.", "category": "result" }, { "sentence": "There should also be need for the government to introduce poverty alleviation strategies targeting resettled farmers since they tend to experience the impact of poverty in more adverse ways.", "category": "result" } ] }, { "paper_id": "28881965", "title": "COSTS AND LENGTH OF STAY ASSOCIATED WITH ANTIMICROBIAL RESISTANCE IN ACUTE KIDNEY INJURY PATIENTS WITH BLOODSTREAM INFECTION", "abstract": "Abstract Introduction: Antimicrobial resistance negatively impacts on prognosis. Intensive care unit (ICU) patients, and particularly those with acute kidney injury (AKI), are at high risk for developing nosocomial bloodstream infections (BSI) due to multi-drug-resistant strains. Economic implications in terms of costs and length of stay (LOS) attributable to antimicrobial resistance are underevaluated. This study aimed to assess whether microbial susceptibility patterns affect costs and LOS in a well-defined cohort of ICU patients with AKI undergoing renal replacement therapy (RRT) who developed nosocomial BSI. Methods: Historical study (1995-2004) enrolling all adult RRT-dependent ICU patients with AKI and nosocomial BSI. Costs were considered as invoiced in the Belgian reimbursement system, and LOS was used as a surrogate marker for hospital resource allocation. Results: Of the 1330 patients with AKI undergoing RRT, 92 had microbiologic evidence of nosocomial BSI (57/92, 62% due to a multi-drug-resistant microorganism). Main patient characteristics were equal in both groups. As compared to patients with antimicrobial- susceptible BSI, patients with antimicrobialresistant BSI were more likely to acquire Grampositive infection (72.6% vs 25.5%, P<0.001). No differences were found neither in LOS (ICU before BSI, ICU, hospital before BSI, hospital, hospital after BSI, and time on RRT; all P>0.05) or hospital costs (all P>0.05) when comparing patients with antimicrobial-resistant vs antimicrobial-susceptible BSI. However, although not statistically significant, patients with BSI caused by resistant Gram-negative-, Candida-, or anaerobic bacteria incurred substantial higher costs than those without. Conclusion: In a cohort of ICU patients with AKI and nosocomial BSI undergoing RRT, patients with antimicrobial-resistant vs antimicrobial-susceptible Gram-positive BSI did not have longer hospital stays, or higher hospital costs. Patients with resistant “other” (i.e. Gram-negative, Candida, or anaerobic) BSI were found to have a distinct trend towards increased resources use as compared to patients with susceptible “other” BSI, respectively.", "classified_sentences": [ { "sentence": "Abstract Introduction: Antimicrobial resistance negatively impacts on prognosis.", "category": "background" }, { "sentence": "Intensive care unit (ICU) patients, and particularly those with acute kidney injury (AKI), are at high risk for developing nosocomial bloodstream infections (BSI) due to multi-drug-resistant strains.", "category": "background" }, { "sentence": "Economic implications in terms of costs and length of stay (LOS) attributable to antimicrobial resistance are underevaluated.", "category": "background" }, { "sentence": "This study aimed to assess whether microbial susceptibility patterns affect costs and LOS in a well-defined cohort of ICU patients with AKI undergoing renal replacement therapy (RRT) who developed nosocomial BSI.", "category": "method" }, { "sentence": "Methods: Historical study (1995-2004) enrolling all adult RRT-dependent ICU patients with AKI and nosocomial BSI.", "category": "method" }, { "sentence": "Costs were considered as invoiced in the Belgian reimbursement system, and LOS was used as a surrogate marker for hospital resource allocation.", "category": "method" }, { "sentence": "Results: Of the 1330 patients with AKI undergoing RRT, 92 had microbiologic evidence of nosocomial BSI (57/92, 62% due to a multi-drug-resistant microorganism).", "category": "result" }, { "sentence": "Main patient characteristics were equal in both groups.", "category": "result" }, { "sentence": "As compared to patients with antimicrobial- susceptible BSI, patients with antimicrobialresistant BSI were more likely to acquire Grampositive infection (72.6% vs 25.5%, P<0.001).", "category": "result" }, { "sentence": "No differences were found neither in LOS (ICU before BSI, ICU, hospital before BSI, hospital, hospital after BSI, and time on RRT; all P>0.05) or hospital costs (all P>0.05) when comparing patients with antimicrobial-resistant vs antimicrobial-susceptible BSI.", "category": "result" }, { "sentence": "However, although not statistically significant, patients with BSI caused by resistant Gram-negative-, Candida-, or anaerobic bacteria incurred substantial higher costs than those without.", "category": "result" }, { "sentence": "Conclusion: In a cohort of ICU patients with AKI and nosocomial BSI undergoing RRT, patients with antimicrobial-resistant vs antimicrobial-susceptible Gram-positive BSI did not have longer hospital stays, or higher hospital costs.", "category": "result" }, { "sentence": "Patients with resistant “other” (i.e. Gram-negative, Candida, or anaerobic) BSI were found to have a distinct trend towards increased resources use as compared to patients with susceptible “other” BSI, respectively.", "category": "result" } ] }, { "paper_id": "164215758", "title": "Systematic Modeling of Sludge Filtration Process Using Dimensional Analysis Technique", "abstract": "The handling and disposal of sludge; the largest constituent removed in the process of treating wastewater is one of the greatest challenges facing the environmental engineer. The sludge has high water content and compressibility attribute and as such it is expedient to dewater it to reduce its volume and prevent environmental health hazard. This study presents a sludge filtration equation which incorporates the compressibility coefficient using a modified dimensional analysis technique. The equation was validated using experimental data from a pilot scale sand drying bed equipment yielding a close agreement between the theoretical and experimental values of the slope and intercept with a correlation coefficient ranging from 0.940. 98. The experimental slope and intercept was found to be (1260913.48 s/m 6 , 4872.53 s/m 3 ) (5359604.57 s/m 6 , 844882.56 s/m 3 ), (112117050.4 s/m 6 , -2135816.16 s/m 3 ), (145562880 s/m 6 , -30497917.03 s/m 3 ) while the theoretical slopes and intercepts are (1257426.75 s/m 6 , 5270.26 s/m 3 ),( 4579418.42 s/m 6 , 905658.24 s/m 3 ), (112117075 s/m 6 , 21358166.74 s/m 3 ), (206699290.5 s/m 6 , -4589555.58 s/m 3 ) respectively. The equation also accounts for the compressibility attribute believed to affect sludge filtration process.", "classified_sentences": [ { "sentence": "The handling and disposal of sludge; the largest constituent removed in the process of treating wastewater is one of the greatest challenges facing the environmental engineer.", "category": "background" }, { "sentence": "The sludge has high water content and compressibility attribute and as such it is expedient to dewater it to reduce its volume and prevent environmental health hazard.", "category": "background" }, { "sentence": "This study presents a sludge filtration equation which incorporates the compressibility coefficient using a modified dimensional analysis technique.", "category": "method" }, { "sentence": "The equation was validated using experimental data from a pilot scale sand drying bed equipment yielding a close agreement between the theoretical and experimental values of the slope and intercept with a correlation coefficient ranging from 0.940.", "category": "result" }, { "sentence": "98.", "category": "result" }, { "sentence": "The experimental slope and intercept was found to be (1260913.48 s/m 6 , 4872.53 s/m 3 ) (5359604.57 s/m 6 , 844882.56 s/m 3 ), (112117050.4 s/m 6 , -2135816.16 s/m 3 ), (145562880 s/m 6 , -30497917.03 s/m 3 ) while the theoretical slopes and intercepts are (1257426.75 s/m 6 , 5270.26 s/m 3 ),( 4579418.42 s/m 6 , 905658.24 s/m 3 ), (112117075 s/m 6 , 21358166.74 s/m 3 ), (206699290.5 s/m 6 , -4589555.58 s/m 3 ) respectively.", "category": "result" }, { "sentence": "The equation also accounts for the compressibility attribute believed to affect sludge filtration process.", "category": "method" } ] }, { "paper_id": "30616147", "title": "Paclitaxel plus doxorubicin in metastatic breast Ca: the Milan experience.", "abstract": "A pilot study conducted at the National Cancer Institute in Milan, Italy assessed the efficacy of six or eight cycles of paclitaxel (Taxol) 200 mg/m2 q3wks plus doxorubicin (Adriamycin) (60 mg/m2q3wks) in 49 women with metastatic breast cancer who had received no prior chemotherapy. This study suggested that paclitaxel and doxorubicin should be considered for previously untreated patients who have been initially diagnosed with metastatic breast cancer and have a good performance status. The probability of complete response in these patients can be enhanced by continuing treatment with single-agent paclitaxel, and the risk of cardiotoxicity can be minimized by keeping the cumulative dose of doxorubicin below 360 mg/m2. Based on the experience in Milan, this combination is one of the most effective regimens for the treatment of women with breast cancer. Indeed, the convenience, efficacy, and tolerability of the combination justify the large trials that are currently evaluating its effects in women with operable breast cancer.", "classified_sentences": [ { "sentence": "A pilot study conducted at the National Cancer Institute in Milan, Italy assessed the efficacy of six or eight cycles of paclitaxel (Taxol) 200 mg/m2 q3wks plus doxorubicin (Adriamycin) (60 mg/m2q3wks) in 49 women with metastatic breast cancer who had received no prior chemotherapy.", "category": "method" }, { "sentence": "This study suggested that paclitaxel and doxorubicin should be considered for previously untreated patients who have been initially diagnosed with metastatic breast cancer and have a good performance status.", "category": "result" }, { "sentence": "The probability of complete response in these patients can be enhanced by continuing treatment with single-agent paclitaxel, and the risk of cardiotoxicity can be minimized by keeping the cumulative dose of doxorubicin below 360 mg/m2.", "category": "result" }, { "sentence": "Based on the experience in Milan, this combination is one of the most effective regimens for the treatment of women with breast cancer.", "category": "result" }, { "sentence": "Indeed, the convenience, efficacy, and tolerability of the combination justify the large trials that are currently evaluating its effects in women with operable breast cancer.", "category": "result" } ] }, { "paper_id": "34378973", "title": "Prognostic relevance of CCAs/Ph- in CML settled.", "abstract": "In this issue of Blood, Issa et al1 report that patients with chronic-phase chronic myeloid leukemia (CML) in remission after treatment with tyrosine kinase inhibitors (TKIs) with clonal chromosomal aberrations (CCAs) in Philadelphia chromosome–negative (Ph−) metaphases had a significantly worse survival than similar patients without CCAs. This study is notable for the large number of patients with CCAs (n = 58) and the length of follow up (median, 7.6 years) (see figure).", "classified_sentences": [ { "sentence": "In this issue of Blood, Issa et al1 report that patients with chronic-phase chronic myeloid leukemia (CML) in remission after treatment with tyrosine kinase inhibitors (TKIs) with clonal chromosomal aberrations (CCAs) in Philadelphia chromosome–negative (Ph−) metaphases had a significantly worse survival than similar patients without CCAs.", "category": "background" }, { "sentence": "This study is notable for the large number of patients with CCAs (n = 58) and the length of follow up (median, 7.6 years) (see figure).", "category": "method" } ] }, { "paper_id": "37743218", "title": "Vision as a Fundamentally Statistical Machine", "abstract": "As a vital driving force of systems neuroscience, visual neuroscience had its conceptual framework established more than 40 years ago based on Hubel and Wiesel’s groundbreaking work on the receptive-field properties of visual neurons (Hubel & Wiesel, 1977). This framework was subsequently strengthened by David Marr's influential book (Marr, 2010). In this paradigm, visual neurons are conceived to perform bottom-up, imagebased processing to build a series of symbolic representations of visual stimuli. This paradigm, however, is deeply misleading since the generative sources in the threedimensional (3D) physical world of any stimulus, to which visual animals must respond successfully, cannot be determined by image-based processing (due to the inverse optics problem). This is perhaps the reason why \"Now, thirty years later, the main problems that occupied Marr remain fundamental open problems in the study of perception\" (Marr, 2010), as assessed by two prominent vision scientists and Marr's close associates.", "classified_sentences": [ { "sentence": "As a vital driving force of systems neuroscience, visual neuroscience had its conceptual framework established more than 40 years ago based on Hubel and Wiesel’s groundbreaking work on the receptive-field properties of visual neurons (Hubel & Wiesel, 1977).", "category": "background" }, { "sentence": "This framework was subsequently strengthened by David Marr's influential book (Marr, 2010).", "category": "background" }, { "sentence": "In this paradigm, visual neurons are conceived to perform bottom-up, image-based processing to build a series of symbolic representations of visual stimuli.", "category": "background" }, { "sentence": "This paradigm, however, is deeply misleading since the generative sources in the three-dimensional (3D) physical world of any stimulus, to which visual animals must respond successfully, cannot be determined by image-based processing (due to the inverse optics problem).", "category": "background" }, { "sentence": "This is perhaps the reason why \"Now, thirty years later, the main problems that occupied Marr remain fundamental open problems in the study of perception\" (Marr, 2010), as assessed by two prominent vision scientists and Marr's close associates.", "category": "background" } ] }, { "paper_id": "46347940", "title": "The primary percutaneous coronary intervention for acute anterior myocardial infarction in a middle-aged male patient with bilateral coronary artery to pulmonary artery fistulas", "abstract": "A 38-year-old man admitted to emergency department with 2 h of typical substernal chest pain, shortness of breath and nausea. The ECG revealed sinus rhythm with a 3 mm ST elevation in precordial leads V1–V6. The coronary angiography revealed acute total occlusion in left anterior descending artery (LAD) with normal circumflex and right coronary artery (RCA) along with bilateral fistulas arising from the proximal LAD and ostial RCA draining into the main pulmonary artery. Therefore, primary percutaneous coronary intervention and bare metal stent implantation was performed to culprit LAD lesion. The electrocardiographically gated 64-slice multidetector-row CT showed two large, tortuous abnormal vessels which arose from the both ostial part of the RCA and LAD draining into the main pulmonary artery. We report an unusual case of bilateral coronary artery to pulmonary artery fistulas leading to acute anterior myocardial infarction in a middle-aged male patient.", "classified_sentences": [ { "sentence": "A 38-year-old man admitted to emergency department with 2 h of typical substernal chest pain, shortness of breath and nausea.", "category": "background" }, { "sentence": "The ECG revealed sinus rhythm with a 3 mm ST elevation in precordial leads V1–V6.", "category": "background" }, { "sentence": "The coronary angiography revealed acute total occlusion in left anterior descending artery (LAD) with normal circumflex and right coronary artery (RCA) along with bilateral fistulas arising from the proximal LAD and ostial RCA draining into the main pulmonary artery.", "category": "background" }, { "sentence": "Therefore, primary percutaneous coronary intervention and bare metal stent implantation was performed to culprit LAD lesion.", "category": "method" }, { "sentence": "The electrocardiographically gated 64-slice multidetector-row CT showed two large, tortuous abnormal vessels which arose from the both ostial part of the RCA and LAD draining into the main pulmonary artery.", "category": "method" }, { "sentence": "We report an unusual case of bilateral coronary artery to pulmonary artery fistulas leading to acute anterior myocardial infarction in a middle-aged male patient.", "category": "result" } ] }, { "paper_id": "49434640", "title": "Dynamic Match Kernel With Deep Convolutional Features for Image Retrieval", "abstract": "For image retrieval methods based on bag of visual words, much attention has been paid to enhancing the discriminative powers of the local features. Although retrieved images are usually similar to a query in minutiae, they may be significantly different from a semantic perspective, which can be effectively distinguished by convolutional neural networks (CNN). Such images should not be considered as relevant pairs. To tackle this problem, we propose to construct a dynamic match kernel by adaptively calculating the matching thresholds between query and candidate images based on the pairwise distance among deep CNN features. In contrast to the typical static match kernel which is independent to the global appearance of retrieved images, the dynamic one leverages the semantical similarity as a constraint for determining the matches. Accordingly, we propose a semantic-constrained retrieval framework by incorporating the dynamic match kernel, which focuses on matched patches between relevant images and filters out the ones for irrelevant pairs. Furthermore, we demonstrate that the proposed kernel complements recent methods, such as hamming embedding, multiple assignment, local descriptors aggregation, and graph-based re-ranking, while it outperforms the static one under various settings on off-the-shelf evaluation metrics. We also propose to evaluate the matched patches both quantitatively and qualitatively. Extensive experiments on five benchmark data sets and large-scale distractors validate the merits of the proposed method against the state-of-the-art methods for image retrieval.", "classified_sentences": [ { "sentence": "For image retrieval methods based on bag of visual words, much attention has been paid to enhancing the discriminative powers of the local features.", "category": "background" }, { "sentence": "Although retrieved images are usually similar to a query in minutiae, they may be significantly different from a semantic perspective, which can be effectively distinguished by convolutional neural networks (CNN).", "category": "background" }, { "sentence": "Such images should not be considered as relevant pairs.", "category": "background" }, { "sentence": "To tackle this problem, we propose to construct a dynamic match kernel by adaptively calculating the matching thresholds between query and candidate images based on the pairwise distance among deep CNN features.", "category": "method" }, { "sentence": "In contrast to the typical static match kernel which is independent to the global appearance of retrieved images, the dynamic one leverages the semantical similarity as a constraint for determining the matches.", "category": "method" }, { "sentence": "Accordingly, we propose a semantic-constrained retrieval framework by incorporating the dynamic match kernel, which focuses on matched patches between relevant images and filters out the ones for irrelevant pairs.", "category": "method" }, { "sentence": "Furthermore, we demonstrate that the proposed kernel complements recent methods, such as hamming embedding, multiple assignment, local descriptors aggregation, and graph-based re-ranking, while it outperforms the static one under various settings on off-the-shelf evaluation metrics.", "category": "result" }, { "sentence": "We also propose to evaluate the matched patches both quantitatively and qualitatively.", "category": "method" }, { "sentence": "Extensive experiments on five benchmark data sets and large-scale distractors validate the merits of the proposed method against the state-of-the-art methods for image retrieval.", "category": "result" } ] }, { "paper_id": "52800221", "title": "Iterative quantization: A procrustean approach to learning binary codes", "abstract": "This paper addresses the problem of learning similarity-preserving binary codes for efficient retrieval in large-scale image collections. We propose a simple and efficient alternating minimization scheme for finding a rotation of zero-centered data so as to minimize the quantization error of mapping this data to the vertices of a zero-centered binary hypercube. This method, dubbed iterative quantization (ITQ), has connections to multi-class spectral clustering and to the orthogonal Procrustes problem, and it can be used both with unsupervised data embeddings such as PCA and supervised embeddings such as canonical correlation analysis (CCA). Our experiments show that the resulting binary coding schemes decisively outperform several other state-of-the-art methods.", "classified_sentences": [ { "sentence": "This paper addresses the problem of learning similarity-preserving binary codes for efficient retrieval in large-scale image collections.", "category": "background" }, { "sentence": "We propose a simple and efficient alternating minimization scheme for finding a rotation of zero-centered data so as to minimize the quantization error of mapping this data to the vertices of a zero-centered binary hypercube.", "category": "method" }, { "sentence": "This method, dubbed iterative quantization (ITQ), has connections to multi-class spectral clustering and to the orthogonal Procrustes problem, and it can be used both with unsupervised data embeddings such as PCA and supervised embeddings such as canonical correlation analysis (CCA).", "category": "method" }, { "sentence": "Our experiments show that the resulting binary coding schemes decisively outperform several other state-of-the-art methods.", "category": "result" } ] }, { "paper_id": "53049289", "title": "Ergonomic Risk Assessment of Sonographers during Abdominal and Pelvic Sonography", "abstract": "Introduction: Identification of scan specific risk factors is crucial for controlling the increasing incidence ofwork-related musculoskeletal disorders in sonographers. Objectives: The purpose of this study was to conduct a physical ergonomic assessment of sonographers during abdominal and pelvic sonographies by RULA method and to study if any correlation exists between symptoms and RULA. Settings and Design: Cross sectional study design; Multicentric, tertiary health care centers. Methodology: 50 sonographers performing procedures for at least one year and consenting to participate were included in the study. Information of demographic and workload data along with past and present history of musculoskeletal symptoms which they attributed to work-related practices was obtained via questionnaires. RULA assessment was conducted while the sonologist was performing an abdominal/obstetric sonography scan, as it encompasses the major workload of sonologist. Statistical Analysis: Analysis of symptom prevalence and mean RULA scores was done and further statistical analysis was performed by Pearson’s correlation statistics to check if any relationship exists between the RULA and symptoms. Result:58% sonographers reported neck pain, 32% low back and 13% upper back pain, 40% wrist and hand, 14% shoulder pain in the last week. The average total right RULA score and total left RULA score were 6.3 and 5.6 respectively. Statistical correlation represents mildpositivecorrelation (r=0.42, p<0.002*) between neck, trunk and leg RULA score and upper limb symptoms and weak (r=0.47,p<0.000*) relationship between right upper limb RULA scoreand spine symptoms. Conclusion: The one-week prevalence of work-related symptoms were 84% andone-year prevalencewas 58% in the sonographers participating in this study.The high RULA scores(>5)recorded in this study demand urgent attention and need to implement ergonomic changes in the working postures of sonographers.", "classified_sentences": [ { "sentence": "Introduction: Identification of scan specific risk factors is crucial for controlling the increasing incidence ofwork-related musculoskeletal disorders in sonographers.", "category": "background" }, { "sentence": "Objectives: The purpose of this study was to conduct a physical ergonomic assessment of sonographers during abdominal and pelvic sonographies by RULA method and to study if any correlation exists between symptoms and RULA.", "category": "method" }, { "sentence": "Settings and Design: Cross sectional study design; Multicentric, tertiary health care centers.", "category": "method" }, { "sentence": "Methodology: 50 sonographers performing procedures for at least one year and consenting to participate were included in the study.", "category": "method" }, { "sentence": "Information of demographic and workload data along with past and present history of musculoskeletal symptoms which they attributed to work-related practices was obtained via questionnaires.", "category": "method" }, { "sentence": "RULA assessment was conducted while the sonologist was performing an abdominal/obstetric sonography scan, as it encompasses the major workload of sonologist.", "category": "method" }, { "sentence": "Statistical Analysis: Analysis of symptom prevalence and mean RULA scores was done and further statistical analysis was performed by Pearson’s correlation statistics to check if any relationship exists between the RULA and symptoms.", "category": "method" }, { "sentence": "Result:58% sonographers reported neck pain, 32% low back and 13% upper back pain, 40% wrist and hand, 14% shoulder pain in the last week.", "category": "result" }, { "sentence": "The average total right RULA score and total left RULA score were 6.3 and 5.6 respectively.", "category": "result" }, { "sentence": "Statistical correlation represents mildpositivecorrelation (r=0.42, p<0.002*) between neck, trunk and leg RULA score and upper limb symptoms and weak (r=0.47,p<0.000*) relationship between right upper limb RULA scoreand spine symptoms.", "category": "result" }, { "sentence": "Conclusion: The one-week prevalence of work-related symptoms were 84% andone-year prevalencewas 58% in the sonographers participating in this study.The high RULA scores(>5)recorded in this study demand urgent attention and need to implement ergonomic changes in the working postures of sonographers.", "category": "result" } ] }, { "paper_id": "53051765", "title": "Developing team skills with self- and peer assessment: Are benefits inversely related to team function?", "abstract": "Purpose – Self‐ and peer assessment has proved effective in promoting the development of teamwork and other professional skills in undergraduate students. However, in previous research approximately 30 percent of students reported that its use produced no perceived improvement in their teamwork experience. It was hypothesised that a significant number of these students were probably members of a team that would have functioned well without self‐ and peer assessment and hence the process did not improve their teamwork experience. This paper aims to report the testing of this hypothesis.Design/methodology/approach – The paper reviews some of the literature on self‐ and peer assessment, outlines the online self‐ and peer assessment tool SPARKPLUS, and analyses the results of a post‐subject survey of students in a large multi‐disciplinary engineering design subject.Findings – It was found that students who were neutral as to whether self‐ and peer assessment improved their teamwork experience cannot be assume.", "classified_sentences": [ { "sentence": "Purpose – Self‐and peer assessment has proved effective in promoting the development of teamwork and other professional skills in undergraduate students.", "category": "background" }, { "sentence": "However, in previous research approximately 30 percent of students reported that its use produced no perceived improvement in their teamwork experience.", "category": "background" }, { "sentence": "It was hypothesised that a significant number of these students were probably members of a team that would have functioned well without self‐and peer assessment and hence the process did not improve their teamwork experience.", "category": "background" }, { "sentence": "This paper aims to report the testing of this hypothesis.", "category": "method" }, { "sentence": "Design/methodology/approach – The paper reviews some of the literature on self‐and peer assessment, outlines the online self‐and peer assessment tool SPARKPLUS, and analyses the results of a post‐subject survey of students in a large multi‐disciplinary engineering design subject.", "category": "method" }, { "sentence": "Findings – It was found that students who were neutral as to whether self‐and peer assessment improved their teamwork experience cannot be assume.", "category": "result" } ] }, { "paper_id": "53079390", "title": "Similar but not the Same: Word Sense Disambiguation Improves Event Detection via Neural Representation Matching", "abstract": "Event detection (ED) and word sense disambiguation (WSD) are two similar tasks in that they both involve identifying the classes (i.e. event types or word senses) of some word in a given sentence. It is thus possible to extract the knowledge hidden in the data for WSD, and utilize it to improve the performance on ED. In this work, we propose a method to transfer the knowledge learned on WSD to ED by matching the neural representations learned for the two tasks. Our experiments on two widely used datasets for ED demonstrate the effectiveness of the proposed method.", "classified_sentences": [ { "sentence": "Event detection (ED) and word sense disambiguation (WSD) are two similar tasks in that they both involve identifying the classes (i.e. event types or word senses) of some word in a given sentence.", "category": "background" }, { "sentence": "It is thus possible to extract the knowledge hidden in the data for WSD, and utilize it to improve the performance on ED. In this work, we propose a method to transfer the knowledge learned on WSD to ED by matching the neural representations learned for the two tasks.", "category": "method" }, { "sentence": "Our experiments on two widely used datasets for ED demonstrate the effectiveness of the proposed method.", "category": "result" } ] }, { "paper_id": "53554040", "title": "Modern History of Cathodic Arc Coating", "abstract": "Introduction Coatings based on the condensation of cathodic arc plasmas are widely used for a range of applications including hard and wear resistant coatings on tools and automotive parts and corrosion resistant and decorative coatings on building supplies such as faucets and handles. The origin of the technology can be traced back to the 18th century. Many observations were closely related to the development of electrical power sources and early electrical engineering. Major advancements on cathodic arc physics were made after the electron was discovered (J.J. Thomson 1897) and the structure of the atom was roughly understood (E. Rutherford 1911). The early history of arc discharges and first arc-based coatings is described in other publications [1-3]. The actual industrial use of cathodic arc technology evolved in the 1970s in the Soviet Union and 1980s in the Western hemisphere. In this contribution, the modern history is briefly reviewed, including the development of industrial arc coaters for hard and decorative coatings, the need for the implementation of macroparticle filtering, and the use of arc plasmas for the fabrication of multilayers, composite coatings, and nanostructures.", "classified_sentences": [ { "sentence": "Introduction Coatings based on the condensation of cathodic arc plasmas are widely used for a range of applications including hard and wear resistant coatings on tools and automotive parts and corrosion resistant and decorative coatings on building supplies such as faucets and handles.", "category": "background" }, { "sentence": "The origin of the technology can be traced back to the 18th century.", "category": "background" }, { "sentence": "Many observations were closely related to the development of electrical power sources and early electrical engineering.", "category": "background" }, { "sentence": "Major advancements on cathodic arc physics were made after the electron was discovered (J.J. Thomson 1897) and the structure of the atom was roughly understood (E. Rutherford 1911).", "category": "background" }, { "sentence": "The early history of arc discharges and first arc-based coatings is described in other publications [1-3].", "category": "background" }, { "sentence": "The actual industrial use of cathodic arc technology evolved in the 1970s in the Soviet Union and 1980s in the Western hemisphere.", "category": "background" }, { "sentence": "In this contribution, the modern history is briefly reviewed, including the development of industrial arc coaters for hard and decorative coatings, the need for the implementation of macroparticle filtering, and the use of arc plasmas for the fabrication of multilayers, composite coatings, and nanostructures.", "category": "background" } ] }, { "paper_id": "53757775", "title": "A Scalable Optimization Mechanism for Pairwise Based Discrete Hashing", "abstract": "Maintaining the pairwise relationship among originally high-dimensional data into a low-dimensional binary space is a popular strategy to learn binary codes. One simple and intuitive method is to utilize two identical code matrices produced by hash functions to approximate a pairwise real label matrix. However, the resulting quartic problem in term of hash functions is difficult to directly solve due to the non-convex and non-smooth nature of the objective. In this paper, unlike previous optimization methods using various relaxation strategies, we aim to directly solve the original quartic problem using a novel alternative optimization mechanism to linearize the quartic problem by introducing a linear regression model. Additionally, we find that gradually learning each batch of binary codes in a sequential mode, i.e. batch by batch, is greatly beneficial to the convergence of binary code learning. Based on this significant discovery and the proposed strategy, we introduce a scalable symmetric discrete hashing algorithm that gradually and smoothly updates each batch of binary codes. To further improve the smoothness, we also propose a greedy symmetric discrete hashing algorithm to update each bit of batch binary codes. Moreover, we extend the proposed optimization mechanism to solve the non-convex optimization problems for binary code learning in many other pairwise based hashing algorithms. Extensive experiments on benchmark single-label and multi-label databases demonstrate the superior performance of the proposed mechanism over recent state-of-the-art methods on two kinds of retrieval tasks: similarity and ranking order. The source codes are available on https://github.com/xsshi2015/Scalable-Pairwise-based-Discrete-Hashing.", "classified_sentences": [ { "sentence": "Maintaining the pairwise relationship among originally high-dimensional data into a low-dimensional binary space is a popular strategy to learn binary codes.", "category": "background" }, { "sentence": "One simple and intuitive method is to utilize two identical code matrices produced by hash functions to approximate a pairwise real label matrix.", "category": "method" }, { "sentence": "However, the resulting quartic problem in term of hash functions is difficult to directly solve due to the non-convex and non-smooth nature of the objective.", "category": "background" }, { "sentence": "In this paper, unlike previous optimization methods using various relaxation strategies, we aim to directly solve the original quartic problem using a novel alternative optimization mechanism to linearize the quartic problem by introducing a linear regression model.", "category": "method" }, { "sentence": "Additionally, we find that gradually learning each batch of binary codes in a sequential mode, i.e. batch by batch, is greatly beneficial to the convergence of binary code learning.", "category": "method" }, { "sentence": "Based on this significant discovery and the proposed strategy, we introduce a scalable symmetric discrete hashing algorithm that gradually and smoothly updates each batch of binary codes.", "category": "method" }, { "sentence": "To further improve the smoothness, we also propose a greedy symmetric discrete hashing algorithm to update each bit of batch binary codes.", "category": "method" }, { "sentence": "Moreover, we extend the proposed optimization mechanism to solve the non-convex optimization problems for binary code learning in many other pairwise based hashing algorithms.", "category": "method" }, { "sentence": "Extensive experiments on benchmark single-label and multi-label databases demonstrate the superior performance of the proposed mechanism over recent state-of-the-art methods on two kinds of retrieval tasks: similarity and ranking order.", "category": "result" }, { "sentence": "The source codes are available on https://github.com/xsshi2015/Scalable-Pairwise-based-Discrete-Hashing.", "category": "result" } ] }, { "paper_id": "54164372", "title": "EXPLORING THE INTERSECTION BETWEEN LITERARY AND DIGITAL LITERACY IN SCHOOL", "abstract": "Two related questions are explored in this paper: What is the motivation for a literary blog activity in a “book-flooding” program? Under what conditions can such a blog activity be successfully implemented in the classroom? In order to answer these questions, we address the relation between literary and digital literacy. We see literary and digital literacy as parts of the overall literacy of the students. These parts of literacy are important in themselves, but we present research to show that they also can interact with each other and with other aspects of literacy in ways that have an accelerating effect on the students‟ literacy as a whole. The research reported from serves as motivation for our own blog activity, which we present here in detail to demonstrate how such a digital literacy can be implemented. The presentation is accompanied by a discussion of some important conditions that are important for using digital tools as part of a potentially literacy enhancing “book-flooding” program in class.", "classified_sentences": [ { "sentence": "Two related questions are explored in this paper: What is the motivation for a literary blog activity in a “book-flooding” program?", "category": "background" }, { "sentence": "Under what conditions can such a blog activity be successfully implemented in the classroom?", "category": "background" }, { "sentence": "In order to answer these questions, we address the relation between literary and digital literacy.", "category": "method" }, { "sentence": "We see literary and digital literacy as parts of the overall literacy of the students.", "category": "background" }, { "sentence": "These parts of literacy are important in themselves, but we present research to show that they also can interact with each other and with other aspects of literacy in ways that have an accelerating effect on the students‟ literacy as a whole.", "category": "background" }, { "sentence": "The research reported from serves as motivation for our own blog activity, which we present here in detail to demonstrate how such a digital literacy can be implemented.", "category": "method" }, { "sentence": "The presentation is accompanied by a discussion of some important conditions that are important for using digital tools as part of a potentially literacy enhancing “book-flooding” program in class.", "category": "method" } ] }, { "paper_id": "54436148", "title": "Comparing neural‐ and N‐gram‐based language models for word segmentation", "abstract": "Word segmentation is the task of inserting or deleting word boundary characters in order to separate character sequences that correspond to words in some language. In this article we propose an approach based on a beam search algorithm and a language model working at the byte/character level, the latter component implemented either as an n‐gram model or a recurrent neural network. The resulting system analyzes the text input with no word boundaries one token at a time, which can be a character or a byte, and uses the information gathered by the language model to determine if a boundary must be placed in the current position or not. Our aim is to use this system in a preprocessing step for a microtext normalization system. This means that it needs to effectively cope with the data sparsity present on this kind of texts. We also strove to surpass the performance of two readily available word segmentation systems: The well‐known and accessible Word Breaker by Microsoft, and the Python module WordSegment by Grant Jenks. The results show that we have met our objectives, and we hope to continue to improve both the precision and the efficiency of our system in the future.", "classified_sentences": [ { "sentence": "Word segmentation is the task of inserting or deleting word boundary characters in order to separate character sequences that correspond to words in some language.", "category": "background" }, { "sentence": "In this article we propose an approach based on a beam search algorithm and a language model working at the byte/character level, the latter component implemented either as an n‐gram model or a recurrent neural network.", "category": "method" }, { "sentence": "The resulting system analyzes the text input with no word boundaries one token at a time, which can be a character or a byte, and uses the information gathered by the language model to determine if a boundary must be placed in the current position or not.", "category": "method" }, { "sentence": "Our aim is to use this system in a preprocessing step for a microtext normalization system.", "category": "method" }, { "sentence": "This means that it needs to effectively cope with the data sparsity present on this kind of texts.", "category": "method" }, { "sentence": "We also strove to surpass the performance of two readily available word segmentation systems: The well‐known and accessible Word Breaker by Microsoft, and the Python module WordSegment by Grant Jenks.", "category": "method" }, { "sentence": "The results show that we have met our objectives, and we hope to continue to improve both the precision and the efficiency of our system in the future.", "category": "result" } ] }, { "paper_id": "56538469", "title": "Childhood and adolescent obesity: Multidisciplinary approaches in a clinical setting", "abstract": "BACKGROUND: The high prevalence of obesity in children and adolescents emphasizes the necessity to develop evidence-based treatment programs that are useful in a clinical setting. AIMS: The overall aim of the thesis was to develop and evaluate multidisciplinary approaches for management of children and adolescents with obesity. The focus was on generalizability of the treatment in a clinical setting as well as to analyze which factors might explain and influence the results. The development of treatment programs took into account clinical necessities such as the waiting list and available resources at the Childhood Obesity Unit in southern Sweden, a tertiary referral centre. The aim of Paper I was to assess the effects of low intensity solution-based single family therapy intervention on self-esteem and body mass index (BMI), BMI z-scores in obese pediatric subjects, and functioning in their families. The aim of the Paper II was to evaluate the efficacy of a Family Weight School – a one-year treatment model based on family therapy and brief solution-focused therapy in four group meetings with extremely and morbidly obese adolescents compared with waiting list controls. The aim of Paper III was to examine factors associated with self-esteem in a clinical sample of obese children. The findings could be used to improve family-based programs. The aim of Paper IV was to evaluate the effect on childhood obesity of a one-week sports camp followed by a six-month support program at local sports clubs. The aim of Paper V was to describe the implementation of the theory of family therapy in a clinical setting in order to provide tools for clinicians in the field of obesity who work with families, alone or in a multidisciplinary team. RESULTS: The single family therapy treatment (Paper I) resulted in a significant decrease in the degree of obesity of the child, as well as improvements in self-esteem and family functioning. These results were obtained after 3.8 sessions. Eighty-one percent of the children (44 out of 54) participated in the follow-up. The Family Weight School (Paper II) resulted in a significant decrease in the degree of obesity in adolescents with BMI z-scores of less than 3.5 (adult equivalent approximately BMI 40), but not in adolescents with BMI z-scores of more than 3.5, compared with a waiting list control group. Ninety percent of the intervention group (65 out of 72) completed the one-year-program. In Paper III we showed that self-esteem in a sample of severely obese children and adolescents referred for treatment is lower after the age of twelve, especially in girls. In Paper IV we have shown that one year after the camp the intervention group had significant decreases in BMI z-score. The control group had also reduced their BMI z-score. No differences were found in baseline values, follow-up values or changes in BMI z-score between groups, or between boys and girls. In Paper V we have described the key elements (approach, language, and process) of the family therapy-based program and tools that are helpful in treating children and adolescents with obesity in a clinical setting. CONCLUSIONS: Family therapy-based interventions may be useful in the treatment of childhood and adolescent obesity in a clinical setting. A future strategy might include the Family Weight School for those with BMI z-scores of less than 3.5 as a first step and the single family therapy for younger children and those who have BMI z-scores of more than 3.5. Treatment should start at a young age with special attention given to girls, since their self-esteem is particularly affected. Furthermore, the residential one-week sports camp combined with six-months of local club support has not proved to be an effective therapeutic intervention. (Less)", "classified_sentences": [ { "sentence": "The high prevalence of obesity in children and adolescents emphasizes the necessity to develop evidence-based treatment programs that are useful in a clinical setting.", "category": "background" }, { "sentence": "The overall aim of the thesis was to develop and evaluate multidisciplinary approaches for management of children and adolescents with obesity.", "category": "method" }, { "sentence": "The focus was on generalizability of the treatment in a clinical setting as well as to analyze which factors might explain and influence the results.", "category": "method" }, { "sentence": "The development of treatment programs took into account clinical necessities such as the waiting list and available resources at the Childhood Obesity Unit in southern Sweden, a tertiary referral centre.", "category": "method" }, { "sentence": "The aim of Paper I was to assess the effects of low intensity solution-based single family therapy intervention on self-esteem and body mass index (BMI), BMI z-scores in obese pediatric subjects, and functioning in their families.", "category": "method" }, { "sentence": "The aim of the Paper II was to evaluate the efficacy of a Family Weight School – a one-year treatment model based on family therapy and brief solution-focused therapy in four group meetings with extremely and morbidly obese adolescents compared with waiting list controls.", "category": "method" }, { "sentence": "The aim of Paper III was to examine factors associated with self-esteem in a clinical sample of obese children.", "category": "method" }, { "sentence": "The findings could be used to improve family-based programs.", "category": "result" }, { "sentence": "The aim of Paper IV was to evaluate the effect on childhood obesity of a one-week sports camp followed by a six-month support program at local sports clubs.", "category": "method" }, { "sentence": "The aim of Paper V was to describe the implementation of the theory of family therapy in a clinical setting in order to provide tools for clinicians in the field of obesity who work with families, alone or in a multidisciplinary team.", "category": "method" }, { "sentence": "The single family therapy treatment (Paper I) resulted in a significant decrease in the degree of obesity of the child, as well as improvements in self-esteem and family functioning.", "category": "result" }, { "sentence": "These results were obtained after 3.8 sessions.", "category": "result" }, { "sentence": "Eighty-one percent of the children (44 out of 54) participated in the follow-up.", "category": "result" }, { "sentence": "The Family Weight School (Paper II) resulted in a significant decrease in the degree of obesity in adolescents with BMI z-scores of less than 3.5 (adult equivalent approximately BMI 40), but not in adolescents with BMI z-scores of more than 3.5, compared with a waiting list control group.", "category": "result" }, { "sentence": "Ninety percent of the intervention group (65 out of 72) completed the one-year-program.", "category": "result" }, { "sentence": "In Paper III we showed that self-esteem in a sample of severely obese children and adolescents referred for treatment is lower after the age of twelve, especially in girls.", "category": "result" }, { "sentence": "In Paper IV we have shown that one year after the camp the intervention group had significant decreases in BMI z-score.", "category": "result" }, { "sentence": "The control group had also reduced their BMI z-score.", "category": "result" }, { "sentence": "No differences were found in baseline values, follow-up values or changes in BMI z-score between groups, or between boys and girls.", "category": "result" }, { "sentence": "In Paper V we have described the key elements (approach, language, and process) of the family therapy-based program and tools that are helpful in treating children and adolescents with obesity in a clinical setting.", "category": "result" }, { "sentence": "Family therapy-based interventions may be useful in the treatment of childhood and adolescent obesity in a clinical setting.", "category": "result" }, { "sentence": "A future strategy might include the Family Weight School for those with BMI z-scores of less than 3.5 as a first step and the single family therapy for younger children and those who have BMI z-scores of more than 3.5.", "category": "result" }, { "sentence": "Treatment should start at a young age with special attention given to girls, since their self-esteem is particularly affected.", "category": "result" }, { "sentence": "Furthermore, the residential one-week sports camp combined with six-months of local club support has not proved to be an effective therapeutic intervention.", "category": "result" } ] }, { "paper_id": "59230724", "title": "Detecting Serendipitous Drug Usage in Social Media with Deep Neural Network Models", "abstract": "Serendipitous drug usage refers to unexpected relief of comorbid diseases or symptoms when patients take a drug for another common or known indication. In the history of drug discovery, serendipity has contributed significantly to new and successful indications for many drugs. Our previous research has identified patient reported serendipitous drug usage in social media. If such information could be computationally identified in social media, it could be helpful for generating and validating drug-repositioning hypotheses. In this study, we framed detection of serendipitous drug usage in social media as a binary classification problem and investigated deep neural network models as a solution. We constructed word-embedding features from drug-review posts in the patient forum of WebMD, using the word2vec algorithm. We adopted the convolutional neural network (CNN), long short-term memory network (LSTM), and convolutional long short-term memory network (CLSTM) and redesigned them by adding contextual information that we extracted from drug-review posts, information filtering tools, medical ontology, and medical knowledge. We trained, tuned, and evaluated our deep neural network models on a gold standard dataset containing 15,714 sentences, of which 447 contained serendipitous drug usages. Additionally, we compared our deep neural networks to support vector machine, random forest, and AdaBoost.M1 algorithms. The results showed that adding context information helped to reduce the false-positive rate of deep neural network models. In the presence of an extremely imbalanced dataset and limited instances of serendipitous drug usage, deep neural network models did not outperform other machine learning models with n-gram and context features. However, deep neural network models could more effectively utilize word embedding in feature construction. This advantage made deep neural networks worthy of further investigation and improvement.", "classified_sentences": [ { "sentence": "Serendipitous drug usage refers to unexpected relief of comorbid diseases or symptoms when patients take a drug for another common or known indication.", "category": "background" }, { "sentence": "In the history of drug discovery, serendipity has contributed significantly to new and successful indications for many drugs.", "category": "background" }, { "sentence": "Our previous research has identified patient reported serendipitous drug usage in social media.", "category": "background" }, { "sentence": "If such information could be computationally identified in social media, it could be helpful for generating and validating drug-repositioning hypotheses.", "category": "background" }, { "sentence": "In this study, we framed detection of serendipitous drug usage in social media as a binary classification problem and investigated deep neural network models as a solution.", "category": "method" }, { "sentence": "We constructed word-embedding features from drug-review posts in the patient forum of WebMD, using the word2vec algorithm.", "category": "method" }, { "sentence": "We adopted the convolutional neural network (CNN), long short-term memory network (LSTM), and convolutional long short-term memory network (CLSTM) and redesigned them by adding contextual information that we extracted from drug-review posts, information filtering tools, medical ontology, and medical knowledge.", "category": "method" }, { "sentence": "We trained, tuned, and evaluated our deep neural network models on a gold standard dataset containing 15,714 sentences, of which 447 contained serendipitous drug usages.", "category": "method" }, { "sentence": "Additionally, we compared our deep neural networks to support vector machine, random forest, and AdaBoost.M1 algorithms.", "category": "method" }, { "sentence": "The results showed that adding context information helped to reduce the false-positive rate of deep neural network models.", "category": "result" }, { "sentence": "In the presence of an extremely imbalanced dataset and limited instances of serendipitous drug usage, deep neural network models did not outperform other machine learning models with n-gram and context features.", "category": "result" }, { "sentence": "However, deep neural network models could more effectively utilize word embedding in feature construction.", "category": "result" }, { "sentence": "This advantage made deep neural networks worthy of further investigation and improvement.", "category": "result" } ] }, { "paper_id": "195064593", "title": "Beyond Word2Vec: Embedding Words and Phrases in Same Vector Space", "abstract": "Word embeddings are being used for several linguistic problems and NLP tasks. Improvements in solutions to such problems are great because of the recent break-throughs in vector representation of words and research in vector space models. However, vector embeddings of phrases keeping semantics intact with words has been challenging. We propose a novel methodology using Siamese deep neural networks to embed multi-word units and fine-tune the current state-of-the-art word embed-dings keeping both in the same vector space. We show several semantic relations between words and phrases using the embeddings generated by our system and evaluate that the similarity of words and their corresponding paraphrases are maximized using the modified embeddings.", "classified_sentences": [ { "sentence": "Word embeddings are being used for several linguistic problems and NLP tasks.", "category": "background" }, { "sentence": "Improvements in solutions to such problems are great because of the recent break-throughs in vector representation of words and research in vector space models.", "category": "background" }, { "sentence": "However, vector embeddings of phrases keeping semantics intact with words has been challenging.", "category": "background" }, { "sentence": "We propose a novel methodology using Siamese deep neural networks to embed multi-word units and fine-tune the current state-of-the-art word embed-dings keeping both in the same vector space.", "category": "method" }, { "sentence": "We show several semantic relations between words and phrases using the embeddings generated by our system and evaluate that the similarity of words and their corresponding paraphrases are maximized using the modified embeddings.", "category": "result" } ] }, { "paper_id": "195745567", "title": "Loop mediated isothermal amplification : an innovative gene amplification technique for plant diseases", "abstract": "Nowadays, molecular diagnostic methods of plant pathogens evolved in a fast way, the use of rapid and easy to use detection method is fundamental to prevent pathogens cross border and spread to enhance food quality and security. Real – Time Loop-mediated isothermal amplification (RT-LAMP) is a novel molecular detection method that specifically detects genomic DNA by using a set of six oligonucleotide primers specific to different regions of a target gene and Bacillus stearothermophilus (Bst) DNA polymerase protein. This method has been recently modified to be use as a RT-LAMP and then widely applied in many fields for on-site detection and ability to be used in cross border control of plant health, such as quarantine disease diagnosis. The application of rapid and simple DNA extraction method (10 min at 65°C) shortened the detection assay to less than one hour. During the period 2015–2017 are tested in Albania more than 100 samples for: Xylella fastidiosa in olive trees, Flavescence doree in vineyards. Olive samples were taken in Vlora, Durrës, Saranda, etc. All the samples have been negative. Samples for the Flavescence doree have been taken in Lezhë, Vlorë, Durrës in Albania and Istria in Croatia. The samples from Albania resulted negative while the samples from Croatia (17 samples) were 53% positive. Every procedure has been confirmed with positive and negative controls to use diagnostic kit denominated Xylella Screen Glow & FD from Enbiotech s.r. l. , Italy. RT-LAMP method is more sensitive than convectional and RT PCR, advantage of LAMP is the isothermal reaction condition, hereby LAMP is affordable because of no need to have expensive thermal cycler. Although recommended reagent storage temperature is 20oC, reagents can be stored at environment temperature. Hereby there is no need to have cold chain for reagent distribution. The results indicate that the RT-LAMP assay is extremely rapid, cost-effective, highly sensitive and specific and has potential application in plant pathology surveillance. This new method is widely used nowadays by many laboratories and is recognized by EPPO as a standard method for some pathogens. Test performance study is being developed today in the circles of European scientists.", "classified_sentences": [ { "sentence": "Nowadays, molecular diagnostic methods of plant pathogens evolved in a fast way, the use of rapid and easy to use detection method is fundamental to prevent pathogens cross border and spread to enhance food quality and security.", "category": "background" }, { "sentence": "Real – Time Loop-mediated isothermal amplification (RT-LAMP) is a novel molecular detection method that specifically detects genomic DNA by using a set of six oligonucleotide primers specific to different regions of a target gene and Bacillus stearothermophilus (Bst) DNA polymerase protein.", "category": "method" }, { "sentence": "This method has been recently modified to be use as a RT-LAMP and then widely applied in many fields for on-site detection and ability to be used in cross border control of plant health, such as quarantine disease diagnosis.", "category": "method" }, { "sentence": "The application of rapid and simple DNA extraction method (10 min at 65°C) shortened the detection assay to less than one hour.", "category": "method" }, { "sentence": "During the period 2015–2017 are tested in Albania more than 100 samples for: Xylella fastidiosa in olive trees, Flavescence doree in vineyards.", "category": "result" }, { "sentence": "Olive samples were taken in Vlora, Durrës, Saranda, etc. All the samples have been negative.", "category": "result" }, { "sentence": "Samples for the Flavescence doree have been taken in Lezhë, Vlorë, Durrës in Albania and Istria in Croatia.", "category": "result" }, { "sentence": "The samples from Albania resulted negative while the samples from Croatia (17 samples) were 53% positive.", "category": "result" }, { "sentence": "Every procedure has been confirmed with positive and negative controls to use diagnostic kit denominated Xylella Screen Glow & FD from Enbiotech s.r.", "category": "method" }, { "sentence": ", Italy.", "category": "method" }, { "sentence": "RT-LAMP method is more sensitive than convectional and RT PCR, advantage of LAMP is the isothermal reaction condition, hereby LAMP is affordable because of no need to have expensive thermal cycler.", "category": "method" }, { "sentence": "Although recommended reagent storage temperature is 20oC, reagents can be stored at environment temperature.", "category": "method" }, { "sentence": "Hereby there is no need to have cold chain for reagent distribution.", "category": "method" }, { "sentence": "The results indicate that the RT-LAMP assay is extremely rapid, cost-effective, highly sensitive and specific and has potential application in plant pathology surveillance.", "category": "result" }, { "sentence": "This new method is widely used nowadays by many laboratories and is recognized by EPPO as a standard method for some pathogens.", "category": "result" }, { "sentence": "Test performance study is being developed today in the circles of European scientists.", "category": "background" } ] }, { "paper_id": "65358785", "title": "Coupled charge and spin dynamics in high-density ensembles of nitrogen-vacancy centers in diamond", "abstract": "We studied the spin depolarization of ensembles of nitrogen-vacancy (NV) centers in nitrogen-rich single crystal diamonds. We found a strong dependence of the evolution of the polarized state in the dark on the concentration of NV centers. At low excitation power, we observed a simple exponential decay profile in the low-density regime and a paradoxical inverted exponential profile in the high-density regime. At higher excitation power, we observed complex behavior, with an initial sharp rise in luminescence signal after the preparation pulse followed by a slower exponential decay. Magnetic field and excitation laser power-dependent measurements suggest that the rapid initial increase of the luminescence signal is related to recharging of the nitrogen-vacancy centers (from neutral to negatively charged) in the dark. The slow relaxing component corresponds to the longitudinal spin relaxation of the NV ensemble. The shape of the decay profile reflects the interplay between two mechanisms: the NV charge state conversion in the dark and the longitudinal spin relaxation. These mechanisms, in turn, are influenced by ionization, recharging and polarization dynamics during excitation. Interestingly, we found that charge dynamics are dominant in NV-dense samples even at very feeble excitation power. These observations may be important for the use of ensembles of NV centers in precession magnetometry and sensing applications.", "classified_sentences": [ { "sentence": "We studied the spin depolarization of ensembles of nitrogen-vacancy (NV) centers in nitrogen-rich single crystal diamonds.", "category": "background" }, { "sentence": "We found a strong dependence of the evolution of the polarized state in the dark on the concentration of NV centers.", "category": "result" }, { "sentence": "At low excitation power, we observed a simple exponential decay profile in the low-density regime and a paradoxical inverted exponential profile in the high-density regime.", "category": "result" }, { "sentence": "At higher excitation power, we observed complex behavior, with an initial sharp rise in luminescence signal after the preparation pulse followed by a slower exponential decay.", "category": "result" }, { "sentence": "Magnetic field and excitation laser power-dependent measurements suggest that the rapid initial increase of the luminescence signal is related to recharging of the nitrogen-vacancy centers (from neutral to negatively charged) in the dark.", "category": "result" }, { "sentence": "The slow relaxing component corresponds to the longitudinal spin relaxation of the NV ensemble.", "category": "result" }, { "sentence": "The shape of the decay profile reflects the interplay between two mechanisms: the NV charge state conversion in the dark and the longitudinal spin relaxation.", "category": "result" }, { "sentence": "These mechanisms, in turn, are influenced by ionization, recharging and polarization dynamics during excitation.", "category": "result" }, { "sentence": "Interestingly, we found that charge dynamics are dominant in NV-dense samples even at very feeble excitation power.", "category": "result" }, { "sentence": "These observations may be important for the use of ensembles of NV centers in precession magnetometry and sensing applications.", "category": "background" } ] }, { "paper_id": "67872619", "title": "Screen-Printed Piezoresistive Sensors for Monitoring Pressure Distribution in Wheelchair", "abstract": "Prolonged sitting inadequacies cause pressure ulcer to many individuals, especially to disadvantaged with reduced mobility. The measurement of distributed pressure and detection of irregular sitting postures is essential for preventing the risk of developing pressure ulcer. In this paper, a pressure sensing system capable of recognizing sitting postures by means of measuring interface pressure through printed pressure sensors is presented. A thin and flexible large area sensor is screen-printed using silver flake and carbon particle inks and comprises 16 sensing elements. For the evaluation of practical usability, the sensor characterization is carried out by conducting stability, repeatability, drift, and bending tests. The performance of the sensor is checked under varying environmental conditions. Sitting posture detection accuracy above 80 % is achieved using a classification algorithm for four different sitting postures. Pressure distribution is monitored at a scanning rate of 10 Hz. A low-power and small form factor of readout electronics enables a compact packaging inside the seat cushion. The presented sensor design targets smart wheelchairs, but it is extendable to much larger areas, for example, to be used in beds. The proposed sensing system would be of a great assistance for caregivers and health professionals.", "classified_sentences": [ { "sentence": "Prolonged sitting inadequacies cause pressure ulcer to many individuals, especially to disadvantaged with reduced mobility.", "category": "background" }, { "sentence": "The measurement of distributed pressure and detection of irregular sitting postures is essential for preventing the risk of developing pressure ulcer.", "category": "background" }, { "sentence": "In this paper, a pressure sensing system capable of recognizing sitting postures by means of measuring interface pressure through printed pressure sensors is presented.", "category": "method" }, { "sentence": "A thin and flexible large area sensor is screen-printed using silver flake and carbon particle inks and comprises 16 sensing elements.", "category": "method" }, { "sentence": "For the evaluation of practical usability, the sensor characterization is carried out by conducting stability, repeatability, drift, and bending tests.", "category": "method" }, { "sentence": "The performance of the sensor is checked under varying environmental conditions.", "category": "method" }, { "sentence": "Sitting posture detection accuracy above 80 % is achieved using a classification algorithm for four different sitting postures.", "category": "result" }, { "sentence": "Pressure distribution is monitored at a scanning rate of 10 Hz.", "category": "result" }, { "sentence": "A low-power and small form factor of readout electronics enables a compact packaging inside the seat cushion.", "category": "method" }, { "sentence": "The presented sensor design targets smart wheelchairs, but it is extendable to much larger areas, for example, to be used in beds.", "category": "method" }, { "sentence": "The proposed sensing system would be of a great assistance for caregivers and health professionals.", "category": "result" } ] }, { "paper_id": "203818879", "title": "Effect of Methanol Extract of Gomphrena celesioides on Chloroquine-Induced Hepatotoxicity and Oxidative Stress in Male Wistar Rats", "abstract": "Hepatic injury has been reported to be associated with chloroquine therapy. Gomphrena celesioides has been claimed to have pleiotropic protective properties in the liver by traditional herbal practitioner but there is no scientific evidence to this claim. This investigation therefore sought to evaluate the effect of Gomphrena celesioides administration on chloroquine-induced hepatic injury in rats. Forty adult male rats were divided into five groups of eight rats and were treated orally once daily. Rats in group one received 1 ml/kg body weight of 0.9% normal saline; rats in group two received 250 mg/kg body weight of chloroquine for three days; groups three, four and five rats were pre-treated with 200 mg/kg, 400 mg/kg and 800 mg/kg body weight of methanol extract of Gomphrena celesiodes for three days and on the fourth day were given 250 mg/kg body weight of chloroquine for three days. The experiment lasted for seven days. Liver injury was manifested biochemically by a significant increase in serum level or activities of hepatic markers (aminotransferases, alkaline phosphate, bilirubin, cholesterol and gamma glutamyl transferase). In addition, hepatic tissue from chloroquine-treated rats showed a significant increase in lipid peroxidation with a decrease in hepatic superoxide dismutase, catalase, glutathione peroxidase, glutathione-S-transferase and glutathione reservoirs. Moreover, the liver histopathologic evaluation revealed significant in chloroquine-treated rats. Gomphrena celesioides administration significantly alleviated chloroquine-induced pathologic changes in serum biochemistry and liver tissue. The results also suggest that Gomphrena celesioides possesses protective properties against chloroquine-induced liver injury via mitigation of drug-induced oxidative stress and its consequent events.", "classified_sentences": [ { "sentence": "Hepatic injury has been reported to be associated with chloroquine therapy.", "category": "background" }, { "sentence": "Gomphrena celesioides has been claimed to have pleiotropic protective properties in the liver by traditional herbal practitioner but there is no scientific evidence to this claim.", "category": "background" }, { "sentence": "This investigation therefore sought to evaluate the effect of Gomphrena celesioides administration on chloroquine-induced hepatic injury in rats.", "category": "method" }, { "sentence": "Forty adult male rats were divided into five groups of eight rats and were treated orally once daily.", "category": "method" }, { "sentence": "Rats in group one received 1 ml/kg body weight of 0.9% normal saline; rats in group two received 250 mg/kg body weight of chloroquine for three days; groups three, four and five rats were pre-treated with 200 mg/kg, 400 mg/kg and 800 mg/kg body weight of methanol extract of Gomphrena celesiodes for three days and on the fourth day were given 250 mg/kg body weight of chloroquine for three days.", "category": "method" }, { "sentence": "The experiment lasted for seven days.", "category": "method" }, { "sentence": "Liver injury was manifested biochemically by a significant increase in serum level or activities of hepatic markers (aminotransferases, alkaline phosphate, bilirubin, cholesterol and gamma glutamyl transferase).", "category": "result" }, { "sentence": "In addition, hepatic tissue from chloroquine-treated rats showed a significant increase in lipid peroxidation with a decrease in hepatic superoxide dismutase, catalase, glutathione peroxidase, glutathione-S-transferase and glutathione reservoirs.", "category": "result" }, { "sentence": "Moreover, the liver histopathologic evaluation revealed significant in chloroquine-treated rats.", "category": "result" }, { "sentence": "Gomphrena celesioides administration significantly alleviated chloroquine-induced pathologic changes in serum biochemistry and liver tissue.", "category": "result" }, { "sentence": "The results also suggest that Gomphrena celesioides possesses protective properties against chloroquine-induced liver injury via mitigation of drug-induced oxidative stress and its consequent events.", "category": "result" } ] }, { "paper_id": "69661863", "title": "A DGS pattern including DMS behavior for compact unit-cell designs", "abstract": "In this paper, a new (defected ground structure) DGS unit-cell is introduced and its performance is analyzed. The dumbbell-shaped DGS cell is modified by etching off slots in the ground plane to include (defected microstrip structure) DMS behavior without etching any defect in the microstrip line. The degree of freedom of cell design allows setting the bandgap center frequency while enabling a degree of miniaturization. A circuit-model is proposed to provide accurate cell responses. Simulations of the unit-cell with DGS and DMS behavior show moderate electromagnetic (EM) noise from the ground plane. Advantages are illustrated in a fabricated cell occupying approximately 40% less area.", "classified_sentences": [ { "sentence": "In this paper, a new (defected ground structure) DGS unit-cell is introduced and its performance is analyzed.", "category": "method" }, { "sentence": "The dumbbell-shaped DGS cell is modified by etching off slots in the ground plane to include (defected microstrip structure) DMS behavior without etching any defect in the microstrip line.", "category": "method" }, { "sentence": "The degree of freedom of cell design allows setting the bandgap center frequency while enabling a degree of miniaturization.", "category": "method" }, { "sentence": "A circuit-model is proposed to provide accurate cell responses.", "category": "method" }, { "sentence": "Simulations of the unit-cell with DGS and DMS behavior show moderate electromagnetic (EM) noise from the ground plane.", "category": "result" }, { "sentence": "Advantages are illustrated in a fabricated cell occupying approximately 40% less area.", "category": "result" } ] }, { "paper_id": "206187116", "title": "When I say … mastery learning", "abstract": "In 1968, as a college sophomore, I enrolled in a basic statistics course with 30 other students. The course was taught using mastery learning principles: it was delivered in 16 units sequenced by week; it provided practice and study opportunities in several forms; it used formative and summative testing; very high standards were expressed as minimum passing test scores; instructor feedback and coaching were provided, and its delivery allowed for variations in the time or opportunities needed to achieve the goal of each unit. After teaching and practice, we reported for unit testing at 10.00 am every Tuesday on a pass-the-test, see-you-next-week basis. Retests, as needed, were scheduled for Fridays at 5.00 pm (party time for US college students) or, as a last resort, at 7.00 am on Sunday mornings (doomsday option: never needed). All 30 students passed the course with an A grade, with no differences among us. We all felt great about this experience of success. The only downside was that the professor, Dr Jack Michael, was reprimanded by the dean for grading too leniently. ‘How can everyone be a high achiever? ’ groused the dean. ‘Someone must fail! ’", "classified_sentences": [ { "sentence": "In 1968, as a college sophomore, I enrolled in a basic statistics course with 30 other students.", "category": "background" }, { "sentence": "The course was taught using mastery learning principles: it was delivered in 16 units sequenced by week; it provided practice and study opportunities in several forms; it used formative and summative testing; very high standards were expressed as minimum passing test scores; instructor feedback and coaching were provided, and its delivery allowed for variations in the time or opportunities needed to achieve the goal of each unit.", "category": "method" }, { "sentence": "After teaching and practice, we reported for unit testing at 10.00 am every Tuesday on a pass-the-test, see-you-next-week basis.", "category": "method" }, { "sentence": "Retests, as needed, were scheduled for Fridays at 5.00 pm (party time for US college students) or, as a last resort, at 7.00 am on Sunday mornings (doomsday option: never needed).", "category": "method" }, { "sentence": "All 30 students passed the course with an A grade, with no differences among us.", "category": "result" }, { "sentence": "We all felt great about this experience of success.", "category": "result" }, { "sentence": "The only downside was that the professor, Dr Jack Michael, was reprimanded by the dean for grading too leniently.", "category": "result" }, { "sentence": "'How can everyone be a high achiever?", "category": "result" }, { "sentence": "' groused the dean.", "category": "result" }, { "sentence": "'Someone must fail!", "category": "result" } ] }, { "paper_id": "206592854", "title": "Deep hashing for compact binary codes learning", "abstract": "In this paper, we propose a new deep hashing (DH) approach to learn compact binary codes for large scale visual search. Unlike most existing binary codes learning methods which seek a single linear projection to map each sample into a binary vector, we develop a deep neural network to seek multiple hierarchical non-linear transformations to learn these binary codes, so that the nonlinear relationship of samples can be well exploited. Our model is learned under three constraints at the top layer of the deep network: 1) the loss between the original real-valued feature descriptor and the learned binary vector is minimized, 2) the binary codes distribute evenly on each bit, and 3) different bits are as independent as possible. To further improve the discriminative power of the learned binary codes, we extend DH into supervised DH (SDH) by including one discriminative term into the objective function of DH which simultaneously maximizes the inter-class variations and minimizes the intra-class variations of the learned binary codes. Experimental results show the superiority of the proposed approach over the state-of-the-arts.", "classified_sentences": [ { "sentence": "In this paper, we propose a new deep hashing (DH) approach to learn compact binary codes for large scale visual search.", "category": "method" }, { "sentence": "Unlike most existing binary codes learning methods which seek a single linear projection to map each sample into a binary vector, we develop a deep neural network to seek multiple hierarchical non-linear transformations to learn these binary codes, so that the nonlinear relationship of samples can be well exploited.", "category": "method" }, { "sentence": "Our model is learned under three constraints at the top layer of the deep network: 1) the loss between the original real-valued feature descriptor and the learned binary vector is minimized, 2) the binary codes distribute evenly on each bit, and 3) different bits are as independent as possible.", "category": "method" }, { "sentence": "To further improve the discriminative power of the learned binary codes, we extend DH into supervised DH (SDH) by including one discriminative term into the objective function of DH which simultaneously maximizes the inter-class variations and minimizes the intra-class variations of the learned binary codes.", "category": "method" }, { "sentence": "Experimental results show the superiority of the proposed approach over the state-of-the-arts.", "category": "result" } ] }, { "paper_id": "208172406", "title": "Visitors ’ impulse shopping behavior at Kuala Lumpur International Airport ( KLIA ) 2", "abstract": "Visitors’ impulse shopping behavior is widely happening throughout the daily routine without anyone realizing it. The objective of the paper is to examine the relationship between store environmental characteristics, namely, ambient, design, and social and the visitors’ impulsive shopping behavior. Apart from that, the mediating effect of visitors’ emotional responses on the relationship between store environmental characteristics and the visitors’ impulse shopping behavior was also examined. Findings indicated that store environmental characteristics have a significant relationship with the visitor’s impulse shopping behavior. The higher the store environmental characteristics effects such as ambient characteristics, design characteristics and social characteristics, the higher the rate of visitor’s impulse shopping behavior will be. It was also revealed that store environmental characteristics have a significant relationship with the visitors’ emotional responses. The better the store environmental characteristics effects such as ambient characteristics, design characteristics and social characteristics, the better the conditions of visitors’ emotional responses. Results also indicated that visitors’ emotional responses have a significant relationship with the visitor’s impulse shopping behavior. The better the visitors’ emotional responses, the higher the possibilities of visitor’s impulse shopping behavior will be. And finally, the finding revealed that the visitors’ emotional responses mediate the relationship between store environmental characteristics and the visitors’ impulse shopping behavior.", "classified_sentences": [ { "sentence": "Visitors’ impulse shopping behavior is widely happening throughout the daily routine without anyone realizing it.", "category": "background" }, { "sentence": "The objective of the paper is to examine the relationship between store environmental characteristics, namely, ambient, design, and social and the visitors’ impulsive shopping behavior.", "category": "method" }, { "sentence": "Apart from that, the mediating effect of visitors’ emotional responses on the relationship between store environmental characteristics and the visitors’ impulse shopping behavior was also examined.", "category": "method" }, { "sentence": "Findings indicated that store environmental characteristics have a significant relationship with the visitor’s impulse shopping behavior.", "category": "result" }, { "sentence": "The higher the store environmental characteristics effects such as ambient characteristics, design characteristics and social characteristics, the higher the rate of visitor’s impulse shopping behavior will be.", "category": "result" }, { "sentence": "It was also revealed that store environmental characteristics have a significant relationship with the visitors’ emotional responses.", "category": "result" }, { "sentence": "The better the store environmental characteristics effects such as ambient characteristics, design characteristics and social characteristics, the better the conditions of visitors’ emotional responses.", "category": "result" }, { "sentence": "Results also indicated that visitors’ emotional responses have a significant relationship with the visitor’s impulse shopping behavior.", "category": "result" }, { "sentence": "The better the visitors’ emotional responses, the higher the possibilities of visitor’s impulse shopping behavior will be.", "category": "result" }, { "sentence": "And finally, the finding revealed that the visitors’ emotional responses mediate the relationship between store environmental characteristics and the visitors’ impulse shopping behavior.", "category": "result" } ] }, { "paper_id": "212562512", "title": "MITOGENIC EFFECT OF THE SALIVARY GLAND EXTRACTS OF OCTOPUS", "abstract": "Objective: Octopuses from Mumbai waters have been less studied for their bioactive potential. The present study aimed to check the effect of the posterior salivary gland extracts of two species of octopus Cistopus indicus and Octopus fusiformis on cell proliferation. Methods: The posterior salivary glands were extracted in different solvents and partially purified by column chromatography. The samples were tested at different concentrations for their effect on the growing root meristems of Allium cepa and on chicken chorio-allantoic membrane. Results: The extracts increased the mitotic index (MI) with reference to the control system. The mitogenic effect of the extracts decreased with increasing recovery time. The salivary gland extracts of the octopuses, at lower concentrations, did not exhibit pro-angiogenic or anti-angiogenic effect on CAM vasculature. The acetic acid extracts, at a concentration of 0.75 mg/mL, has significantly promoted the proliferation of blood vessels. Conclusion: The study is indicative of the presence of mitotic promoting factor(s) in low concentrations in the posterior salivary glands. The effect is short-lived.", "classified_sentences": [ { "sentence": "Objective: Octopuses from Mumbai waters have been less studied for their bioactive potential.", "category": "background" }, { "sentence": "The present study aimed to check the effect of the posterior salivary gland extracts of two species of octopus Cistopus indicus and Octopus fusiformis on cell proliferation.", "category": "background" }, { "sentence": "Methods: The posterior salivary glands were extracted in different solvents and partially purified by column chromatography.", "category": "method" }, { "sentence": "The samples were tested at different concentrations for their effect on the growing root meristems of Allium cepa and on chicken chorio-allantoic membrane.", "category": "method" }, { "sentence": "Results: The extracts increased the mitotic index (MI) with reference to the control system.", "category": "result" }, { "sentence": "The mitogenic effect of the extracts decreased with increasing recovery time.", "category": "result" }, { "sentence": "The salivary gland extracts of the octopuses, at lower concentrations, did not exhibit pro-angiogenic or anti-angiogenic effect on CAM vasculature.", "category": "result" }, { "sentence": "The acetic acid extracts, at a concentration of 0.75 mg/mL, has significantly promoted the proliferation of blood vessels.", "category": "result" }, { "sentence": "Conclusion: The study is indicative of the presence of mitotic promoting factor(s) in low concentrations in the posterior salivary glands.", "category": "result" }, { "sentence": "The effect is short-lived.", "category": "result" } ] }, { "paper_id": "213188349", "title": "Review of the Numerical Modeling of Compression Molding of Sheet Molding Compound", "abstract": "A review of the numerical modeling of the compression molding of the sheet molding compound (SMC) is presented. The focus of this review is the practical difficulties of modeling cases with high fiber content, an area in which there is relatively little documented work. In these cases, the prediction of the flows become intricate due to several reasons, mainly the complex rheology of the compound and large temperature gradients, but also the orientation of fibers and the micromechanics of the interactions between the fluid and the fibers play major roles. The details of this during moldings are discussed. Special attention is given to the impact on viscosity from the high fiber volume fraction, and the various models for this. One additional area of interest is the modeling of the fiber orientation.", "classified_sentences": [ { "sentence": "A review of the numerical modeling of the compression molding of the sheet molding compound (SMC) is presented.", "category": "background" }, { "sentence": "The focus of this review is the practical difficulties of modeling cases with high fiber content, an area in which there is relatively little documented work.", "category": "background" }, { "sentence": "In these cases, the prediction of the flows become intricate due to several reasons, mainly the complex rheology of the compound and large temperature gradients, but also the orientation of fibers and the micromechanics of the interactions between the fluid and the fibers play major roles.", "category": "background" }, { "sentence": "The details of this during moldings are discussed.", "category": "method" }, { "sentence": "Special attention is given to the impact on viscosity from the high fiber volume fraction, and the various models for this.", "category": "method" }, { "sentence": "One additional area of interest is the modeling of the fiber orientation.", "category": "method" } ] }, { "paper_id": "216054721", "title": "Alcohol Screening for Women in their Childbearing Years: What are Health Care Providers doing in Canada?", "abstract": "Health care providers (HCPs) have an important role in screening for alcohol use across the lifespan, particularly during the childbearing years, and providing brief intervention to yield optimal outcomes and prevent the potential teratogenic effects of prenatal alcohol exposure. Objective The purpose of this study was to describe the current alcohol screening practices of Canadian HCPs who care for pregnant women and women of childbearing age. Methods An online survey was administered in 2017 with the aim to identify current knowledge, attitudes, practices and beliefs among Canadian HCPs on screening, brief intervention and referral to treatment (SBIRT) for alcohol use for this population. The bilingual survey was disseminated by 4 national professional associations. A total of 634 interprofessional clinicians completed to the survey. Descriptive analysis was completed for the respondent's profession and their practices related to alcohol SBIRT. Cross-tabulation analyses explored the use of different screening questionnaires. Results Most respondents reported asking about alcohol use; however, there was a low overall use of screening questionnaires for both women of childbearing age and those who are pregnant. Low screening rates may equate to missed opportunities for intervention. Low rates of brief intervention and referral were noted even in circumstances where at-risk drinking was identified, with only 16.4% of respondents intervening when pregnant women reported at-risk alcohol consumption. Conclusion Continued efforts are needed to improve alcohol screening practices among women's HCPs across Canada. Priority areas for training include: understanding validated alcohol screening questionnaires; incorporating brief intervention into routine care; and developing local referral pathways.", "classified_sentences": [ { "sentence": "Health care providers (HCPs) have an important role in screening for alcohol use across the lifespan, particularly during the childbearing years, and providing brief intervention to yield optimal outcomes and prevent the potential teratogenic effects of prenatal alcohol exposure.", "category": "background" }, { "sentence": "Objective The purpose of this study was to describe the current alcohol screening practices of Canadian HCPs who care for pregnant women and women of childbearing age.", "category": "background" }, { "sentence": "Methods An online survey was administered in 2017 with the aim to identify current knowledge, attitudes, practices and beliefs among Canadian HCPs on screening, brief intervention and referral to treatment (SBIRT) for alcohol use for this population.", "category": "method" }, { "sentence": "The bilingual survey was disseminated by 4 national professional associations.", "category": "method" }, { "sentence": "A total of 634 interprofessional clinicians completed to the survey.", "category": "method" }, { "sentence": "Descriptive analysis was completed for the respondent's profession and their practices related to alcohol SBIRT.", "category": "method" }, { "sentence": "Cross-tabulation analyses explored the use of different screening questionnaires.", "category": "method" }, { "sentence": "Results Most respondents reported asking about alcohol use; however, there was a low overall use of screening questionnaires for both women of childbearing age and those who are pregnant.", "category": "result" }, { "sentence": "Low screening rates may equate to missed opportunities for intervention.", "category": "result" }, { "sentence": "Low rates of brief intervention and referral were noted even in circumstances where at-risk drinking was identified, with only 16.4% of respondents intervening when pregnant women reported at-risk alcohol consumption.", "category": "result" }, { "sentence": "Conclusion Continued efforts are needed to improve alcohol screening practices among women's HCPs across Canada.", "category": "result" }, { "sentence": "Priority areas for training include: understanding validated alcohol screening questionnaires; incorporating brief intervention into routine care; and developing local referral pathways.", "category": "result" } ] }, { "paper_id": "219938042", "title": "Une note sur l’analyse du constituant pour le français (A Note on constituent parsing for French)", "abstract": "Cet article traite des analyses d’erreurs quantitatives et qualitatives sur les résultats de l’analyse syntaxique des constituants pour le français. Pour cela, nous étendons l’approche de Kummerfeld et al. (2012) pour français, et nous présentons les détails de l’analyse. Nous entraînons les systèmes d’analyse syntaxique statistiques et neuraux avec le corpus arboré pour français, et nous évaluons les résultats d’analyse. Le corpus arboré pour le français fournit des étiquettes syntagmatiques à grain fin, et les caractéristiques grammaticales du corpus affectent des erreurs d’analyse syntaxique.", "classified_sentences": [ { "sentence": "Cet article traite des analyses d’erreurs quantitatives et qualitatives sur les résultats de l’analyse syntaxique des constituants pour le français.", "category": "background" }, { "sentence": "Pour cela, nous étendons l’approche de Kummerfeld et al. (2012) pour français, et nous présentons les détails de l’analyse.", "category": "method" }, { "sentence": "Nous entraînons les systèmes d’analyse syntaxique statistiques et neuraux avec le corpus arboré pour français, et nous évaluons les résultats d’analyse.", "category": "method" }, { "sentence": "Le corpus arboré pour le français fournit des étiquettes syntagmatiques à grain fin, et les caractéristiques grammaticales du corpus affectent des erreurs d’analyse syntaxique.", "category": "result" } ] }, { "paper_id": "86548959", "title": "Evaluation of Rootstock Resistance to Fusarium Wilt and Gummy Stem Blight and Effect on Yield and Quality of a Grafted 'Inodorus' Melon", "abstract": "Grafting represents an effective tool for controlling the race 1,2 of Fusarium oxysporum f. sp. melonis (FOM) and Didymella bryoniae in melon (Cucumis melo L.). Although not considered a soilborne pathogen, D. bryoniae survives on plant remains in the soil. The lack of effective resistant commercial hybrids and the gradual reduced use of soil fumigation with methyl bromide increase the risk of damages by both these pathogens. We determined the effectiveness of eight commercial rootstocks, 'RS 841', 'P 360', 'ES 99-13', 'Elsi' (Cucurbita maxima Duchesne · Cucurbita moschata Duchesne), and 'Belimo', 'Energia', 'Griffin', 'ES liscio' (Cucumis melo genotypes), for their resistance to FOM and D. bryoniae. During 2003 and 2004 growing seasons, the inodorus F1 hybrid Incas was grafted onto each of these commercial rootstocks and then evaluated, under greenhouse conditions, in terms of productivity and fruit quality. Cucurbita rootstocks ('RS 841', 'P 360', 'ES 99-13', 'Elsi') were highly resistant both to the race 1,2 of FOM (100% survival) and to D. bryoniae (almost absent crown lesions and low leaf disease index); this reaction clearly differed from that of both the C. melo rootstocks ('Belimo', 'Energia', 'Griffin', 'ES liscio') and the control Incas. In both years, the highest yield was recorded in the graft combination Incas/'RS 841' with 5.6 and 8.1 kgm -2 during 2003 and 2004, respectively. The Cucurbita rootstock 'RS 841' produced yields higher than C. melo rootstocks ('Belimo', 'Energia', 'Griffin', 'ES liscio') and the control Incas. Fruit dry matter, titratable acidity, total soluble solid contents, fruit firmness, and Hunter color (L* (brightness), a* (redness), and b* (yellowness) param- eters) of grafted melons were similar to those of the plants grown on their own roots.", "classified_sentences": [ { "sentence": "Grafting represents an effective tool for controlling the race 1,2 of Fusarium oxysporum f.sp.melonis (FOM) and Didymella bryoniae in melon (Cucumis melo L.).", "category": "background" }, { "sentence": "sp.", "category": "background" }, { "sentence": "melonis (FOM) and Didymella bryoniae in melon (Cucumis melo L.).", "category": "background" }, { "sentence": "Although not considered a soilborne pathogen, D.bryoniae survives on plant remains in the soil.", "category": "background" }, { "sentence": "The lack of effective resistant commercial hybrids and the gradual reduced use of soil fumigation with methyl bromide increase the risk of damages by both these pathogens.", "category": "background" }, { "sentence": "We determined the effectiveness of eight commercial rootstocks, 'RS 841', 'P 360', 'ES 99-13', 'Elsi' (Cucurbita maxima Duchesne · Cucurbita moschata Duchesne), and 'Belimo', 'Energia', 'Griffin', 'ES liscio' (Cucumis melo genotypes), for their resistance to FOM and D.bryoniae.", "category": "method" }, { "sentence": "During 2003 and 2004 growing seasons, the inodorus F1 hybrid Incas was grafted onto each of these commercial rootstocks and then evaluated, under greenhouse conditions, in terms of productivity and fruit quality.", "category": "method" }, { "sentence": "Cucurbita rootstocks ('RS 841', 'P 360', 'ES 99-13', 'Elsi') were highly resistant both to the race 1,2 of FOM (100% survival) and to D.bryoniae (almost absent crown lesions and low leaf disease index); this reaction clearly differed from that of both the C.melo rootstocks ('Belimo', 'Energia', 'Griffin', 'ES liscio') and the control Incas.", "category": "result" }, { "sentence": "In both years, the highest yield was recorded in the graft combination Incas/'RS 841' with 5.6 and 8.1 kgm -2 during 2003 and 2004, respectively.", "category": "result" }, { "sentence": "The Cucurbita rootstock 'RS 841' produced yields higher than C.melo rootstocks ('Belimo', 'Energia', 'Griffin', 'ES liscio') and the control Incas.", "category": "result" }, { "sentence": "Fruit dry matter, titratable acidity, total soluble solid contents, fruit firmness, and Hunter color (L* (brightness), a* (redness), and b* (yellowness) param- eters) of grafted melons were similar to those of the plants grown on their own roots.", "category": "result" } ] }, { "paper_id": "87693043", "title": "Influences of Environmental Variability, Genetics and Plant Size on Variation in Sexual and Clonal Reproduction and Allocation of Resources in Three Wetland Plant Species", "abstract": "Optimal Partitioning Theory (OPT) states organisms will give more resources to structures and functions that enhance fitness. OPT can be applied to reproduction in clonal plants, which allocate resources between two modes of reproduction—sexual through fruits and clonal through spacers and ramets. In nutrient rich environments, clonal growth allows offspring to stay in beneficial surroundings, while in nutrient poor conditions, sexual reproduction can allow escape and generation of new, potentially more fit offspring. I tested this hypothesis by comparing clonal and sexual reproductive allocation in Penthorum sedoides under differing nutrient levels over two generations. Genotypic and environmental influences on reproductive variation in Lythrum salicaria and Penthorum sedoides were separated by comparing clones within and between treatments. Allocation to fruits was higher in the control than the fertilized group, but only in the second year, providing partial support to an increase in sexual allocation in lower resource conditions. Allocation to spacer mass and ramet mass increased under high nutrients, while number of ramets did not, also providing limited support to the predictions of OPT. Genotype had little effect on sexual and clonal variation. Variation due to fertilizer was more influential, demonstrating plasticity in reproductive expression.", "classified_sentences": [ { "sentence": "Optimal Partitioning Theory (OPT) states organisms will give more resources to structures and functions that enhance fitness.", "category": "background" }, { "sentence": "OPT can be applied to reproduction in clonal plants, which allocate resources between two modes of reproduction—sexual through fruits and clonal through spacers and ramets.", "category": "background" }, { "sentence": "In nutrient rich environments, clonal growth allows offspring to stay in beneficial surroundings, while in nutrient poor conditions, sexual reproduction can allow escape and generation of new, potentially more fit offspring.", "category": "background" }, { "sentence": "I tested this hypothesis by comparing clonal and sexual reproductive allocation in Penthorum sedoides under differing nutrient levels over two generations.", "category": "method" }, { "sentence": "Genotypic and environmental influences on reproductive variation in Lythrum salicaria and Penthorum sedoides were separated by comparing clones within and between treatments.", "category": "method" }, { "sentence": "Allocation to fruits was higher in the control than the fertilized group, but only in the second year, providing partial support to an increase in sexual allocation in lower resource conditions.", "category": "result" }, { "sentence": "Allocation to spacer mass and ramet mass increased under high nutrients, while number of ramets did not, also providing limited support to the predictions of OPT.", "category": "result" }, { "sentence": "Genotype had little effect on sexual and clonal variation.", "category": "result" }, { "sentence": "Variation due to fertilizer was more influential, demonstrating plasticity in reproductive expression.", "category": "result" } ] }, { "paper_id": "221936134", "title": "Study of the spatial and temporal variability of rainfall in the Middle and Lower Cheliff (Algeria)", "abstract": "Abstract. The purpose of this study is to identify homogeneous rainfall regions and to study the spatial and temporal variability of rainfall in the Cheliff basin using the regional vector method and the statistical approach (Pettitt test, Lee Heghinian test and Hubert segmentation) and the geostatistic approach (inverse distance weighting method). In terms of results, the regional vector method highlighted six (6) homogeneous rainfall regions. The downward trend occurred in the study area in 1972, affecting a few coastal stations. In 1976, this decline extended to the South West and throughout the coastal region. In 1980, the drop covered the entire basin. This decline has resulted in an estimated deficit of 30 % on average in the eastern region, the coastal region and the Mina. However, the central part of the basin experienced a 20 % decrease compared to the period before the break (1968–1980). The same spatial irregularity in rainfall was observed during the pre-break and post-break periods (1981–2010). On the other hand, throughout the basin, the areas corresponding to the rainfall ranges identified during the 1968–1980 period experienced an average decrease of 100 mm during the post-break period, except in the eastern region, where the decrease exceeded 200 mm.", "classified_sentences": [ { "sentence": "Abstract.", "category": "background" }, { "sentence": "The purpose of this study is to identify homogeneous rainfall regions and to study the spatial and temporal variability of rainfall in the Cheliff basin using the regional vector method and the statistical approach (Pettitt test, Lee Heghinian test and Hubert segmentation) and the geostatistic approach (inverse distance weighting method).", "category": "method" }, { "sentence": "In terms of results, the regional vector method highlighted six (6) homogeneous rainfall regions.", "category": "result" }, { "sentence": "The downward trend occurred in the study area in 1972, affecting a few coastal stations.", "category": "result" }, { "sentence": "In 1976, this decline extended to the South West and throughout the coastal region.", "category": "result" }, { "sentence": "In 1980, the drop covered the entire basin.", "category": "result" }, { "sentence": "This decline has resulted in an estimated deficit of 30 % on average in the eastern region, the coastal region and the Mina.", "category": "result" }, { "sentence": "However, the central part of the basin experienced a 20 % decrease compared to the period before the break (1968–1980).", "category": "result" }, { "sentence": "The same spatial irregularity in rainfall was observed during the pre-break and post-break periods (1981–2010).", "category": "result" }, { "sentence": "On the other hand, throughout the basin, the areas corresponding to the rainfall ranges identified during the 1968–1980 period experienced an average decrease of 100 mm during the post-break period, except in the eastern region, where the decrease exceeded 200 mm.", "category": "result" } ] }, { "paper_id": "88512931", "title": "Comparison of non-homogeneous regression models for probabilistic wind speed forecasting", "abstract": "In weather forecasting, non-homogeneous regression (NR) is used to statistically post-process forecast ensembles in order to obtain calibrated predictive distributions. For wind speed forecasts, the regression model is given by a truncated normal (TN) distribution, where location and spread derive from the ensemble. This article proposes two alternative approaches which utilise the generalised extreme value (GEV) distribution. A direct alternative to the TN regression is to apply a predictive distribution from the GEV family, while a regime-switching approach based on the median of the forecast ensemble incorporates both distributions. In a case study on daily maximum wind speed over Germany with the forecast ensemble from the European Centre for Medium-Range Weather Forecasts (ECMWF), all three approaches significantly improve the calibration as well as the overall skill of the raw ensemble with the regime-switching approach showing the highest skill in the upper tail.", "classified_sentences": [ { "sentence": "In weather forecasting, non-homogeneous regression (NR) is used to statistically post-process forecast ensembles in order to obtain calibrated predictive distributions.", "category": "background" }, { "sentence": "For wind speed forecasts, the regression model is given by a truncated normal (TN) distribution, where location and spread derive from the ensemble.", "category": "method" }, { "sentence": "This article proposes two alternative approaches which utilise the generalised extreme value (GEV) distribution.", "category": "method" }, { "sentence": "A direct alternative to the TN regression is to apply a predictive distribution from the GEV family, while a regime-switching approach based on the median of the forecast ensemble incorporates both distributions.", "category": "method" }, { "sentence": "In a case study on daily maximum wind speed over Germany with the forecast ensemble from the European Centre for Medium-Range Weather Forecasts (ECMWF), all three approaches significantly improve the calibration as well as the overall skill of the raw ensemble with the regime-switching approach showing the highest skill in the upper tail.", "category": "result" } ] }, { "paper_id": "226281977", "title": "An HVS-Oriented Saliency Map Prediction Modeling", "abstract": "Visual attention is one of the most significant characteristics for selecting and understanding the outside world. The nature complex scenes, including larger redundancy and human vision, can't be processing all information simultaneously because of the information bottleneck. The visual system mainly focuses on dominant parts of the scenes to reduce the input visual redundancy information. It's commonly known as visual attention prediction or visual saliency map. This paper proposes a new saliency prediction architecture inspired by human low-level visual cortex function. The model considered the opponent color channel, wavelet energy map, and contrast sensitivity function for extract image features and maximum approach to real visual neural network function in the brain. The proposed model is evaluated several datasets, including MIT1003, MIT300, TORONTO, and SID4VAM to explain its efficiency. The proposed model results are quantitatively and qualitatively compared to other state-of-the-art salience prediction models and their achieved out-performing of visual saliency prediction.", "classified_sentences": [ { "sentence": "Visual attention is one of the most significant characteristics for selecting and understanding the outside world.", "category": "background" }, { "sentence": "The nature complex scenes, including larger redundancy and human vision, can't be processing all information simultaneously because of the information bottleneck.", "category": "background" }, { "sentence": "The visual system mainly focuses on dominant parts of the scenes to reduce the input visual redundancy information.", "category": "background" }, { "sentence": "It's commonly known as visual attention prediction or visual saliency map.", "category": "background" }, { "sentence": "This paper proposes a new saliency prediction architecture inspired by human low-level visual cortex function.", "category": "method" }, { "sentence": "The model considered the opponent color channel, wavelet energy map, and contrast sensitivity function for extract image features and maximum approach to real visual neural network function in the brain.", "category": "method" }, { "sentence": "The proposed model is evaluated several datasets, including MIT1003, MIT300, TORONTO, and SID4VAM to explain its efficiency.", "category": "result" }, { "sentence": "The proposed model results are quantitatively and qualitatively compared to other state-of-the-art salience prediction models and their achieved out-performing of visual saliency prediction.", "category": "result" } ] }, { "paper_id": "231663640", "title": "Curcumin Mitigates Hypertension, Endothelial Dysfunction and Oxidative Stress in Rats with Chronic Exposure to Lead and Cadmium.", "abstract": "Lead (Pb) and cadmium (Cd) are environmental pollutants and nonessential elements in the body. Both metals induce the development of hypertension which is associated with oxidative stress. Curcumin (CUR) is a polyphenolic compound with strong antioxidant activity. The present study evaluated the effect of CUR on oxidative stress, alteration of vascular responsiveness and hypertension induced by exposure to either Pb, Cd or the combination of Pb and Cd. Male Sprague-Dawley rats were exposed to low level of lead acetate (100 mg/L) and/or cadmium chloride (10 mg/L) in the drinking water for 16 weeks. The control animals received deionized water as drinking water. CUR (100 mg/kg) or propylene glycol as vehicle was intragastrically administered once daily for the last 4 weeks. Exposure to Pb, Cd or the combination induced increases in blood pressure and peripheral vascular resistance, and decreased the blood pressure response to intravenous infusion to acetylcholine. Supplementation with CUR significantly reduced blood pressure, alleviated oxidative stress, and increased plasma nitrate/nitrite and glutathione in the blood. The effects of CUR were associated with the improvement of vascular responsiveness, upregulation of the endothelial nitric oxide synthase and downregulation of the NADPH oxidase expression. Furthermore, CUR reduced the metal levels in blood, aorta, liver and kidney. Altogether, exposure to the combination of Pb and Cd aggravated hypertension and oxidative stress, and CUR effectively ameliorated these adverse events in metal exposed animals. Data indicate that CUR may be useful as a dietary supplement for protection against the noxious effects of the heavy metals.", "classified_sentences": [ { "sentence": "Lead (Pb) and cadmium (Cd) are environmental pollutants and nonessential elements in the body.", "category": "background" }, { "sentence": "Both metals induce the development of hypertension which is associated with oxidative stress.", "category": "background" }, { "sentence": "Curcumin (CUR) is a polyphenolic compound with strong antioxidant activity.", "category": "background" }, { "sentence": "The present study evaluated the effect of CUR on oxidative stress, alteration of vascular responsiveness and hypertension induced by exposure to either Pb, Cd or the combination of Pb and Cd.", "category": "method" }, { "sentence": "Male Sprague-Dawley rats were exposed to low level of lead acetate (100 mg/L) and/or cadmium chloride (10 mg/L) in the drinking water for 16 weeks.", "category": "method" }, { "sentence": "The control animals received deionized water as drinking water.", "category": "method" }, { "sentence": "CUR (100 mg/kg) or propylene glycol as vehicle was intragastrically administered once daily for the last 4 weeks.", "category": "method" }, { "sentence": "Exposure to Pb, Cd or the combination induced increases in blood pressure and peripheral vascular resistance, and decreased the blood pressure response to intravenous infusion to acetylcholine.", "category": "result" }, { "sentence": "Supplementation with CUR significantly reduced blood pressure, alleviated oxidative stress, and increased plasma nitrate/nitrite and glutathione in the blood.", "category": "result" }, { "sentence": "The effects of CUR were associated with the improvement of vascular responsiveness, upregulation of the endothelial nitric oxide synthase and downregulation of the NADPH oxidase expression.", "category": "result" }, { "sentence": "Furthermore, CUR reduced the metal levels in blood, aorta, liver and kidney.", "category": "result" }, { "sentence": "Altogether, exposure to the combination of Pb and Cd aggravated hypertension and oxidative stress, and CUR effectively ameliorated these adverse events in metal exposed animals.", "category": "result" }, { "sentence": "Data indicate that CUR may be useful as a dietary supplement for protection against the noxious effects of the heavy metals.", "category": "result" } ] }, { "paper_id": "231721722", "title": "Birth Defects and Long-Term Neurodevelopmental Abnormalities in Infants Born During the Zika Virus Epidemic in the Dominican Republic", "abstract": "Background: When acquired during pregnancy, Zika virus (ZIKV) infection can cause substantial fetal morbidity, however, little is known about the long-term neurodevelopmental abnormalities of infants with congenital ZIKV exposure without microcephaly at birth. Methods: We conducted a cross sectional study to characterize infants born with microcephaly, and a retrospective cohort study of infants who appeared well at birth, but had possible congenital ZIKV exposure. We analyzed data from the Dominican Ministry of Health’s (MoH) National System of Epidemiological Surveillance. Neurodevelopmental abnormalities were assessed by pediatric neurologists over an 18-month period using Denver Developmental Screening Test II. Results: Of 800 known live births from 1,364 women with suspected or confirmed ZIKV infection during pregnancy, 87 (11%) infants had confirmed microcephaly. Mean head circumference (HC) at birth was 28.1 cm (SD ± 2.1 cm) and 41% had a HC on the zero percentile for gestational age. Of 42 infants with possible congenital ZIKV exposure followed longitudinally, 52% had neurodevelopmental abnormalities, including two cases of postnatal onset microcephaly, during follow-up. Most abnormalities resolved, though two infants (4%) had neurodevelopmental abnormalities that were likely associated with ZIKV infection and persisted through 15–18 months. Conclusions: In the DR epidemic, 11% of infants born to women reported to the MoH with suspected or confirmed ZIKV during pregnancy had microcephaly. Some 4% of ZKV-exposed infants developed postnatal neurocognitive abnormalities. Monitoring of the cohort through late childhood and adolescence is needed.", "classified_sentences": [ { "sentence": "Background: When acquired during pregnancy, Zika virus (ZIKV) infection can cause substantial fetal morbidity, however, little is known about the long-term neurodevelopmental abnormalities of infants with congenital ZIKV exposure without microcephaly at birth.", "category": "background" }, { "sentence": "Methods: We conducted a cross sectional study to characterize infants born with microcephaly, and a retrospective cohort study of infants who appeared well at birth, but had possible congenital ZIKV exposure.", "category": "method" }, { "sentence": "We analyzed data from the Dominican Ministry of Health’s (MoH) National System of Epidemiological Surveillance.", "category": "method" }, { "sentence": "Neurodevelopmental abnormalities were assessed by pediatric neurologists over an 18-month period using Denver Developmental Screening Test II.", "category": "method" }, { "sentence": "Results: Of 800 known live births from 1,364 women with suspected or confirmed ZIKV infection during pregnancy, 87 (11%) infants had confirmed microcephaly.", "category": "result" }, { "sentence": "Mean head circumference (HC) at birth was 28.1 cm (SD ± 2.1 cm) and 41% had a HC on the zero percentile for gestational age.", "category": "result" }, { "sentence": "Of 42 infants with possible congenital ZIKV exposure followed longitudinally, 52% had neurodevelopmental abnormalities, including two cases of postnatal onset microcephaly, during follow-up.", "category": "result" }, { "sentence": "Most abnormalities resolved, though two infants (4%) had neurodevelopmental abnormalities that were likely associated with ZIKV infection and persisted through 15–18 months.", "category": "result" }, { "sentence": "Conclusions: In the DR epidemic, 11% of infants born to women reported to the MoH with suspected or confirmed ZIKV during pregnancy had microcephaly.", "category": "result" }, { "sentence": "Some 4% of ZKV-exposed infants developed postnatal neurocognitive abnormalities.", "category": "result" }, { "sentence": "Monitoring of the cohort through late childhood and adolescence is needed.", "category": "result" } ] }, { "paper_id": "231786409", "title": "Introduction to Neural Transfer Learning with Transformers for Social Science Text Analysis", "abstract": "Transformer-based models for transfer learning have the potential to achieve high prediction accuracies on text-based supervised learning tasks with relatively few training data instances. These models are thus likely to benefit social scientists that seek to have as accurate as possible text-based measures but only have limited resources for annotating training data. To enable social scientists to leverage these potential benefits for their research, this paper explains how these methods work, why they might be advantageous, and what their limitations are. Additionally, three Transformer-based models for transfer learning, BERT (Devlin et al. 2019), RoBERTa (Liu et al. 2019), and the Longformer (Beltagy et al. 2020), are compared to conventional machine learning algorithms on three applications. Across all evaluated tasks, textual styles, and training data set sizes, the conventional models are consistently outperformed by transfer learning with Transformers, thereby demonstrating the benefits these models can bring to textbased social science research.", "classified_sentences": [ { "sentence": "Transformer-based models for transfer learning have the potential to achieve high prediction accuracies on text-based supervised learning tasks with relatively few training data instances.", "category": "background" }, { "sentence": "These models are thus likely to benefit social scientists that seek to have as accurate as possible text-based measures but only have limited resources for annotating training data.", "category": "background" }, { "sentence": "To enable social scientists to leverage these potential benefits for their research, this paper explains how these methods work, why they might be advantageous, and what their limitations are.", "category": "method" }, { "sentence": "Additionally, three Transformer-based models for transfer learning, BERT (Devlin et al. 2019), RoBERTa (Liu et al. 2019), and the Longformer (Beltagy et al. 2020), are compared to conventional machine learning algorithms on three applications.", "category": "method" }, { "sentence": "Across all evaluated tasks, textual styles, and training data set sizes, the conventional models are consistently outperformed by transfer learning with Transformers, thereby demonstrating the benefits these models can bring to textbased social science research.", "category": "result" } ] }, { "paper_id": "233231184", "title": "Mutual Information Preserving Back-propagation: Learn to Invert for Faithful Attribution", "abstract": "Back-propagation based visualizations have been proposed to interpret deep neural networks (DNNs), some of which produce interpretations with good visual quality. However, there exist doubts about whether these intuitive visualizations are related to network decisions. Recent studies have confirmed this suspicion by verifying that almost all these modified back-propagation visualizations are not faithful to the model's decision-making process. Besides, these visualizations produce vague \"relative importance scores\", among which low values can't guarantee to be independent of the final prediction. Hence, it's highly desirable to develop a novel back-propagation method that guarantees theoretical faithfulness and produces a quantitative attribution score with a clear understanding. To achieve the goal, we resort to mutual information theory to generate the interpretations, studying how much information of output is encoded in each input neuron. The basic idea is to learn a source signal by back-propagation such that the mutual information between input and output should be as much as possible preserved in the mutual information between input and the source signal. In addition, we propose a Mutual Information Preserving Inverse Network, termed MIP-IN, in which the parameters of each layer are recursively trained to learn how to invert. During the inversion, forward relu operation is adopted to adapt the general interpretations to the specific input. We then empirically demonstrate that the inverted source signal satisfies completeness and minimality property, which are crucial for a faithful interpretation. Furthermore, the empirical study validates the effectiveness of interpretations generated by MIP-IN.", "classified_sentences": [ { "sentence": "Back-propagation based visualizations have been proposed to interpret deep neural networks (DNNs), some of which produce interpretations with good visual quality.", "category": "background" }, { "sentence": "However, there exist doubts about whether these intuitive visualizations are related to network decisions.", "category": "background" }, { "sentence": "Recent studies have confirmed this suspicion by verifying that almost all these modified back-propagation visualizations are not faithful to the model's decision-making process.", "category": "background" }, { "sentence": "Besides, these visualizations produce vague \"relative importance scores\", among which low values can't guarantee to be independent of the final prediction.", "category": "background" }, { "sentence": "Hence, it's highly desirable to develop a novel back-propagation method that guarantees theoretical faithfulness and produces a quantitative attribution score with a clear understanding.", "category": "background" }, { "sentence": "To achieve the goal, we resort to mutual information theory to generate the interpretations, studying how much information of output is encoded in each input neuron.", "category": "method" }, { "sentence": "The basic idea is to learn a source signal by back-propagation such that the mutual information between input and output should be as much as possible preserved in the mutual information between input and the source signal.", "category": "method" }, { "sentence": "In addition, we propose a Mutual Information Preserving Inverse Network, termed MIP-IN, in which the parameters of each layer are recursively trained to learn how to invert.", "category": "method" }, { "sentence": "During the inversion, forward relu operation is adopted to adapt the general interpretations to the specific input.", "category": "method" }, { "sentence": "We then empirically demonstrate that the inverted source signal satisfies completeness and minimality property, which are crucial for a faithful interpretation.", "category": "result" }, { "sentence": "Furthermore, the empirical study validates the effectiveness of interpretations generated by MIP-IN.", "category": "result" } ] }, { "paper_id": "233839149", "title": "Polish adaptation of emotional Stroop test in assessment of pedophilia - a pilot study.", "abstract": "OBJECTIVES A pilot study was conducted in order to construct a Polish adaptation of emotional Stroop test in assessment of pedophilia. METHODS The study consisted of two stages. The first stage involved creating test material by ranking words in adequate lists by competent experts. The second stage consisted of empirical verification of the principle of emotional Stroop test in a non-clinical population. RESULTS Based on the assessment of five competent experts, words were ordered from the most to the least sexually arousing (Kendall's W from 0.368 to 0.693). Six ranked lists were obtained, and the competent experts were subsequently asked to assess whether these lists were suitable for the study (Lawshe's Content Validity Ratio from 0.6 to 1.0). Two categories of words were merged. Five ranked lists were obtained, and the competent experts were subsequently asked again to assess whether these lists were suitable for the study (Lawshe's Content Validity Ratio 1.0). The created lists of words were approved by allcompetent experts. Based on the experimental study conducted on a non-clinical population, it was shown that, in accordance with the principle of the test, the mean response time for sexually related words was longer that for neutral words. The mean response time for children-related words did not differ significantly from response time for neutral words. CONCLUSIONS Based on the study with competent experts and conducted experiments, an initial Polish adaptation of the emotional Stroop test for diagnosis of pedophilia has been created. Further studies with persons with pedophilia are needed to implement the test in clinical setting.", "classified_sentences": [ { "sentence": "A pilot study was conducted in order to construct a Polish adaptation of emotional Stroop test in assessment of pedophilia.", "category": "background" }, { "sentence": "The study consisted of two stages.", "category": "method" }, { "sentence": "The first stage involved creating test material by ranking words in adequate lists by competent experts.", "category": "method" }, { "sentence": "The second stage consisted of empirical verification of the principle of emotional Stroop test in a non-clinical population.", "category": "method" }, { "sentence": "Based on the assessment of five competent experts, words were ordered from the most to the least sexually arousing (Kendall's W from 0.368 to 0.693).", "category": "result" }, { "sentence": "Six ranked lists were obtained, and the competent experts were subsequently asked to assess whether these lists were suitable for the study (Lawshe's Content Validity Ratio from 0.6 to 1.0).", "category": "result" }, { "sentence": "Two categories of words were merged.", "category": "method" }, { "sentence": "Five ranked lists were obtained, and the competent experts were subsequently asked again to assess whether these lists were suitable for the study (Lawshe's Content Validity Ratio 1.0).", "category": "result" }, { "sentence": "The created lists of words were approved by allcompetent experts.", "category": "result" }, { "sentence": "Based on the experimental study conducted on a non-clinical population, it was shown that, in accordance with the principle of the test, the mean response time for sexually related words was longer that for neutral words.", "category": "result" }, { "sentence": "The mean response time for children-related words did not differ significantly from response time for neutral words.", "category": "result" }, { "sentence": "Based on the study with competent experts and conducted experiments, an initial Polish adaptation of the emotional Stroop test for diagnosis of pedophilia has been created.", "category": "result" }, { "sentence": "Further studies with persons with pedophilia are needed to implement the test in clinical setting.", "category": "result" } ] }, { "paper_id": "235351712", "title": "Deciphering Symbiotic Interactions of “Candidatus Aenigmarchaeota” with Inferred Horizontal Gene Transfers and Co-occurrence Networks", "abstract": "Recent advances in sequencing technology promoted the blowout discovery of super tiny microbes in the Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota (DPANN) superphylum. However, the unculturable properties of the majority of microbes impeded our investigation of their behavior and symbiotic lifestyle in the corresponding community. ABSTRACT “Candidatus Aenigmarchaeota” (“Ca. Aenigmarchaeota”) represents one of the earliest proposed evolutionary branches within the Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota (DPANN) superphylum. However, their ecological roles and potential host-symbiont interactions are still poorly understood. Here, eight metagenome-assembled genomes (MAGs) were reconstructed from hot spring ecosystems, and further in-depth comparative and evolutionary genomic analyses were conducted on these MAGs and other genomes downloaded from public databases. Although with limited metabolic capacities, we reported that “Ca. Aenigmarchaeota” in thermal environments harbor more genes related to carbohydrate metabolism than “Ca. Aenigmarchaeota” in nonthermal environments. Evolutionary analyses suggested that members from the Thaumarchaeota, Aigarchaeota, Crenarchaeota, and Korarchaeota (TACK) superphylum and Euryarchaeota contribute substantially to the niche expansion of “Ca. Aenigmarchaeota” via horizontal gene transfer (HGT), especially genes related to virus defense and stress responses. Based on co-occurrence network results and recent genetic exchanges among community members, we conjectured that “Ca. Aenigmarchaeota” may be symbionts associated with one MAG affiliated with the genus Pyrobaculum, though host specificity might be wide and variable across different “Ca. Aenigmarchaeota” organisms. This study provides significant insight into possible DPANN-host interactions and ecological roles of “Ca. Aenigmarchaeota. ” IMPORTANCE Recent advances in sequencing technology promoted the blowout discovery of super tiny microbes in the Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota (DPANN) superphylum. However, the unculturable properties of the majority of microbes impeded our investigation of their behavior and symbiotic lifestyle in the corresponding community. By integrating horizontal gene transfer (HGT) detection and co-occurrence network analysis on “Candidatus Aenigmarchaeota” (“Ca. Aenigmarchaeota”), we made one of the first attempts to infer their putative interaction partners and further decipher the potential functional and genetic interactions between the symbionts. We revealed that HGTs contributed by members from the Thaumarchaeota, Aigarchaeota, Crenarchaeota, and Korarchaeota (TACK) superphylum and Euryarchaeota conferred “Ca. Aenigmarchaeota” with the ability to survive under different environmental stresses, such as virus infection, high temperature, and oxidative stress. This study demonstrates that the interaction partners might be inferable by applying informatics analyses on metagenomic sequencing data.", "classified_sentences": [ { "sentence": "Recent advances in sequencing technology promoted the blowout discovery of super tiny microbes in the Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota (DPANN) superphylum.", "category": "background" }, { "sentence": "However, the unculturable properties of the majority of microbes impeded our investigation of their behavior and symbiotic lifestyle in the corresponding community.", "category": "background" }, { "sentence": "ABSTRACT “Candidatus Aenigmarchaeota” (“Ca. Aenigmarchaeota”) represents one of the earliest proposed evolutionary branches within the Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota (DPANN) superphylum.", "category": "background" }, { "sentence": "However, their ecological roles and potential host-symbiont interactions are still poorly understood.", "category": "background" }, { "sentence": "Here, eight metagenome-assembled genomes (MAGs) were reconstructed from hot spring ecosystems, and further in-depth comparative and evolutionary genomic analyses were conducted on these MAGs and other genomes downloaded from public databases.", "category": "method" }, { "sentence": "Although with limited metabolic capacities, we reported that “Ca. Aenigmarchaeota” in thermal environments harbor more genes related to carbohydrate metabolism than “Ca. Aenigmarchaeota” in nonthermal environments.", "category": "result" }, { "sentence": "Evolutionary analyses suggested that members from the Thaumarchaeota, Aigarchaeota, Crenarchaeota, and Korarchaeota (TACK) superphylum and Euryarchaeota contribute substantially to the niche expansion of “Ca. Aenigmarchaeota” via horizontal gene transfer (HGT), especially genes related to virus defense and stress responses.", "category": "result" }, { "sentence": "Based on co-occurrence network results and recent genetic exchanges among community members, we conjectured that “Ca. Aenigmarchaeota” may be symbionts associated with one MAG affiliated with the genus Pyrobaculum, though host specificity might be wide and variable across different “Ca. Aenigmarchaeota” organisms.", "category": "result" }, { "sentence": "This study provides significant insight into possible DPANN-host interactions and ecological roles of “Ca. Aenigmarchaeota”.", "category": "result" }, { "sentence": "“ IMPORTANCE Recent advances in sequencing technology promoted the blowout discovery of super tiny microbes in the Diapherotrites, Parvarchaeota, Aenigmarchaeota, Nanoarchaeota, and Nanohaloarchaeota (DPANN) superphylum.", "category": "background" }, { "sentence": "However, the unculturable properties of the majority of microbes impeded our investigation of their behavior and symbiotic lifestyle in the corresponding community.", "category": "background" }, { "sentence": "By integrating horizontal gene transfer (HGT) detection and co-occurrence network analysis on “Candidatus Aenigmarchaeota” (“Ca. Aenigmarchaeota”), we made one of the first attempts to infer their putative interaction partners and further decipher the potential functional and genetic interactions between the symbionts.", "category": "method" }, { "sentence": "We revealed that HGTs contributed by members from the Thaumarchaeota, Aigarchaeota, Crenarchaeota, and Korarchaeota (TACK) superphylum and Euryarchaeota conferred “Ca. Aenigmarchaeota” with the ability to survive under different environmental stresses, such as virus infection, high temperature, and oxidative stress.", "category": "result" }, { "sentence": "This study demonstrates that the interaction partners might be inferable by applying informatics analyses on metagenomic sequencing data.", "category": "result" } ] }, { "paper_id": "235392763", "title": "Volatility modelling in time and space 2020", "abstract": "This thesis contributes to the scientific community in several aspects. We introduce both spatialand spatio-temporal extensions to the family of GARCH and ARMA-GARCH models and present asymptotic statistics for the quasi maximum likelihood estimators [QMLE] for the GARCH extensions. An important property of these extensions are their spatialand spatio-temporal stationarity, which is part of the model specifications. The models all exist on an equidistant d-dimensional grid, be it purely spatial or spatiotemporal. Volatility modelling is important in finance, but we also present applications from other fields of study, e.g. climate, meteorology and even cell biology. In stationary spatial statistics on infinite lattices, a boundary problem arises. This is dealt with, in two of the papers, by assuming a circular model. This means wrapping the spatial part of the grid of observation onto a torus surface by connecting opposing edges, and effectively removing the boundaries so that each site’s neighbours are observed. The torus space is good for visualization and the point is that we regard sites on opposite sides of the rectangle we observe as neighbours. Circulation changes the area of observation from infinite to being closed and finite, and proving asymptotic results becomes easier. Consistency and asymptotic normality of the QMLE is established in the circular situation for GARCH models. The circular model can be used as an approximation of an infinite grid model, in which the circular estimator will be biased. In this setting, we suggest a parametric bootstrap bias correction to compensate for the false links between boundary sites due to circulation. In simulation studies, this approach provides good results for both GARCH and ARMA-GARCH models. For ARMA-GARCH, it is not uncommon to fit an ARMA model to data and a GARCH model to its residuals, but simultaneously estimating all parameters is better. We show by a simulation experiment that the variance of the ARMA-part of the QMLE can be reduced by doing this. The second paper of this thesis is an application of non-stationary GARCH modelling in climate research. We investigate how volatility has developed in a daily temperature series at Svalbard Airport over the last 44 years. During this period the temperature there has increased intensively. We model the volatility using a GARCH model with a trend, where the slope depends on the day of the year. Except for the summer, we find a decreasing temperature variability, i.e. a negative trend. The temperature on Svalbard is getting higher and more stable at the same time and we believe this is due to the reduced sea ice extent in the region. Without the circulation, on an infinite grid and in a potentially purely spatial setting, we turn to half-space GARCH models in the final paper. These models use an ordering of the spatial locations, extending non-deterministic time series to space. The MLE used is based on a modified likelihood, and we show that it is consistent and asymptotically Gaussian. Instead of the standard Lyapunov condition for existence of a stationary solution, a generalization of Nelson’s criteria is used.", "classified_sentences": [ { "sentence": "This thesis contributes to the scientific community in several aspects.", "category": "background" }, { "sentence": "We introduce both spatialand spatio-temporal extensions to the family of GARCH and ARMA-GARCH models and present asymptotic statistics for the quasi maximum likelihood estimators [QMLE] for the GARCH extensions.", "category": "method" }, { "sentence": "An important property of these extensions are their spatialand spatio-temporal stationarity, which is part of the model specifications.", "category": "method" }, { "sentence": "The models all exist on an equidistant d-dimensional grid, be it purely spatial or spatiotemporal.", "category": "method" }, { "sentence": "Volatility modelling is important in finance, but we also present applications from other fields of study, e.g. climate, meteorology and even cell biology.", "category": "background" }, { "sentence": "In stationary spatial statistics on infinite lattices, a boundary problem arises.", "category": "background" }, { "sentence": "This is dealt with, in two of the papers, by assuming a circular model.", "category": "method" }, { "sentence": "This means wrapping the spatial part of the grid of observation onto a torus surface by connecting opposing edges, and effectively removing the boundaries so that each site’s neighbours are observed.", "category": "method" }, { "sentence": "The torus space is good for visualization and the point is that we regard sites on opposite sides of the rectangle we observe as neighbours.", "category": "method" }, { "sentence": "Circulation changes the area of observation from infinite to being closed and finite, and proving asymptotic results becomes easier.", "category": "method" }, { "sentence": "Consistency and asymptotic normality of the QMLE is established in the circular situation for GARCH models.", "category": "result" }, { "sentence": "The circular model can be used as an approximation of an infinite grid model, in which the circular estimator will be biased.", "category": "method" }, { "sentence": "In this setting, we suggest a parametric bootstrap bias correction to compensate for the false links between boundary sites due to circulation.", "category": "method" }, { "sentence": "In simulation studies, this approach provides good results for both GARCH and ARMA-GARCH models.", "category": "result" }, { "sentence": "For ARMA-GARCH, it is not uncommon to fit an ARMA model to data and a GARCH model to its residuals, but simultaneously estimating all parameters is better.", "category": "method" }, { "sentence": "We show by a simulation experiment that the variance of the ARMA-part of the QMLE can be reduced by doing this.", "category": "result" }, { "sentence": "The second paper of this thesis is an application of non-stationary GARCH modelling in climate research.", "category": "background" }, { "sentence": "We investigate how volatility has developed in a daily temperature series at Svalbard Airport over the last 44 years.", "category": "background" }, { "sentence": "During this period the temperature there has increased intensively.", "category": "background" }, { "sentence": "We model the volatility using a GARCH model with a trend, where the slope depends on the day of the year.", "category": "method" }, { "sentence": "Except for the summer, we find a decreasing temperature variability, i.e. a negative trend.", "category": "result" }, { "sentence": "The temperature on Svalbard is getting higher and more stable at the same time and we believe this is due to the reduced sea ice extent in the region.", "category": "result" }, { "sentence": "Without the circulation, on an infinite grid and in a potentially purely spatial setting, we turn to half-space GARCH models in the final paper.", "category": "method" }, { "sentence": "These models use an ordering of the spatial locations, extending non-deterministic time series to space.", "category": "method" }, { "sentence": "The MLE used is based on a modified likelihood, and we show that it is consistent and asymptotically Gaussian.", "category": "result" }, { "sentence": "Instead of the standard Lyapunov condition for existence of a stationary solution, a generalization of Nelson’s criteria is used.", "category": "method" } ] }, { "paper_id": "235418915", "title": "The Double Framing Effect of Emotive Metaphors in Argumentation", "abstract": "In argumentation, metaphors are often considered as ambiguous or deceptive uses of language leading to fallacies of reasoning. However, they can also provide useful insights into creative argumentation, leading to genuinely new knowledge. Metaphors entail a framing effect that implicitly provides a specific perspective to interpret the world, guiding reasoning and evaluation of arguments. In the same vein, emotions could be in sharp contrast with proper reasoning, but they can also be cognitive processes of affective framing, influencing our reasoning and behavior in different meaningful ways. Thus, a double (metaphorical and affective) framing effect might influence argumentation in the case of emotive metaphors, such as “Poverty is a disease” or “Your boss is a dictator,” where specific “emotive words” (disease, dictator) are used as vehicles. We present and discuss the results of two experimental studies designed to explore the role of emotive metaphors in argumentation. The studies investigated whether and to what extent the detection of a fallacious argument is influenced by the presence of a conventional vs. novel emotive metaphor. Participants evaluated a series of verbal arguments containing either “non-emotive” or “emotive” (positive or negative) metaphors as middle terms that “bridge” the premises of the argument. The results show that the affective coherence of the metaphor's vehicle and topic plays a crucial role in participants' reasoning style, leading to global heuristic vs. local analytical interpretive processes in the interplay of the metaphorical and the affective framing effects.", "classified_sentences": [ { "sentence": "In argumentation, metaphors are often considered as ambiguous or deceptive uses of language leading to fallacies of reasoning.", "category": "background" }, { "sentence": "However, they can also provide useful insights into creative argumentation, leading to genuinely new knowledge.", "category": "background" }, { "sentence": "Metaphors entail a framing effect that implicitly provides a specific perspective to interpret the world, guiding reasoning and evaluation of arguments.", "category": "background" }, { "sentence": "In the same vein, emotions could be in sharp contrast with proper reasoning, but they can also be cognitive processes of affective framing, influencing our reasoning and behavior in different meaningful ways.", "category": "background" }, { "sentence": "Thus, a double (metaphorical and affective) framing effect might influence argumentation in the case of emotive metaphors, such as “Poverty is a disease” or “Your boss is a dictator,” where specific “emotive words” (disease, dictator) are used as vehicles.", "category": "background" }, { "sentence": "We present and discuss the results of two experimental studies designed to explore the role of emotive metaphors in argumentation.", "category": "method" }, { "sentence": "The studies investigated whether and to what extent the detection of a fallacious argument is influenced by the presence of a conventional vs. novel emotive metaphor.", "category": "method" }, { "sentence": "Participants evaluated a series of verbal arguments containing either “non-emotive” or “emotive” (positive or negative) metaphors as middle terms that “bridge” the premises of the argument.", "category": "method" }, { "sentence": "The results show that the affective coherence of the metaphor's vehicle and topic plays a crucial role in participants' reasoning style, leading to global heuristic vs. local analytical interpretive processes in the interplay of the metaphorical and the affective framing effects.", "category": "result" } ] }, { "paper_id": "237320542", "title": "Portocaval Shunt in Liver Transplantation, Why not?", "abstract": "*Correspondence: Laura Tortolero, Department of Liver Surgery, Ramón y Cajal Hospital, Carretera de Colmenar Viejo, Km 9, 100, Postal Code: 28034, Madrid, Spain, Tel: (+34) 656 518 646; E-mail: laura_tortolero@yahoo.com Received Date: 02 Jun 2020 Accepted Date: 02 Jul 2020 Published Date: 22 Jul 2020 Citation: Tortolero L, Nuño J, Luengo P, Gajate L. Portocaval Shunt in Liver Transplantation, Why not? Clin Surg. 2020; 5: 2871. Copyright © 2020 Laura Tortolero. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Letter to Editor Published: 22 Jul, 2020", "classified_sentences": [ { "sentence": "*Correspondence: Laura Tortolero, Department of Liver Surgery, Ramón y Cajal Hospital, Carretera de Colmenar Viejo, Km 9, 100, Postal Code: 28034, Madrid, Spain, Tel: (+34) 656 518 646; E-mail: laura_tortolero@yahoo.com Received Date: 02 Jun 2020 Accepted Date: 02 Jul 2020 Published Date: 22 Jul 2020 Citation: Tortolero L, Nuño J, Luengo P, Gajate L.", "category": "background" }, { "sentence": "Portocaval Shunt in Liver Transplantation, Why not?", "category": "background" }, { "sentence": "Clin Surg.", "category": "background" }, { "sentence": "2020; 5: 2871.", "category": "background" }, { "sentence": "Copyright © 2020 Laura Tortolero.", "category": "background" }, { "sentence": "This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.", "category": "background" }, { "sentence": "Letter to Editor Published: 22 Jul, 2020", "category": "background" } ] }, { "paper_id": "237953559", "title": "Retrospective Evalution of Pediatric Brucella Patients Followed and Family Screening with Household Members for Brucella Infection", "abstract": "Gereç ve Yöntemler: Altı yıllık zaman diliminde bruselloz tanısı almış 65 çocuk hastanın dosyaları geriye dönük olarak incelenmiştir. Çalışmaya alınan tüm hastaların ve bu hastalardan aile taraması yapılmış olanların brusella taramaları, demografik ve klinik özellikleri kaydedildi. Çalışmaya alınan 65 hastayla bu hastalardan yeni tanı alan 24’ünün yakın aile bireylerinin demografik ve klinik özellikleri kaydedildi. Tüm hasta ve yakın aile bireylerinin serum tüp aglütinasyon testi Çukurova Üniversitesi Tıp Fakültesi Merkez Laboratuvarında çalışıldı. İstatistiksel değerlendirme Çukurova Üniversitesi Tıp Fakültesi, Biyoistatistik Anabilim Dalında yapıldı.", "classified_sentences": [ { "sentence": "Gereç ve Yöntemler: Altı yıllık zaman diliminde bruselloz tanısı almış 65 çocuk hastanın dosyaları geriye dönük olarak incelenmiştir.", "category": "background" }, { "sentence": "Çalışmaya alınan tüm hastaların ve bu hastalardan aile taraması yapılmış olanların brusella taramaları, demografik ve klinik özellikleri kaydedildi.", "category": "method" }, { "sentence": "Çalışmaya alınan 65 hastayla bu hastalardan yeni tanı alan 24’ünün yakın aile bireylerinin demografik ve klinik özellikleri kaydedildi.", "category": "method" }, { "sentence": "Tüm hasta ve yakın aile bireylerinin serum tüp aglütinasyon testi Çukurova Üniversitesi Tıp Fakültesi Merkez Laboratuvarında çalışıldı.", "category": "method" }, { "sentence": "İstatistiksel değerlendirme Çukurova Üniversitesi Tıp Fakültesi, Biyoistatistik Anabilim Dalında yapıldı.", "category": "method" } ] }, { "paper_id": "238222431", "title": "High Abdominal Perfusion Pressure Using Umbilical Cord Flap in the Management of Gastroschisis", "abstract": "Background: Gastroschisis management remains a controversy. Most surgeons prefer reduction and fascial closure. Others advise staged reduction to avoid a sudden rise in intra-abdominal pressure (IAP). This study aims to evaluate the feasibility of using the umbilical cord as a flap (without skin on the top) for tension-free repair of gastroschisis. Methods: In a prospective study of neonates with gastroschisis repaired between January 2018 to October 2020 in Tanta University Hospital, we used the umbilical cord as a flap after the evacuation of all its blood vessels and suturing the edges of the cord with the skin edges of the defect. They were guided by monitoring abdominal perfusion pressure (APP), peak inspiratory pressure (PIP), central venous pressure (CVP), and urine output during 24 and 48 h postoperatively. The umbilical cord flap is used for tension-free closure of gastroschisis if PIP > 24 mmHg, IAP > 20 cmH2O (15 mmHg), APP <50 mmHg, and CVP > 15cmH2O. Results: In 20 cases that had gastroschisis with a median age of 24 h, we applied the umbilical cord flap in all cases and then purse string (Prolene Zero) with daily tightening till complete closure in seven cases, secondary suturing after 10 days in four cases, and leaving skin creeping until complete closure in nine cases. During the trials of closure, the range of APP was 49–52 mmHg. The range of IAP (IVP) was 15–20 cmH2O (11–15 mmHg), the range of PIP was 22–25 cmH2O, the range of CVP was 13–15 cmH2O, and the range of urine output was 1–1.5 ml/kg/h. Conclusion: The umbilical cord flap is an easy, feasible, and cheap method for tension-free closure of gastroschisis with limiting the PIP ≤ 24 mmHg, IAP ≤ 20 cmH2O (15 mmHg), APP > 50 mmHg, and CVP ≤ 15cmH2O.", "classified_sentences": [ { "sentence": "Background: Gastroschisis management remains a controversy.", "category": "background" }, { "sentence": "Most surgeons prefer reduction and fascial closure.", "category": "background" }, { "sentence": "Others advise staged reduction to avoid a sudden rise in intra-abdominal pressure (IAP).", "category": "background" }, { "sentence": "This study aims to evaluate the feasibility of using the umbilical cord as a flap (without skin on the top) for tension-free repair of gastroschisis.", "category": "method" }, { "sentence": "Methods: In a prospective study of neonates with gastroschisis repaired between January 2018 to October 2020 in Tanta University Hospital, we used the umbilical cord as a flap after the evacuation of all its blood vessels and suturing the edges of the cord with the skin edges of the defect.", "category": "method" }, { "sentence": "They were guided by monitoring abdominal perfusion pressure (APP), peak inspiratory pressure (PIP), central venous pressure (CVP), and urine output during 24 and 48 h postoperatively.", "category": "method" }, { "sentence": "The umbilical cord flap is used for tension-free closure of gastroschisis if PIP > 24 mmHg, IAP > 20 cmH2O (15 mmHg), APP <50 mmHg, and CVP > 15cmH2O.", "category": "method" }, { "sentence": "Results: In 20 cases that had gastroschisis with a median age of 24 h, we applied the umbilical cord flap in all cases and then purse string (Prolene Zero) with daily tightening till complete closure in seven cases, secondary suturing after 10 days in four cases, and leaving skin creeping until complete closure in nine cases.", "category": "result" }, { "sentence": "During the trials of closure, the range of APP was 49–52 mmHg.", "category": "result" }, { "sentence": "The range of IAP (IVP) was 15–20 cmH2O (11–15 mmHg), the range of PIP was 22–25 cmH2O, the range of CVP was 13–15 cmH2O, and the range of urine output was 1–1.5 ml/kg/h.", "category": "result" }, { "sentence": "Conclusion: The umbilical cord flap is an easy, feasible, and cheap method for tension-free closure of gastroschisis with limiting the PIP ≤ 24 mmHg, IAP ≤ 20 cmH2O (15 mmHg), APP > 50 mmHg, and CVP ≤ 15cmH2O.", "category": "result" } ] }, { "paper_id": "108835350", "title": "Measurement of high-field electron transport in silicon carbide", "abstract": "We report recent measurements of the drift velocity of electrons parallel to the basal plane in 6H and 4H silicon carbide (SiC) as a function of applied electric field. The dependence of the low field mobility and saturated drift velocity on temperature are also reported. The saturated drift velocities at room temperature are approximately 1.9/spl times/10/sup 7/ cm/s in 6H-SiC and 2.2/spl times/10/sup 7/ cm/s in 4H-SiC.", "classified_sentences": [ { "sentence": "We report recent measurements of the drift velocity of electrons parallel to the basal plane in 6H and 4H silicon carbide (SiC) as a function of applied electric field.", "category": "method" }, { "sentence": "The dependence of the low field mobility and saturated drift velocity on temperature are also reported.", "category": "method" }, { "sentence": "The saturated drift velocities at room temperature are approximately 1.9/spl times/10/sup 7/ cm/s in 6H-SiC and 2.2/spl times/10/sup 7/ cm/s in 4H-SiC.", "category": "result" } ] }, { "paper_id": "244087909", "title": "Real-Time Binaural Room Modeling for Augmented Reality Applications", "abstract": "This paper proposes and evaluates an integrated method for real-time, head-tracked, 3D binaural audio with synthetic reverberation. Virtual vector base amplitude panning is used to position the sound source and spatialize outputs from a scattering delay network reverb algorithm running in parallel. A unique feature of this approach is its realization of interactive auralization using vector base amplitude panning and a scattering delay network, within acceptable levels of latency, at low computational cost. The rendering model also allows direct parameterization of room geometry and absorption characteristics. Varying levels of reverb complexity can be implemented, and these were evaluated against two distinct aspects of perceived sonic immersion. Outcomes from the evaluation provide benchmarks for how the approach could be deployed adaptively, to balance three real-time spatial audio objectives of envelopment, naturalness, and efficiency, within contrasting physical spaces.", "classified_sentences": [ { "sentence": "This paper proposes and evaluates an integrated method for real-time, head-tracked, 3D binaural audio with synthetic reverberation.", "category": "background" }, { "sentence": "Virtual vector base amplitude panning is used to position the sound source and spatialize outputs from a scattering delay network reverb algorithm running in parallel.", "category": "method" }, { "sentence": "A unique feature of this approach is its realization of interactive auralization using vector base amplitude panning and a scattering delay network, within acceptable levels of latency, at low computational cost.", "category": "method" }, { "sentence": "The rendering model also allows direct parameterization of room geometry and absorption characteristics.", "category": "method" }, { "sentence": "Varying levels of reverb complexity can be implemented, and these were evaluated against two distinct aspects of perceived sonic immersion.", "category": "method" }, { "sentence": "Outcomes from the evaluation provide benchmarks for how the approach could be deployed adaptively, to balance three real-time spatial audio objectives of envelopment, naturalness, and efficiency, within contrasting physical spaces.", "category": "result" } ] }, { "paper_id": "244870302", "title": "OTT_A_342613 5373..5383", "abstract": "1Department of Thoracic Surgery, Shengjing Hospital, China Medical University, Shenyang, People’s Republic of China; 2Department of General Surgery, Shengjing Hospital, China Medical University, Shenyang, People’s Republic of China Background: Gastric cancer (GC) ranks fifth in global cancer incidence and third in cancerrelated mortality. The prognosis of GC patients was poor. Necroptosis is a type of regulated cell death mediated by RIP1, RIP3, and MLKL. Necroptosis was found to be involved in antitumor immunity in the cancer immunotherapy. Methods: LASSO Cox regression analysis was performed to construct a prognostic signature. Bioinformatics analysis was performed to construct a lncRNA-miRNA-mRNA regulatory axis. qRT-PCR was performed to verify the expression and prognosis of hub gene in STAD. Results: Most of necroptosis regulators were upregulated, while the mRNA level of TLR3, ALDH2, and NDRG2 was downregulated in STAD versus gastric tissues. The genetic mutation and copy number variation of necroptosis regulator in STAD were also summarized. GO and KEGG pathways analysis revealed that these necroptosis regulators were mainly involved in programmed necrotic cell death and TNF signaling pathway. A necroptosis-related prognostic signature based on four genes (EZH2, PGAM5, TLR4, and TRAF2) had a good performance in predicting the prognosis of STAD patients. We also identified lncRNA SNHG1/miR-21-5p/TLR4 regulatory axis in the progression in STAD. Verification study suggested that the hub gene TLR4 upregulated in STAD and correlated with a poor overall survival. Moreover, Cox regression analysis revealed that TLR4 expression and clinical stage were independent factors affecting the prognosis of STAD patients. Conclusion: We performed a comprehensive bioinformatics analysis and identified a necroptosis-related prognostic signature and a lncRNA SNHG1/miR-21-5p/TLR4 regulatory axis in STAD. Further study should be performed to confirm our result.", "classified_sentences": [ { "sentence": "1Department of Thoracic Surgery, Shengjing Hospital, China Medical University, Shenyang, People’s Republic of China; 2Department of General Surgery, Shengjing Hospital, China Medical University, Shenyang, People’s Republic of China Background: Gastric cancer (GC) ranks fifth in global cancer incidence and third in cancerrelated mortality.", "category": "background" }, { "sentence": "The prognosis of GC patients was poor.", "category": "background" }, { "sentence": "Necroptosis is a type of regulated cell death mediated by RIP1, RIP3, and MLKL.", "category": "background" }, { "sentence": "Necroptosis was found to be involved in antitumor immunity in the cancer immunotherapy.", "category": "background" }, { "sentence": "Methods: LASSO Cox regression analysis was performed to construct a prognostic signature.", "category": "method" }, { "sentence": "Bioinformatics analysis was performed to construct a lncRNA-miRNA-mRNA regulatory axis.", "category": "method" }, { "sentence": "qRT-PCR was performed to verify the expression and prognosis of hub gene in STAD.", "category": "method" }, { "sentence": "Results: Most of necroptosis regulators were upregulated, while the mRNA level of TLR3, ALDH2, and NDRG2 was downregulated in STAD versus gastric tissues.", "category": "result" }, { "sentence": "The genetic mutation and copy number variation of necroptosis regulator in STAD were also summarized.", "category": "result" }, { "sentence": "GO and KEGG pathways analysis revealed that these necroptosis regulators were mainly involved in programmed necrotic cell death and TNF signaling pathway.", "category": "result" }, { "sentence": "A necroptosis-related prognostic signature based on four genes (EZH2, PGAM5, TLR4, and TRAF2) had a good performance in predicting the prognosis of STAD patients.", "category": "result" }, { "sentence": "We also identified lncRNA SNHG1/miR-21-5p/TLR4 regulatory axis in the progression in STAD.", "category": "result" }, { "sentence": "Verification study suggested that the hub gene TLR4 upregulated in STAD and correlated with a poor overall survival.", "category": "result" }, { "sentence": "Moreover, Cox regression analysis revealed that TLR4 expression and clinical stage were independent factors affecting the prognosis of STAD patients.", "category": "result" }, { "sentence": "Conclusion: We performed a comprehensive bioinformatics analysis and identified a necroptosis-related prognostic signature and a lncRNA SNHG1/miR-21-5p/TLR4 regulatory axis in STAD.", "category": "result" }, { "sentence": "Further study should be performed to confirm our result.", "category": "result" } ] }, { "paper_id": "248887396", "title": "RankGen: Improving Text Generation with Large Ranking Models", "abstract": "Given an input sequence (or prefix), modern language models often assign high probabilities to output sequences that are repetitive, incoherent, or irrelevant to the prefix; as such, model-generated text also contains such artifacts. To address these issues we present RankGen, a 1.2B parameter encoder model for English that scores model generations given a prefix. RankGen can be flexibly incorporated as a scoring function in beam search and used to decode from any pretrained language model. We train RankGen using large-scale contrastive learning to map a prefix close to the ground-truth sequence that follows it and far away from two types of negatives: (1) random sequences from the same document as the prefix, and (2) sequences generated from a large language model conditioned on the prefix. Experiments across four different language models (345M-11B parameters) and two domains show that RankGen significantly outperforms decoding algorithms like nucleus, top-k, and typical sampling on both automatic metrics (85.0 vs 77.3 MAUVE) as well as human evaluations with English writers (74.5% human preference over nucleus sampling). Analysis reveals that RankGen outputs are more relevant to the prefix and improve continuity and coherence compared to baselines. We release our model checkpoints, code, and human preference data with explanations to facilitate future research.", "classified_sentences": [ { "sentence": "Given an input sequence (or prefix), modern language models often assign high probabilities to output sequences that are repetitive, incoherent, or irrelevant to the prefix; as such, model-generated text also contains such artifacts.", "category": "background" }, { "sentence": "To address these issues we present RankGen, a 1.2B parameter encoder model for English that scores model generations given a prefix.", "category": "method" }, { "sentence": "RankGen can be flexibly incorporated as a scoring function in beam search and used to decode from any pretrained language model.", "category": "method" }, { "sentence": "We train RankGen using large-scale contrastive learning to map a prefix close to the ground-truth sequence that follows it and far away from two types of negatives: (1) random sequences from the same document as the prefix, and (2) sequences generated from a large language model conditioned on the prefix.", "category": "method" }, { "sentence": "Experiments across four different language models (345M-11B parameters) and two domains show that RankGen significantly outperforms decoding algorithms like nucleus, top-k, and typical sampling on both automatic metrics (85.0 vs 77.3 MAUVE) as well as human evaluations with English writers (74.5% human preference over nucleus sampling).", "category": "result" }, { "sentence": "Analysis reveals that RankGen outputs are more relevant to the prefix and improve continuity and coherence compared to baselines.", "category": "result" }, { "sentence": "We release our model checkpoints, code, and human preference data with explanations to facilitate future research.", "category": "method" } ] }, { "paper_id": "251566690", "title": "Microbial Contamination and Identification of Bacterial for Mobiles Phones in Iraq", "abstract": "This study was conducted to isolate and identify bacteria contaminants on a mobile phone. The samples were collected randomly from 20 mobile phones. This study was conducted between October to December, 2016 at College of Biotechnology, Al-Nahrain University. The isolated colonies were then sub cultured in nutrient agar and slants in order to obtain pure culture of all the six colonies. Six genera of bacteria were identified from positive cultures. In all, 20 swab samples of mobile phone were randomly examined, 19 bacterial isolates were identified from mobile phones were found contaminated with microbiota. The highest prevalence of Staphylococcus aureus was observed in mobile phones. The research findings indicated that S. aureus (8 isolates), Escherichia coli (4 isolates), Enterobacter spp (2 isolates), Bacillus (1 isolates ), Streptococceus spp (1 isolates), and Pseudomonas spp (3 isolates), were the main isolates frequently associated with the mobile phones. Showed Percentage of bacterial isolates from the samples collected from mobile phones after calculating the total percentage of each isolate, found S. aureus, E. coli, Enterobacter spp, Bacillus, Streptococceus spp and Pseudomonas spp in the percentage of 42.10 %, 21.05 %, 10.52%, 5.26 %, 5.26 % and 15.78 % respectively. The results showed that mobile phones were contaminated with different types of bacteria mentioned above. Gram positive cocci, Streptococcus and Staphylococcus spp. were identified based on morphological characteristics. Gram negative bacilli, E. coli, Enterobacter, Bacillus and Pseudomonas were identified based on morphological characteristics. Nineteen isolates from 20 observed mobile phones belonging to the students. The highest prevalence in male was (13 isolates) and were percentage of bacteria isolated 66.66%, while in female were (6 isolates) and percentage of bacteria isolated 33.33%. Also showed results Percentage of total bacteria isolated of female and male, were 31.57% and 68.42 % respectively.", "classified_sentences": [ { "sentence": "This study was conducted to isolate and identify bacteria contaminants on a mobile phone.", "category": "background" }, { "sentence": "The samples were collected randomly from 20 mobile phones.", "category": "method" }, { "sentence": "This study was conducted between October to December, 2016 at College of Biotechnology, Al-Nahrain University.", "category": "background" }, { "sentence": "The isolated colonies were then sub cultured in nutrient agar and slants in order to obtain pure culture of all the six colonies.", "category": "method" }, { "sentence": "Six genera of bacteria were identified from positive cultures.", "category": "result" }, { "sentence": "In all, 20 swab samples of mobile phone were randomly examined, 19 bacterial isolates were identified from mobile phones were found contaminated with microbiota.", "category": "result" }, { "sentence": "The highest prevalence of Staphylococcus aureus was observed in mobile phones.", "category": "result" }, { "sentence": "The research findings indicated that S.", "category": "result" }, { "sentence": "aureus (8 isolates), Escherichia coli (4 isolates), Enterobacter spp (2 isolates), Bacillus (1 isolates ), Streptococceus spp (1 isolates), and Pseudomonas spp (3 isolates), were the main isolates frequently associated with the mobile phones.", "category": "result" }, { "sentence": "Showed Percentage of bacterial isolates from the samples collected from mobile phones after calculating the total percentage of each isolate, found S.", "category": "result" }, { "sentence": "aureus, E.", "category": "result" }, { "sentence": "coli, Enterobacter spp, Bacillus, Streptococceus spp and Pseudomonas spp in the percentage of 42.10 %, 21.05 %, 10.52%, 5.26 %, 5.26 % and 15.78 % respectively.", "category": "result" }, { "sentence": "The results showed that mobile phones were contaminated with different types of bacteria mentioned above.", "category": "result" }, { "sentence": "Gram positive cocci, Streptococcus and Staphylococcus spp.", "category": "result" }, { "sentence": "were identified based on morphological characteristics.", "category": "result" }, { "sentence": "Gram negative bacilli, E.", "category": "result" }, { "sentence": "coli, Enterobacter, Bacillus and Pseudomonas were identified based on morphological characteristics.", "category": "result" }, { "sentence": "Nineteen isolates from 20 observed mobile phones belonging to the students.", "category": "result" }, { "sentence": "The highest prevalence in male was (13 isolates) and were percentage of bacteria isolated 66.66%, while in female were (6 isolates) and percentage of bacteria isolated 33.33%.", "category": "result" }, { "sentence": "Also showed results Percentage of total bacteria isolated of female and male, were 31.57% and 68.42 % respectively.", "category": "result" } ] }, { "paper_id": "252054165", "title": ": Short-term effects of exposure to particulate matter on hospital admissions for asthma and chronic obstructive pulmonary disease", "abstract": "We investigated the effects of particulate matter (PM) factors on hospitalization rates for asthma and chronic obstructive pulmonary disease (COPD). We obtained data on pollutants—PM10, PM2.5—in Seoul, South Korea. We also investigated data for asthma and COPD exacerbation that required hospitalization from 2006 to 2016. We used a time-stratified case-crossover design and generalized additive models with log transformation to assess adjusted risk, and conditional logistic regression was performed to analyze these data. Our study showed that PM10 and PM2.5, on different best lag days, were associated with increased risks of COPD or asthma hospitalization. The odds ratios (ORs) for each per-unit increase in PM10 and PM2.5 were higher in patients with male asthma (PM10: OR, 1.012; 95% confidence interval [CI], 1.008–1.016 and PM2.5: OR, 1.015; 95% CI, 1008–1.023), preschool asthma (PM10: OR, 1.015; 95% CI, 1.006–1.015 and PM2.5: OR, 1.015; 95% CI, 1.009–1.024), male COPD (PM10: OR, 1.012; 95% CI, 1.005–1.019 and PM2.5: OR, 1.013; 95% CI, 1.000–1.026), and senior COPD (PM10: OR, 1.016; 95% CI, 1.008–1.024 and PM2.5: OR, 1.022; 95% CI, 1.007–1.036). Increasing PM levels increased hospitalizations for asthma and COPD. Additionally, the consequences may be different according to age and sex, and PM2.5 may have a more significant effect on airway disease patients than PM10.", "classified_sentences": [ { "sentence": "We investigated the effects of particulate matter (PM) factors on hospitalization rates for asthma and chronic obstructive pulmonary disease (COPD).", "category": "background" }, { "sentence": "We obtained data on pollutants—PM10, PM2.5—in Seoul, South Korea.", "category": "method" }, { "sentence": "We also investigated data for asthma and COPD exacerbation that required hospitalization from 2006 to 2016.", "category": "method" }, { "sentence": "We used a time-stratified case-crossover design and generalized additive models with log transformation to assess adjusted risk, and conditional logistic regression was performed to analyze these data.", "category": "method" }, { "sentence": "Our study showed that PM10 and PM2.5, on different best lag days, were associated with increased risks of COPD or asthma hospitalization.", "category": "result" }, { "sentence": "The odds ratios (ORs) for each per-unit increase in PM10 and PM2.5 were higher in patients with male asthma (PM10: OR, 1.012; 95% confidence interval [CI], 1.008–1.016 and PM2.5: OR, 1.015; 95% CI, 1008–1.023), preschool asthma (PM10: OR, 1.015; 95% CI, 1.006–1.015 and PM2.5: OR, 1.015; 95% CI, 1.009–1.024), male COPD (PM10: OR, 1.012; 95% CI, 1.005–1.019 and PM2.5: OR, 1.013; 95% CI, 1.000–1.026), and senior COPD (PM10: OR, 1.016; 95% CI, 1.008–1.024 and PM2.5: OR, 1.022; 95% CI, 1.007–1.036).", "category": "result" }, { "sentence": "Increasing PM levels increased hospitalizations for asthma and COPD.", "category": "result" }, { "sentence": "Additionally, the consequences may be different according to age and sex, and PM2.5 may have a more significant effect on airway disease patients than PM10.", "category": "result" } ] }, { "paper_id": "253329540", "title": "Dynamic Selection Network For Rgb-D Salient Object Detection", "abstract": "Existing RGB-D salient object detection (SOD) methods usually use elaborate fusion modules for exploring cross-modal information, which is computationally expensive and ignores the noise depth information. To deal with this issue, we propose a dynamic selection network (DSNet) for RGB-D salient object detection. Specifically, a cross-modal combination module (CCM) is proposed to fuse two modalities with a light computation. Then a dynamic selection module (DSM) adaptively learns the model parameter for the decoding based on the fused features. Furthermore, skip connection is used for hierarchical features combination between encoder and decoder. Experiments on four popular datasets demonstrate our model outperforms other state-of-the-art methods.", "classified_sentences": [ { "sentence": "Existing RGB-D salient object detection (SOD) methods usually use elaborate fusion modules for exploring cross-modal information, which is computationally expensive and ignores the noise depth information.", "category": "background" }, { "sentence": "To deal with this issue, we propose a dynamic selection network (DSNet) for RGB-D salient object detection.", "category": "method" }, { "sentence": "Specifically, a cross-modal combination module (CCM) is proposed to fuse two modalities with a light computation.", "category": "method" }, { "sentence": "Then a dynamic selection module (DSM) adaptively learns the model parameter for the decoding based on the fused features.", "category": "method" }, { "sentence": "Furthermore, skip connection is used for hierarchical features combination between encoder and decoder.", "category": "method" }, { "sentence": "Experiments on four popular datasets demonstrate our model outperforms other state-of-the-art methods.", "category": "result" } ] }, { "paper_id": "254640938", "title": "Recruitment of the cardiac conduction system for optimal resynchronization therapy in failing heart", "abstract": "Heart failure (HF) is a leading health burden around the world. Although pharmacological development has dramatically advanced medication therapy in the field, hemodynamic disorders or mechanical desynchrony deteriorated by intra or interventricular conduction abnormalities remains a critical target beyond the scope of pharmacotherapy. In the past 2 decades, nonpharmacologic treatment for heart failure, such as cardiac resynchronization therapy (CRT) via biventricular pacing (BVP), has been playing an important role in improving the prognosis of heart failure. However, the response rate of BVP-CRT is variable, leaving one-third of patients not benefiting from the therapy as expected. Considering the non-physiological activation pattern of BVP-CRT, more efforts have been made to optimize resynchronization. The most extensively investigated approach is by stimulating the native conduction system, e.g., His-Purkinje conduction system pacing (CSP), including His bundle pacing (HBP) and left bundle branch area pacing (LBBAP). These emerging CRT approaches provide an alternative to traditional BVP-CRT, with multiple proof-of-concept studies indicating the safety and efficacy of its utilization in dyssynchronous heart failure. In this review, we summarize the mechanisms of dyssynchronous HF mediated by conduction disturbance, the rationale and acute effect of CSP for CRT, the recent advancement in clinical research, and possible future directions of CSP. Graphical Abstract Emerging strategies for cardiac resynchronization for dyssynchronous heart failure.", "classified_sentences": [ { "sentence": "Heart failure (HF) is a leading health burden around the world.", "category": "background" }, { "sentence": "Although pharmacological development has dramatically advanced medication therapy in the field, hemodynamic disorders or mechanical desynchrony deteriorated by intra or interventricular conduction abnormalities remains a critical target beyond the scope of pharmacotherapy.", "category": "background" }, { "sentence": "In the past 2 decades, nonpharmacologic treatment for heart failure, such as cardiac resynchronization therapy (CRT) via biventricular pacing (BVP), has been playing an important role in improving the prognosis of heart failure.", "category": "background" }, { "sentence": "However, the response rate of BVP-CRT is variable, leaving one-third of patients not benefiting from the therapy as expected.", "category": "background" }, { "sentence": "Considering the non-physiological activation pattern of BVP-CRT, more efforts have been made to optimize resynchronization.", "category": "method" }, { "sentence": "The most extensively investigated approach is by stimulating the native conduction system, e.g., His-Purkinje conduction system pacing (CSP), including His bundle pacing (HBP) and left bundle branch area pacing (LBBAP).", "category": "method" }, { "sentence": "These emerging CRT approaches provide an alternative to traditional BVP-CRT, with multiple proof-of-concept studies indicating the safety and efficacy of its utilization in dyssynchronous heart failure.", "category": "result" }, { "sentence": "In this review, we summarize the mechanisms of dyssynchronous HF mediated by conduction disturbance, the rationale and acute effect of CSP for CRT, the recent advancement in clinical research, and possible future directions of CSP.", "category": "method" }, { "sentence": "Graphical Abstract Emerging strategies for cardiac resynchronization for dyssynchronous heart failure.", "category": "result" } ] }, { "paper_id": "121053719", "title": "A New Vector Partition of the Probability Score", "abstract": "Abstract A new vector partition of the probability, or Brier, score (PS) is formulated and the nature and properties of this partition are described. The relationships between the terms in this partition and the terms in the original vector partition of the PS are indicated. The new partition consists of three terms: 1) a measure of the uncertainty inherent in the events, or states, on the occasions of concern (namely, the PS for the sample relative frequencies); 2) a measure of the reliability of the forecasts; and 3) a new measure of the resolution of the forecasts. These measures of reliability and resolution are and are not, respectively, equivalent (i.e., linearly related) to the measures of reliability and resolution provided by the original partition. Two sample collections of probability forecasts are used to illustrate the differences and relationships between these partitions. Finally, the two partitions are compared, with particular reference to the attributes of the forecasts with which the pa.", "classified_sentences": [ { "sentence": "Abstract A new vector partition of the probability, or Brier, score (PS) is formulated and the nature and properties of this partition are described.", "category": "method" }, { "sentence": "The relationships between the terms in this partition and the terms in the original vector partition of the PS are indicated.", "category": "method" }, { "sentence": "The new partition consists of three terms: 1) a measure of the uncertainty inherent in the events, or states, on the occasions of concern (namely, the PS for the sample relative frequencies); 2) a measure of the reliability of the forecasts; and 3) a new measure of the resolution of the forecasts.", "category": "method" }, { "sentence": "These measures of reliability and resolution are and are not, respectively, equivalent (i.e., linearly related) to the measures of reliability and resolution provided by the original partition.", "category": "result" }, { "sentence": "Two sample collections of probability forecasts are used to illustrate the differences and relationships between these partitions.", "category": "method" }, { "sentence": "Finally, the two partitions are compared, with particular reference to the attributes of the forecasts with which the pa.", "category": "result" } ] }, { "paper_id": "122716558", "title": "Route to chaos in porous-medium thermal convection", "abstract": "A pseudo-spectral numerical scheme is used to study two-dimensional, single-cell, time-dependent convection in a square cross-section of fluid saturated porous material heated from below. With increasing Rayleigh number R convection evolves from steady S to chaotic NP through the sequence of bifurcations S→P(1)→QP2→P(2)→NP, where P(1) and P(2) are simply periodic regimes and QP2 is a quasi-periodic state with two basic frequencies. The transitions (from onset of convection to chaos) occur at Rayleigh numbers of 4π2, 380–400, 500–520, 560–570, and 850–1000. In the first simply periodic regime the fundamental frequency f1 varies as $R^{\\frac{7}{8}} $ and the average Nusselt number $\\overline{Nu}$ is proportional to $R^{\\frac{2}{3}}$; in P(2), f1 varies as $R^{\\frac{3}{2}}$ and $\\overline{Nu}\\propto R^{\\frac{11}{10}}$. Convection in QP2 exhibits hysteresis, i.e. if the QP2 state is reached from P(1) (P(2)) by increasing (decreasing) R then the frequency with the largest spectral power is the one consistent with the extrapolation of f1 according to $R^{\\frac{7}{8}}(R^{\\frac{3}{2}})$. The chaotic states are characterized by spectral peaks with at least 3 fundamental frequencies superimposed on a broadband background noise. The time dependence of these states arises from the random generation of tongue-like disturbances within the horizontal thermal boundary layers. Transition to the chaotic regime is accompanied by the growth of spectral components that destroy the centre-symmetry of convection in the other states. Over-truncation can lead to spurious transitions and bifurcation sequences; in general it produces overly complex flows.", "classified_sentences": [ { "sentence": "A pseudo-spectral numerical scheme is used to study two-dimensional, single-cell, time-dependent convection in a square cross-section of fluid saturated porous material heated from below.", "category": "method" }, { "sentence": "With increasing Rayleigh number R convection evolves from steady S to chaotic NP through the sequence of bifurcations S→P(1)→QP2→P(2)→NP, where P(1) and P(2) are simply periodic regimes and QP2 is a quasi-periodic state with two basic frequencies.", "category": "result" }, { "sentence": "The transitions (from onset of convection to chaos) occur at Rayleigh numbers of 4π2, 380–400, 500–520, 560–570, and 850–1000.", "category": "result" }, { "sentence": "In the first simply periodic regime the fundamental frequency f1 varies as $R^{\\frac{7}{8}} $ and the average Nusselt number $\\overline{Nu}$ is proportional to $R^{\\frac{2}{3}}$; in P(2), f1 varies as $R^{\\frac{3}{2}}$ and $\\overline{Nu}\\propto R^{\\frac{11}{10}}$.", "category": "result" }, { "sentence": "Convection in QP2 exhibits hysteresis, i.e. if the QP2 state is reached from P(1) (P(2)) by increasing (decreasing) R then the frequency with the largest spectral power is the one consistent with the extrapolation of f1 according to $R^{\\frac{7}{8}}(R^{\\frac{3}{2}})$.", "category": "result" }, { "sentence": "The chaotic states are characterized by spectral peaks with at least 3 fundamental frequencies superimposed on a broadband background noise.", "category": "result" }, { "sentence": "The time dependence of these states arises from the random generation of tongue-like disturbances within the horizontal thermal boundary layers.", "category": "result" }, { "sentence": "Transition to the chaotic regime is accompanied by the growth of spectral components that destroy the centre-symmetry of convection in the other states.", "category": "result" }, { "sentence": "Over-truncation can lead to spurious transitions and bifurcation sequences; in general it produces overly complex flows.", "category": "result" } ] }, { "paper_id": "123059443", "title": "Verification of TIGGE Multimodel and ECMWF Reforecast-Calibrated Probabilistic Precipitation Forecasts over the Contiguous United States*", "abstract": "AbstractProbabilistic quantitative precipitation forecasts (PQPFs) were generated from The Observing System Research and Predictability Experiment (THORPEX) Interactive Grand Global Ensemble (TIGGE) database from July to October 2010 using data from Europe (ECMWF), the United Kingdom [Met Office (UKMO)], the United States (NCEP), and Canada [Canadian Meteorological Centre (CMC)]. Forecasts of 24-h accumulated precipitation were evaluated at 1° grid spacing within the contiguous United States against analysis data based on gauges and bias-corrected radar data.PQPFs from ECMWF’s ensembles generally had the highest skill of the raw ensemble forecasts, followed by CMC. Those of UKMO and NCEP were less skillful. PQPFs from CMC forecasts were the most reliable but the least sharp, and PQPFs from NCEP and UKMO ensembles were the least reliable but sharper.Multimodel PQPFs were more reliable and skillful than individual ensemble prediction system forecasts. The improvement was larger for heavier precipitation eve.", "classified_sentences": [ { "sentence": "AbstractProbabilistic quantitative precipitation forecasts (PQPFs) were generated from The Observing System Research and Predictability Experiment (THORPEX) Interactive Grand Global Ensemble (TIGGE) database from July to October 2010 using data from Europe (ECMWF), the United Kingdom [Met Office (UKMO)], the United States (NCEP), and Canada [Canadian Meteorological Centre (CMC)].", "category": "background" }, { "sentence": "Forecasts of 24-h accumulated precipitation were evaluated at 1° grid spacing within the contiguous United States against analysis data based on gauges and bias-corrected radar data.PQPFs from ECMWF’s ensembles generally had the highest skill of the raw ensemble forecasts, followed by CMC.", "category": "method" }, { "sentence": "Those of UKMO and NCEP were less skillful.", "category": "result" }, { "sentence": "PQPFs from CMC forecasts were the most reliable but the least sharp, and PQPFs from NCEP and UKMO ensembles were the least reliable but sharper.Multimodel PQPFs were more reliable and skillful than individual ensemble prediction system forecasts.", "category": "result" }, { "sentence": "The improvement was larger for heavier precipitation eve.", "category": "result" } ] }, { "paper_id": "123347891", "title": "Functions without short implicents. Part II: Construction", "abstract": "Abstract This paper is a continuation of the paper ‘Functions without short implicents. Part I: lower estimates of weights’. In Part II we propose various methods of construction of n-place Boolean functions not admitting implicents of k variables. The first of the methods proposed is based on the gradient algorithm, the second and the third ones depend on a certain combinatorial principle of construction, while the fourth method is based on a random choice of elements in the function support. The above methods have different efficiency depending on the value of k. We give upper estimates for the minimal value w (n, k) of weights of the so-constructed functions. Together with the lower estimates of w (n, k) from the first part of the paper this allows us to obtain an asymptotically sharp estimate w (n, k) = Θ (ln n) as n → ∞.", "classified_sentences": [ { "sentence": "Abstract This paper is a continuation of the paper ‘Functions without short implicents. Part I: lower estimates of weights’.", "category": "background" }, { "sentence": "In Part II we propose various methods of construction of n-place Boolean functions not admitting implicents of k variables.", "category": "method" }, { "sentence": "The first of the methods proposed is based on the gradient algorithm, the second and the third ones depend on a certain combinatorial principle of construction, while the fourth method is based on a random choice of elements in the function support.", "category": "method" }, { "sentence": "The above methods have different efficiency depending on the value of k.", "category": "method" }, { "sentence": "We give upper estimates for the minimal value w (n, k) of weights of the so-constructed functions.", "category": "result" }, { "sentence": "Together with the lower estimates of w (n, k) from the first part of the paper this allows us to obtain an asymptotically sharp estimate w (n, k) = Θ (ln n) as n → ∞.", "category": "result" } ] }, { "paper_id": "257882460", "title": "Dynamic behavior of earth dams under different kinematic impacts", "abstract": "The paper provides a detailed analysis of the current state of the problem. A mathematical model is presented to determine the dynamic behavior of earth dams, considering the viscoelastic properties of soil, using the hereditary Boltzmann-Volterra theory with the A.R. Rzhanitsyn kernel under periodic kinematic impacts. To solve the problems considered, the finite element method and complex arithmetics were used to reduce integrodifferential equations to a high-order complex algebraic equation. The accuracy of the methods was verified by solving test problems. Steadystate forced vibrations of the Pskem earth dam 195 m high are studied considering the real geometry and soil properties under resonant vibration modes. It was stated that the largest stress amplitudes in the body of the dam occur not only under the first resonance, but they can occur under other dense spectra of the eigenfrequencies of the dam, due to the interaction between close natural modes of vibration. The strength of various sections of the dam body was tested under kinematic impact using the Coulomb- Mohr theory of strength; the most dangerous sections of the dam were identified in terms of the highest stress.", "classified_sentences": [ { "sentence": "The paper provides a detailed analysis of the current state of the problem.", "category": "background" }, { "sentence": "A mathematical model is presented to determine the dynamic behavior of earth dams, considering the viscoelastic properties of soil, using the hereditary Boltzmann-Volterra theory with the A.R.", "category": "method" }, { "sentence": "Rzhanitsyn kernel under periodic kinematic impacts.", "category": "method" }, { "sentence": "To solve the problems considered, the finite element method and complex arithmetics were used to reduce integrodifferential equations to a high-order complex algebraic equation.", "category": "method" }, { "sentence": "The accuracy of the methods was verified by solving test problems.", "category": "method" }, { "sentence": "Steadystate forced vibrations of the Pskem earth dam 195 m high are studied considering the real geometry and soil properties under resonant vibration modes.", "category": "result" }, { "sentence": "It was stated that the largest stress amplitudes in the body of the dam occur not only under the first resonance, but they can occur under other dense spectra of the eigenfrequencies of the dam, due to the interaction between close natural modes of vibration.", "category": "result" }, { "sentence": "The strength of various sections of the dam body was tested under kinematic impact using the Coulomb- Mohr theory of strength; the most dangerous sections of the dam were identified in terms of the highest stress.", "category": "result" } ] }, { "paper_id": "124628167", "title": "Changes in the form of short gravity waves on long waves and tidal currents", "abstract": "Short gravity waves, when superposed on much longer waves of the same type, have a tendency to become both shorter and steeper at the crests of the longer waves, and correspondingly longer and lower in the troughs. In the present paper, by taking into account the non-linear interactions between the two wave trains, the changes in wavelength and amplitude of the shorter wave train are rigorously calculated. The results differ in some essentials from previous estimates by Unna. The variation in energy of the short waves is shown to correspond to work done by the longer waves against the radiation stress of the short waves, which has previously been overlooked. The concept of the radiation stress is likely to be valuable in other problems.", "classified_sentences": [ { "sentence": "Short gravity waves, when superposed on much longer waves of the same type, have a tendency to become both shorter and steeper at the crests of the longer waves, and correspondingly longer and lower in the troughs.", "category": "background" }, { "sentence": "In the present paper, by taking into account the non-linear interactions between the two wave trains, the changes in wavelength and amplitude of the shorter wave train are rigorously calculated.", "category": "method" }, { "sentence": "The results differ in some essentials from previous estimates by Unna.", "category": "result" }, { "sentence": "The variation in energy of the short waves is shown to correspond to work done by the longer waves against the radiation stress of the short waves, which has previously been overlooked.", "category": "result" }, { "sentence": "The concept of the radiation stress is likely to be valuable in other problems.", "category": "result" } ] }, { "paper_id": "260447567", "title": "Binary Constrained Deep Hashing Network for Image Retrieval without Human Annotation", "abstract": "Learning compact binary codes for image retrieval problem using deep neural networks has attracted increasing attention recently. However, training deep hashing networks is challenging due to the binary constraints on the hash codes, the similarity preserving properties, and the requirement for a vast amount of labelled images. To the best of our knowledge, none of the existing methods has tackled all of these challenges completely in a unified framework. In this work, we propose a novel end-toend deep hashing approach, which is trained to produce binary codes directly from image pixels without human intervention. In particular, our main contribution is to propose a novel pairwise loss function, which simultaneously encodes the distances between pairs of binary codes, and the binary quantization error. We propose an efficient parameter learning algorithm for this loss function. In addition, to provide similar/dissimilar images for our pairwise loss function, we exploit 3D models reconstructed from unlabeled images for automatic generation of enormous similar/dissimilar pairs. Extensive experiments on three image retrieval benchmark datasets demonstrate the superior performance of the proposed method.", "classified_sentences": [ { "sentence": "Learning compact binary codes for image retrieval problem using deep neural networks has attracted increasing attention recently.", "category": "background" }, { "sentence": "However, training deep hashing networks is challenging due to the binary constraints on the hash codes, the similarity preserving properties, and the requirement for a vast amount of labelled images.", "category": "background" }, { "sentence": "To the best of our knowledge, none of the existing methods has tackled all of these challenges completely in a unified framework.", "category": "background" }, { "sentence": "In this work, we propose a novel end-toend deep hashing approach, which is trained to produce binary codes directly from image pixels without human intervention.", "category": "method" }, { "sentence": "In particular, our main contribution is to propose a novel pairwise loss function, which simultaneously encodes the distances between pairs of binary codes, and the binary quantization error.", "category": "method" }, { "sentence": "We propose an efficient parameter learning algorithm for this loss function.", "category": "method" }, { "sentence": "In addition, to provide similar/dissimilar images for our pairwise loss function, we exploit 3D models reconstructed from unlabeled images for automatic generation of enormous similar/dissimilar pairs.", "category": "method" }, { "sentence": "Extensive experiments on three image retrieval benchmark datasets demonstrate the superior performance of the proposed method.", "category": "result" } ] }, { "paper_id": "264147507", "title": "Antibiogram Profile and Detection of Resistance Genes in Pseudomonas aeruginosa Recovered from Hospital Wastewater Effluent", "abstract": "The nosocomial pathogen Pseudomonas aeruginosa (P. aeruginosa) is characterized by increased prevalence in hospital wastewater and is a public health concern. Untreated wastewater severely challenges human health when discharged into nearby aquatic ecosystems. The antibiogram profiles and resistance genes of P. aeruginosa were evaluated in this study. Wastewater effluents were obtained from a hospital within a six-month sampling period. After the samples were processed and analysed, P. aeruginosa was identified by polymerase chain reaction (PCR) by amplifying OprI and OprL genes. The Kirby–Bauer diffusion technique was employed to check the susceptibility profiles of P. aeruginosa which were further interpreted using CLSI guidelines. A total of 21 resistance genes were investigated among the isolates. The sum of 81 positive P. aeruginosa were isolated in this study. This study’s mean count of Pseudomonas aeruginosa ranged from 2.4 × 105 to 6.5 × 105 CFU/mL. A significant proportion of the isolates were susceptible to imipenem (93%), tobramycin (85%), norfloxacin (85%), aztreonam (70%), ciprofloxacin (51%), meropenem (47%), levofloxacin (43%), and gentamicin (40%). Meanwhile, a low susceptibility was recorded for amikacin and ceftazidime. The overall multiple antibiotics resistance index (MARI) ranged from 0.3 to 0.9, with 75% of the multidrug-resistant isolates. The assessment of β-lactam-resistant genes revealed blaOXA-1 (3.7%) and blaSHV (2.4%). The frequency of carbapenem genes was 6.6% for blaIMP, 6.6% for blaKPC, 6.6% for blaoxa-48, 2.2% for blaNDM-1, 2.2% for blaGES, and 2.2% for blaVIM. Of the aminoglycoside genes screened, 8.6% harboured strA, 11.5% harboured aadA, and 1.5% harboured aph(3)-Ia(aphA1). Only one non-β-lactamase gene (qnrA) was detected, with a prevalence of 4.9%. The findings of this study revealed a high prevalence of multidrug-resistant P. aeruginosa and resistance determinants potentially posing environmental health risks.", "classified_sentences": [ { "sentence": "The nosocomial pathogen Pseudomonas aeruginosa (P. aeruginosa) is characterized by increased prevalence in hospital wastewater and is a public health concern.", "category": "background" }, { "sentence": "Untreated wastewater severely challenges human health when discharged into nearby aquatic ecosystems.", "category": "background" }, { "sentence": "The antibiogram profiles and resistance genes of P. aeruginosa were evaluated in this study.", "category": "method" }, { "sentence": "Wastewater effluents were obtained from a hospital within a six-month sampling period.", "category": "method" }, { "sentence": "After the samples were processed and analysed, P. aeruginosa was identified by polymerase chain reaction (PCR) by amplifying OprI and OprL genes.", "category": "method" }, { "sentence": "The Kirby–Bauer diffusion technique was employed to check the susceptibility profiles of P. aeruginosa which were further interpreted using CLSI guidelines.", "category": "method" }, { "sentence": "A total of 21 resistance genes were investigated among the isolates.", "category": "method" }, { "sentence": "The sum of 81 positive P. aeruginosa were isolated in this study.", "category": "result" }, { "sentence": "This study’s mean count of Pseudomonas aeruginosa ranged from 2.4 × 105 to 6.5 × 105 CFU/mL.", "category": "result" }, { "sentence": "A significant proportion of the isolates were susceptible to imipenem (93%), tobramycin (85%), norfloxacin (85%), aztreonam (70%), ciprofloxacin (51%), meropenem (47%), levofloxacin (43%), and gentamicin (40%).", "category": "result" }, { "sentence": "Meanwhile, a low susceptibility was recorded for amikacin and ceftazidime.", "category": "result" }, { "sentence": "The overall multiple antibiotics resistance index (MARI) ranged from 0.3 to 0.9, with 75% of the multidrug-resistant isolates.", "category": "result" }, { "sentence": "The assessment of β-lactam-resistant genes revealed blaOXA-1 (3.7%) and blaSHV (2.4%).", "category": "result" }, { "sentence": "The frequency of carbapenem genes was 6.6% for blaIMP, 6.6% for blaKPC, 6.6% for blaoxa-48, 2.2% for blaNDM-1, 2.2% for blaGES, and 2.2% for blaVIM.", "category": "result" }, { "sentence": "Of the aminoglycoside genes screened, 8.6% harboured strA, 11.5% harboured aadA, and 1.5% harboured aph(3)-Ia(aphA1).", "category": "result" }, { "sentence": "Only one non-β-lactamase gene (qnrA) was detected, with a prevalence of 4.9%.", "category": "result" }, { "sentence": "The findings of this study revealed a high prevalence of multidrug-resistant P. aeruginosa and resistance determinants potentially posing environmental health risks.", "category": "result" } ] }, { "paper_id": "265313972", "title": "Binary Neural Networks in FPGAs: Architectures, Tool Flows and Hardware Comparisons", "abstract": "Binary neural networks (BNNs) are variations of artificial/deep neural network (ANN/DNN) architectures that constrain the real values of weights to the binary set of numbers {−1,1}. By using binary values, BNNs can convert matrix multiplications into bitwise operations, which accelerates both training and inference and reduces hardware complexity and model sizes for implementation. Compared to traditional deep learning architectures, BNNs are a good choice for implementation in resource-constrained devices like FPGAs and ASICs. However, BNNs have the disadvantage of reduced performance and accuracy because of the tradeoff due to binarization. Over the years, this has attracted the attention of the research community to overcome the performance gap of BNNs, and several architectures have been proposed. In this paper, we provide a comprehensive review of BNNs for implementation in FPGA hardware. The survey covers different aspects, such as BNN architectures and variants, design and tool flows for FPGAs, and various applications for BNNs. The final part of the paper gives some benchmark works and design tools for implementing BNNs in FPGAs based on established datasets used by the research community.", "classified_sentences": [ { "sentence": "Binary neural networks (BNNs) are variations of artificial/deep neural network (ANN/DNN) architectures that constrain the real values of weights to the binary set of numbers {−1,1}.", "category": "background" }, { "sentence": "By using binary values, BNNs can convert matrix multiplications into bitwise operations, which accelerates both training and inference and reduces hardware complexity and model sizes for implementation.", "category": "background" }, { "sentence": "Compared to traditional deep learning architectures, BNNs are a good choice for implementation in resource-constrained devices like FPGAs and ASICs.", "category": "background" }, { "sentence": "However, BNNs have the disadvantage of reduced performance and accuracy because of the tradeoff due to binarization.", "category": "background" }, { "sentence": "Over the years, this has attracted the attention of the research community to overcome the performance gap of BNNs, and several architectures have been proposed.", "category": "background" }, { "sentence": "In this paper, we provide a comprehensive review of BNNs for implementation in FPGA hardware.", "category": "method" }, { "sentence": "The survey covers different aspects, such as BNN architectures and variants, design and tool flows for FPGAs, and various applications for BNNs.", "category": "method" }, { "sentence": "The final part of the paper gives some benchmark works and design tools for implementing BNNs in FPGAs based on established datasets used by the research community.", "category": "method" } ] }, { "paper_id": "268319153", "title": "Simulation Study on Temperature and Stress Fields in Mg-Gd-Y-Zn-Zr Alloy during CMT Additive Manufacturing Process", "abstract": "A new heat source combination, consisting of a uniform body heat source and a tilted double ellipsoidal heat source, has been developed for cold metal transfer (CMT) wire-arc additive manufacturing of Mg-Gd-Y-Zn-Zr alloy. Simulations were conducted to analyze the temperature field and stress distribution during the process. The optimal combination of feeding speed and welding speed was found to be 8 m/min and 8 mm/s, respectively, resulting in the lowest thermal accumulation and residual stress. Z-axis residual stress was identified as the main component of residual stress. Electron Backscatter Diffraction (EBSD) testing showed weak texture strength, and Kernel Average Misorientation (KAM) analysis revealed that the 1st layer had the highest residual stress, while the 11th layer had higher residual stress than the 6th layer. Microhardness in the 1st, 11th, and 6th layers varies due to residual stress impacts on dislocation density. Higher residual stress increases dislocation density, raising microhardness in components. The experimental results were highly consistent with the simulated results.", "classified_sentences": [ { "sentence": "A new heat source combination, consisting of a uniform body heat source and a tilted double ellipsoidal heat source, has been developed for cold metal transfer (CMT) wire-arc additive manufacturing of Mg-Gd-Y-Zn-Zr alloy.", "category": "method" }, { "sentence": "Simulations were conducted to analyze the temperature field and stress distribution during the process.", "category": "method" }, { "sentence": "The optimal combination of feeding speed and welding speed was found to be 8 m/min and 8 mm/s, respectively, resulting in the lowest thermal accumulation and residual stress.", "category": "result" }, { "sentence": "Z-axis residual stress was identified as the main component of residual stress.", "category": "result" }, { "sentence": "Electron Backscatter Diffraction (EBSD) testing showed weak texture strength, and Kernel Average Misorientation (KAM) analysis revealed that the 1st layer had the highest residual stress, while the 11th layer had higher residual stress than the 6th layer.", "category": "result" }, { "sentence": "Microhardness in the 1st, 11th, and 6th layers varies due to residual stress impacts on dislocation density.", "category": "result" }, { "sentence": "Higher residual stress increases dislocation density, raising microhardness in components.", "category": "result" }, { "sentence": "The experimental results were highly consistent with the simulated results.", "category": "result" } ] }, { "paper_id": "233072", "title": "A Deep Hashing Learning Network", "abstract": "Hashing-based methods seek compact and efficient binary codes that preserve the neighborhood structure in the original data space. For most existing hashing methods, an image is first encoded as a vector of hand-crafted visual feature, followed by a hash projection and quantization step to get the compact binary vector. Most of the hand-crafted features just encode the low-level information of the input, the feature may not preserve the semantic similarities of images pairs. Meanwhile, the hashing function learning process is independent with the feature representation, so the feature may not be optimal for the hashing projection. In this paper, we propose a supervised hashing method based on a well designed deep convolutional neural network, which tries to learn hashing code and compact representations of data simultaneously. The proposed model learn the binary codes by adding a compact sigmoid layer before the loss layer. Experiments on several image data sets show that the proposed model outperforms other state-of-the-art methods.", "classified_sentences": [ { "sentence": "Hashing-based methods seek compact and efficient binary codes that preserve the neighborhood structure in the original data space.", "category": "background" }, { "sentence": "For most existing hashing methods, an image is first encoded as a vector of hand-crafted visual feature, followed by a hash projection and quantization step to get the compact binary vector.", "category": "background" }, { "sentence": "Most of the hand-crafted features just encode the low-level information of the input, the feature may not preserve the semantic similarities of images pairs.", "category": "background" }, { "sentence": "Meanwhile, the hashing function learning process is independent with the feature representation, so the feature may not be optimal for the hashing projection.", "category": "background" }, { "sentence": "In this paper, we propose a supervised hashing method based on a well designed deep convolutional neural network, which tries to learn hashing code and compact representations of data simultaneously.", "category": "method" }, { "sentence": "The proposed model learn the binary codes by adding a compact sigmoid layer before the loss layer.", "category": "method" }, { "sentence": "Experiments on several image data sets show that the proposed model outperforms other state-of-the-art methods.", "category": "result" } ] }, { "paper_id": "350786", "title": "Visual Saliency Based on Scale-Space Analysis in the Frequency Domain", "abstract": "We address the issue of visual saliency from three perspectives. First, we consider saliency detection as a frequency domain analysis problem. Second, we achieve this by employing the concept of nonsaliency. Third, we simultaneously consider the detection of salient regions of different size. The paper proposes a new bottom-up paradigm for detecting visual saliency, characterized by a scale-space analysis of the amplitude spectrum of natural images. We show that the convolution of the image amplitude spectrum with a low-pass Gaussian kernel of an appropriate scale is equivalent to an image saliency detector. The saliency map is obtained by reconstructing the 2D signal using the original phase and the amplitude spectrum, filtered at a scale selected by minimizing saliency map entropy. A Hypercomplex Fourier Transform performs the analysis in the frequency domain. Using available databases, we demonstrate experimentally that the proposed model can predict human fixation data. We also introduce a new image database and use it to show that the saliency detector can highlight both small and large salient regions, as well as inhibit repeated distractors in cluttered images. In addition, we show that it is able to predict salient regions on which people focus their attention.", "classified_sentences": [ { "sentence": "We address the issue of visual saliency from three perspectives.", "category": "background" }, { "sentence": "First, we consider saliency detection as a frequency domain analysis problem.", "category": "method" }, { "sentence": "Second, we achieve this by employing the concept of nonsaliency.", "category": "method" }, { "sentence": "Third, we simultaneously consider the detection of salient regions of different size.", "category": "method" }, { "sentence": "The paper proposes a new bottom-up paradigm for detecting visual saliency, characterized by a scale-space analysis of the amplitude spectrum of natural images.", "category": "method" }, { "sentence": "We show that the convolution of the image amplitude spectrum with a low-pass Gaussian kernel of an appropriate scale is equivalent to an image saliency detector.", "category": "method" }, { "sentence": "The saliency map is obtained by reconstructing the 2D signal using the original phase and the amplitude spectrum, filtered at a scale selected by minimizing saliency map entropy.", "category": "method" }, { "sentence": "A Hypercomplex Fourier Transform performs the analysis in the frequency domain.", "category": "method" }, { "sentence": "Using available databases, we demonstrate experimentally that the proposed model can predict human fixation data.", "category": "result" }, { "sentence": "We also introduce a new image database and use it to show that the saliency detector can highlight both small and large salient regions, as well as inhibit repeated distractors in cluttered images.", "category": "result" }, { "sentence": "In addition, we show that it is able to predict salient regions on which people focus their attention.", "category": "result" } ] }, { "paper_id": "941594", "title": "Predicting Success in Goal-Driven Human-Human Dialogues", "abstract": "In goal-driven dialogue systems, success is often defined based on a structured definition of the goal. This requires that the dialogue system be constrained to handle a specific class of goals and that there be a mechanism to measure success with respect to that goal. However, in many human-human dialogues the diversity of goals makes it infeasible to define success in such a way. To address this scenario, we consider the task of automatically predicting success in goal-driven human-human dialogues using only the information communicated between participants in the form of text. We build a dataset from stackoverflow.com which consists of exchanges between two users in the technical domain where ground-truth success labels are available. We then propose a turn-based hierarchical neural network model that can be used to predict success without requiring a structured goal definition. We show this model outperforms rule-based heuristics and other baselines as it is able to detect patterns over the course of a dialogue and capture notions such as gratitude.", "classified_sentences": [ { "sentence": "In goal-driven dialogue systems, success is often defined based on a structured definition of the goal.", "category": "background" }, { "sentence": "This requires that the dialogue system be constrained to handle a specific class of goals and that there be a mechanism to measure success with respect to that goal.", "category": "background" }, { "sentence": "However, in many human-human dialogues the diversity of goals makes it infeasible to define success in such a way.", "category": "background" }, { "sentence": "To address this scenario, we consider the task of automatically predicting success in goal-driven human-human dialogues using only the information communicated between participants in the form of text.", "category": "method" }, { "sentence": "We build a dataset from stackoverflow.com which consists of exchanges between two users in the technical domain where ground-truth success labels are available.", "category": "method" }, { "sentence": "We then propose a turn-based hierarchical neural network model that can be used to predict success without requiring a structured goal definition.", "category": "method" }, { "sentence": "We show this model outperforms rule-based heuristics and other baselines as it is able to detect patterns over the course of a dialogue and capture notions such as gratitude.", "category": "result" } ] }, { "paper_id": "1549343", "title": "The Revised Arabic PropBank", "abstract": "The revised Arabic PropBank (APB) reflects a number of changes to the data and the process of PropBanking. Several changes stem from Treebank revisions. An automatic process was put in place to map existing annotation to the new trees. We have revised the original 493 Frame Files from the Pilot APB and added 1462 new files for a total of 1955 Frame Files with 2446 framesets. In addition to a heightened attention to sense distinctions this cycle includes a greater attempt to address complicated predicates such as light verb constructions and multi-word expressions. New tools facilitate the data tagging and also simplify frame creation.", "classified_sentences": [ { "sentence": "The revised Arabic PropBank (APB) reflects a number of changes to the data and the process of PropBanking.", "category": "background" }, { "sentence": "Several changes stem from Treebank revisions.", "category": "background" }, { "sentence": "An automatic process was put in place to map existing annotation to the new trees.", "category": "method" }, { "sentence": "We have revised the original 493 Frame Files from the Pilot APB and added 1462 new files for a total of 1955 Frame Files with 2446 framesets.", "category": "result" }, { "sentence": "In addition to a heightened attention to sense distinctions this cycle includes a greater attempt to address complicated predicates such as light verb constructions and multi-word expressions.", "category": "method" }, { "sentence": "New tools facilitate the data tagging and also simplify frame creation.", "category": "method" } ] }, { "paper_id": "1976080", "title": "The improved two-dimensional Gabor filter based interest objects detection", "abstract": "Saliency using natural statistics model has a good effect on obtaining the region of interest. However, it still has limitations such as insufficient image contrast, inability to cope with local information and so on. In order to overcome these challenges, an improved algorithm based on the two-dimensional Gabor filter and color maps is proposed. The two-dimensional Gabor filter has a close approach to simple receptive fields in human striate cortex. It is used to tackle the problem mentioned above. Experiments have demonstrated the feasibility and effectiveness of the proposed algorithm for interest objects detection.", "classified_sentences": [ { "sentence": "Saliency using natural statistics model has a good effect on obtaining the region of interest.", "category": "background" }, { "sentence": "However, it still has limitations such as insufficient image contrast, inability to cope with local information and so on.", "category": "background" }, { "sentence": "In order to overcome these challenges, an improved algorithm based on the two-dimensional Gabor filter and color maps is proposed.", "category": "method" }, { "sentence": "The two-dimensional Gabor filter has a close approach to simple receptive fields in human striate cortex.", "category": "method" }, { "sentence": "It is used to tackle the problem mentioned above.", "category": "method" }, { "sentence": "Experiments have demonstrated the feasibility and effectiveness of the proposed algorithm for interest objects detection.", "category": "result" } ] }, { "paper_id": "2633904", "title": "Add-On Strategies for Fine-Grained Pedestrian Classification", "abstract": "In this paper, we present four add-on strategies for the fine-grained pedestrian classification task. These strategies are: (1) super-resolution based image preprocessing, which helps to recover the image details; (2) patch dividing based deep feature extraction, which extracts features in a way that preserves the spatial layout of input images; (3) pose- wise classifier sharing, which learns robust classifiers and makes robust predictions using pose information; and (4) graphical model based inference, which utilizes the interdependence between different subcategories to update raw estimations. The proposed strategies are independent and flexible, which make it easy to implement them in practice. We evaluated these strategies on the CRP dataset and confirmed that all of them lead to improvements over the baseline. We also confirmed an improvement over the state-of-the-art when all strategies are combined together.", "classified_sentences": [ { "sentence": "In this paper, we present four add-on strategies for the fine-grained pedestrian classification task.", "category": "method" }, { "sentence": "These strategies are: (1) super-resolution based image preprocessing, which helps to recover the image details; (2) patch dividing based deep feature extraction, which extracts features in a way that preserves the spatial layout of input images; (3) pose- wise classifier sharing, which learns robust classifiers and makes robust predictions using pose information; and (4) graphical model based inference, which utilizes the interdependence between different subcategories to update raw estimations.", "category": "method" }, { "sentence": "The proposed strategies are independent and flexible, which make it easy to implement them in practice.", "category": "method" }, { "sentence": "We evaluated these strategies on the CRP dataset and confirmed that all of them lead to improvements over the baseline.", "category": "result" }, { "sentence": "We also confirmed an improvement over the state-of-the-art when all strategies are combined together.", "category": "result" } ] }, { "paper_id": "6740820", "title": "Interpretation of Rank Histograms for Verifying Ensemble Forecasts", "abstract": "Abstract Rank histograms are a tool for evaluating ensemble forecasts. They are useful for determining the reliability of ensemble forecasts and for diagnosing errors in its mean and spread. Rank histograms are generated by repeatedly tallying the rank of the verification (usually an observation) relative to values from an ensemble sorted from lowest to highest. However, an uncritical use of the rank histogram can lead to misinterpretations of the qualities of that ensemble. For example, a flat rank histogram, usually taken as a sign of reliability, can still be generated from unreliable ensembles. Similarly, a U-shaped rank histogram, commonly understood as indicating a lack of variability in the ensemble, can also be a sign of conditional bias. It is also shown that flat rank histograms can be generated for some model variables if the variance of the ensemble is correctly specified, yet if covariances between model grid points are improperly specified, rank histograms for combinations of model variables.", "classified_sentences": [ { "sentence": "Abstract Rank histograms are a tool for evaluating ensemble forecasts.", "category": "background" }, { "sentence": "They are useful for determining the reliability of ensemble forecasts and for diagnosing errors in its mean and spread.", "category": "background" }, { "sentence": "Rank histograms are generated by repeatedly tallying the rank of the verification (usually an observation) relative to values from an ensemble sorted from lowest to highest.", "category": "method" }, { "sentence": "However, an uncritical use of the rank histogram can lead to misinterpretations of the qualities of that ensemble.", "category": "background" }, { "sentence": "For example, a flat rank histogram, usually taken as a sign of reliability, can still be generated from unreliable ensembles.", "category": "background" }, { "sentence": "Similarly, a U-shaped rank histogram, commonly understood as indicating a lack of variability in the ensemble, can also be a sign of conditional bias.", "category": "background" }, { "sentence": "It is also shown that flat rank histograms can be generated for some model variables if the variance of the ensemble is correctly specified, yet if covariances between model grid points are improperly specified, rank histograms for combinations of model variables.", "category": "result" } ] }, { "paper_id": "7860810", "title": "Fast and Accurate Preordering for SMT using Neural Networks", "abstract": "We propose the use of neural networks to model source-side preordering for faster and better statistical machine translation. The neural network trains a logistic regression model to predict whether two sibling nodes of the source-side parse tree should be swapped in order to obtain a more monotonic parallel corpus, based on samples extracted from the word-aligned parallel corpus. For multiple language pairs and domains, we show that this yields the best reordering performance against other state-of-the-art techniques, resulting in improved translation quality and very fast decoding.", "classified_sentences": [ { "sentence": "We propose the use of neural networks to model source-side preordering for faster and better statistical machine translation.", "category": "method" }, { "sentence": "The neural network trains a logistic regression model to predict whether two sibling nodes of the source-side parse tree should be swapped in order to obtain a more monotonic parallel corpus, based on samples extracted from the word-aligned parallel corpus.", "category": "method" }, { "sentence": "For multiple language pairs and domains, we show that this yields the best reordering performance against other state-of-the-art techniques, resulting in improved translation quality and very fast decoding.", "category": "result" } ] }, { "paper_id": "8534821", "title": "Nonparametric bottom-up saliency detection by self-resemblance", "abstract": "We present a novel bottom-up saliency detection algorithm. Our method computes so-called local regression kernels (i.e., local features) from the given image, which measure the likeness of a pixel to its surroundings. Visual saliency is then computed using the said “self-resemblance” measure. The framework results in a saliency map where each pixel indicates the statistical likelihood of saliency of a feature matrix given its surrounding feature matrices. As a similarity measure, matrix cosine similarity (a generalization of cosine similarity) is employed. State of the art performance is demonstrated on commonly used human eye fixation data [3] and some psychological patterns.", "classified_sentences": [ { "sentence": "We present a novel bottom-up saliency detection algorithm.", "category": "method" }, { "sentence": "Our method computes so-called local regression kernels (i.e., local features) from the given image, which measure the likeness of a pixel to its surroundings.", "category": "method" }, { "sentence": "Visual saliency is then computed using the said “self-resemblance” measure.", "category": "method" }, { "sentence": "The framework results in a saliency map where each pixel indicates the statistical likelihood of saliency of a feature matrix given its surrounding feature matrices.", "category": "result" }, { "sentence": "As a similarity measure, matrix cosine similarity (a generalization of cosine similarity) is employed.", "category": "method" }, { "sentence": "State of the art performance is demonstrated on commonly used human eye fixation data [3] and some psychological patterns.", "category": "result" } ] }, { "paper_id": "10168768", "title": "A Large Coverage Verb Taxonomy for Arabic", "abstract": "In this article I present a lexicon for Arabic verbs which exploits Levin’s verb-classes (Levin, 1993) and the basic development procedure used by (Schuler, 2005). The verb lexicon in its current state has 173 classes which contain 4392 verbs and 498 frames providing information about verb root, the deverbal form of the verb, the participle, thematic roles, subcategorisation frames and syntactic and semantic descriptions of each verb. The taxonomy is available in XML format. It can be ported to MYSQL, YAML or JSON and accessed either in Arabic characters or in the Buckwalter transliteration.", "classified_sentences": [ { "sentence": "In this article I present a lexicon for Arabic verbs which exploits Levin’s verb-classes (Levin, 1993) and the basic development procedure used by (Schuler, 2005).", "category": "method" }, { "sentence": "The verb lexicon in its current state has 173 classes which contain 4392 verbs and 498 frames providing information about verb root, the deverbal form of the verb, the participle, thematic roles, subcategorisation frames and syntactic and semantic descriptions of each verb.", "category": "result" }, { "sentence": "The taxonomy is available in XML format.", "category": "result" }, { "sentence": "It can be ported to MYSQL, YAML or JSON and accessed either in Arabic characters or in the Buckwalter transliteration.", "category": "method" } ] }, { "paper_id": "11289341", "title": "Spherical hashing", "abstract": "Many binary code encoding schemes based on hashing have been actively studied recently, since they can provide efficient similarity search, especially nearest neighbor search, and compact data representations suitable for handling large scale image databases in many computer vision problems. Existing hashing techniques encode high-dimensional data points by using hyperplane-based hashing functions. In this paper we propose a novel hypersphere-based hashing function, spherical hashing, to map more spatially coherent data points into a binary code compared to hyperplane-based hashing functions. Furthermore, we propose a new binary code distance function, spherical Hamming distance, that is tailored to our hypersphere-based binary coding scheme, and design an efficient iterative optimization process to achieve balanced partitioning of data points for each hash function and independence between hashing functions. Our extensive experiments show that our spherical hashing technique significantly outperforms six state-of-the-art hashing techniques based on hyperplanes across various image benchmarks of sizes ranging from one to 75 million of GIST descriptors. The performance gains are consistent and large, up to 100% improvements. The excellent results confirm the unique merits of the proposed idea in using hyperspheres to encode proximity regions in high-dimensional spaces. Finally, our method is intuitive and easy to implement.", "classified_sentences": [ { "sentence": "Many binary code encoding schemes based on hashing have been actively studied recently, since they can provide efficient similarity search, especially nearest neighbor search, and compact data representations suitable for handling large scale image databases in many computer vision problems.", "category": "background" }, { "sentence": "Existing hashing techniques encode high-dimensional data points by using hyperplane-based hashing functions.", "category": "background" }, { "sentence": "In this paper we propose a novel hypersphere-based hashing function, spherical hashing, to map more spatially coherent data points into a binary code compared to hyperplane-based hashing functions.", "category": "method" }, { "sentence": "Furthermore, we propose a new binary code distance function, spherical Hamming distance, that is tailored to our hypersphere-based binary coding scheme, and design an efficient iterative optimization process to achieve balanced partitioning of data points for each hash function and independence between hashing functions.", "category": "method" }, { "sentence": "Our extensive experiments show that our spherical hashing technique significantly outperforms six state-of-the-art hashing techniques based on hyperplanes across various image benchmarks of sizes ranging from one to 75 million of GIST descriptors.", "category": "result" }, { "sentence": "The performance gains are consistent and large, up to 100% improvements.", "category": "result" }, { "sentence": "The excellent results confirm the unique merits of the proposed idea in using hyperspheres to encode proximity regions in high-dimensional spaces.", "category": "result" }, { "sentence": "Finally, our method is intuitive and easy to implement.", "category": "method" } ] }, { "paper_id": "11412140", "title": "Image Modification Based on a Visual Saliency Map for Guiding Visual Attention", "abstract": "SUMMARY It is commonly believed that improved interaction be-tween humans and electronic device, it is e ff ective to draw the viewer’s attention to a particular object. Augmented reality (AR) applications can call attention to real objects by overlaying highlight e ff ects or visual stimuli (such as arrows) on a physical scene. Sometimes, more subtle e ff ects would be desirable, in which case it would be necessary to smoothly and naturally guide the user’s gaze without external stimuli. Here, a novel image modifi-cation method is proposed for directing a viewer’s gaze to specific regions of interest. The proposed method uses saliency analysis and color modulation to create modified images in which the region of interest is the most salient region in the entire image. The proposed saliency map model that is used during saliency analysis reduces computational costs and improves the naturalness of the image using the LAB color space and simplified normalization. During color modulation, the modulation value of each LAB component is determined in order to consider the relationship between the LAB components and the saliency value. With the image obtained in this manner, the viewer’s attention is smoothly attracted to a specific region very naturally. Gaze measurements as well as a subjective experiments were conducted to prove the e ff ectiveness of the proposed method. These results show that a viewer’s visual attention is indeed attracted toward the specified region without any sense of discomfort or disruption when the proposed method is used.", "classified_sentences": [ { "sentence": "SUMMARY It is commonly believed that improved interaction be-tween humans and electronic device, it is e ff ective to draw the viewer’s attention to a particular object.", "category": "background" }, { "sentence": "Augmented reality (AR) applications can call attention to real objects by overlaying highlight e ff ects or visual stimuli (such as arrows) on a physical scene.", "category": "background" }, { "sentence": "Sometimes, more subtle e ff ects would be desirable, in which case it would be necessary to smoothly and naturally guide the user’s gaze without external stimuli.", "category": "background" }, { "sentence": "Here, a novel image modifi-cation method is proposed for directing a viewer’s gaze to specific regions of interest.", "category": "method" }, { "sentence": "The proposed method uses saliency analysis and color modulation to create modified images in which the region of interest is the most salient region in the entire image.", "category": "method" }, { "sentence": "The proposed saliency map model that is used during saliency analysis reduces computational costs and improves the naturalness of the image using the LAB color space and simplifified normalization.", "category": "method" }, { "sentence": "During color modulation, the modulation value of each LAB component is determined in order to consider the relationship between the LAB components and the saliency value.", "category": "method" }, { "sentence": "With the image obtained in this manner, the viewer’s attention is smoothly attracted to a specific region very naturally.", "category": "method" }, { "sentence": "Gaze measurements as well as a subjective experiments were conducted to prove the e ff ectiveness of the proposed method.", "category": "result" }, { "sentence": "These results show that a viewer’s visual attention is indeed attracted toward the specified region without any sense of discomfort or disruption when the proposed method is used.", "category": "result" } ] }, { "paper_id": "11588315", "title": "English-Czech Machine Translation Using", "abstract": "English to Czech machine translation as it is implemented in the TectoMT system consists of three phases: analysis, transfer and synthesis. The system uses tectogrammatical (deep-syntactic dependency) trees as the transfer medium. Each phase is divided into so-called blocks, which are processing units that solve linguistically interpretable tasks (e.g., statistical part-of-speech tagging or rule-based placement of clitics). This paper shortly introduces linguistic layers of language description which are used for the translation and describes basic concepts of the TectoMT framework. The translation results are evaluated using both automatic metric BLEU and human judgments from the WMT 2010 evaluation.", "classified_sentences": [ { "sentence": "English to Czech machine translation as it is implemented in the TectoMT system consists of three phases: analysis, transfer and synthesis.", "category": "background" }, { "sentence": "The system uses tectogrammatical (deep-syntactic dependency) trees as the transfer medium.", "category": "method" }, { "sentence": "Each phase is divided into so-called blocks, which are processing units that solve linguistically interpretable tasks (e.g., statistical part-of-speech tagging or rule-based placement of clitics).", "category": "method" }, { "sentence": "This paper shortly introduces linguistic layers of language description which are used for the translation and describes basic concepts of the TectoMT framework.", "category": "background" }, { "sentence": "The translation results are evaluated using both automatic metric BLEU and human judgments from the WMT 2010 evaluation.", "category": "result" } ] }, { "paper_id": "12418427", "title": "Multi-Tagging for Lexicalized-Grammar Parsing", "abstract": "With performance above 97% accuracy for newspaper text, part of speech (POS) tagging might be considered a solved problem. Previous studies have shown that allowing the parser to resolve POS tag ambiguity does not improve performance. However, for grammar formalisms which use more fine-grained grammatical categories, for example TAG and CCG, tagging accuracy is much lower. In fact, for these formalisms, premature ambiguity resolution makes parsing infeasible.We describe a multi-tagging approach which maintains a suitable level of lexical category ambiguity for accurate and efficient CCG parsing. We extend this multi-tagging approach to the POS level to overcome errors introduced by automatically assigned POS tags. Although POS tagging accuracy seems high, maintaining some POS tag ambiguity in the language processing pipeline results in more accurate CCG supertagging.", "classified_sentences": [ { "sentence": "With performance above 97% accuracy for newspaper text, part of speech (POS) tagging might be considered a solved problem.", "category": "background" }, { "sentence": "Previous studies have shown that allowing the parser to resolve POS tag ambiguity does not improve performance.", "category": "background" }, { "sentence": "However, for grammar formalisms which use more fine-grained grammatical categories, for example TAG and CCG, tagging accuracy is much lower.", "category": "background" }, { "sentence": "In fact, for these formalisms, premature ambiguity resolution makes parsing infeasible.We describe a multi-tagging approach which maintains a suitable level of lexical category ambiguity for accurate and efficient CCG parsing.", "category": "method" }, { "sentence": "We extend this multi-tagging approach to the POS level to overcome errors introduced by automatically assigned POS tags.", "category": "method" }, { "sentence": "Although POS tagging accuracy seems high, maintaining some POS tag ambiguity in the language processing pipeline results in more accurate CCG supertagging.", "category": "result" } ] }, { "paper_id": "12894551", "title": "An Ensemble Diversity Approach to Binary Hashing ❦", "abstract": "Introduction Information retrieval tasks such as searching for a query image or document in a database are essentially a nearest-neighbor search. When the dimensionality of the query and the size of the database is large, approximate search is necessary. We focus on binary hashing, where the query and database are mapped onto low-dimensional binary vectors, where the search is performed. This has two speedups: computing Hamming distances (with hardware support) is much faster than computing distances between high-dimensional floating-point vectors; and the entire database becomes much smaller, so it may reside in fast memory rather than disk (for example, a database of 1 billion real vectors of dimension 500 takes 2 TB in floating point but 8 GB as 64-bit codes).", "classified_sentences": [ { "sentence": "Introduction Information retrieval tasks such as searching for a query image or document in a database are essentially a nearest-neighbor search.", "category": "background" }, { "sentence": "When the dimensionality of the query and the size of the database is large, approximate search is necessary.", "category": "background" }, { "sentence": "We focus on binary hashing, where the query and database are mapped onto low-dimensional binary vectors, where the search is performed.", "category": "method" }, { "sentence": "This has two speedups: computing Hamming distances (with hardware support) is much faster than computing distances between high-dimensional floating-point vectors; and the entire database becomes much smaller, so it may reside in fast memory rather than disk (for example, a database of 1 billion real vectors of dimension 500 takes 2 TB in floating point but 8 GB as 64-bit codes).", "category": "result" } ] }, { "paper_id": "13938168", "title": "Domain adaptation for parsing in automatic speech recognition", "abstract": "This paper addresses the problem of adapting a parser trained on out-of-domain data for use in automatic speech recognition (ASR) rescoring and error detection tasks. Using a self-training approach and adaptation with weakly-supervised data, we obtain improvements in ASR rescoring of confusion networks. Features extracted from the parser output are also used to improve detection of general ASR errors and out-of-vocabulary word regions in conjunction with a maximum entropy classifier.", "classified_sentences": [ { "sentence": "This paper addresses the problem of adapting a parser trained on out-of-domain data for use in automatic speech recognition (ASR) rescoring and error detection tasks.", "category": "background" }, { "sentence": "Using a self-training approach and adaptation with weakly-supervised data, we obtain improvements in ASR rescoring of confusion networks.", "category": "method" }, { "sentence": "Features extracted from the parser output are also used to improve detection of general ASR errors and out-of-vocabulary word regions in conjunction with a maximum entropy classifier.", "category": "method" } ] }, { "paper_id": "14068050", "title": "Exploiting Randomized Prim’s Algorithm and Background Contrast forSaliency Detection", "abstract": "The geodesic saliency method in the literature was based on the boundary and connectivity priority, which as- sumed that most of the background regions can touch the image boundaries. It cannot deal with the images with complex backgrounds or variant textures. To address such problem, we propose an improved saliency detection method by involv- ing the important foreground priority. First, the statistical results of randomized Prim's algorithm are used to generate a coarse conspicuity map, which aims to roughly estimate the potential foreground. Then, the image is over-segmented into some individual superpixels and an affinity propagation clustering method is used to group the superpixels having a simi- lar color appearance together. This is followed by the foreground probability map computation through the spatial interac- tion information between the coarse conspicuity map and superpixel based color clusters. The final saliency map is gener- ated by integrating the above foreground probability map and background color contrast in a unified way. The quantitative and qualitative comparisons on the benchmark dataset MSRA-1000 and SED show that our method outperforms many re- cent proposed state-of-the-art approaches significantly.", "classified_sentences": [ { "sentence": "The geodesic saliency method in the literature was based on the boundary and connectivity priority, which as- sumed that most of the background regions can touch the image boundaries.", "category": "background" }, { "sentence": "It cannot deal with the images with complex backgrounds or variant textures.", "category": "background" }, { "sentence": "To address such problem, we propose an improved saliency detection method by involv- ing the important foreground priority.", "category": "method" }, { "sentence": "First, the statistical results of randomized Prim's algorithm are used to generate a coarse conspicuity map, which aims to roughly estimate the potential foreground.", "category": "method" }, { "sentence": "Then, the image is over-segmented into some individual superpixels and an affinity propagation clustering method is used to group the superpixels having a simi- lar color appearance together.", "category": "method" }, { "sentence": "This is followed by the foreground probability map computation through the spatial interac- tion information between the coarse conspicuity map and superpixel based color clusters.", "category": "method" }, { "sentence": "The final saliency map is gener- ated by integrating the above foreground probability map and background color contrast in a unified way.", "category": "method" }, { "sentence": "The quantitative and qualitative comparisons on the benchmark dataset MSRA-1000 and SED show that our method outperforms many re- cent proposed state-of-the-art approaches significantly.", "category": "result" } ] }, { "paper_id": "14120418", "title": "Topic-Aware Deep Compositional Models for Sentence Classification", "abstract": "In recent years, deep compositional models have emerged as a popular technique for representation learning of sentence in computational linguistic and natural language processing. These models normally train various forms of neural networks on top of pretrained word embeddings using a task-specific corpus. However, most of these works neglect the multisense nature of words in the pretrained word embeddings. In this paper we introduce topic models to enrich the word embeddings for multisenses of words. The integration of the topic model with various semantic compositional processes leads to topic-aware convolutional neural network and topic-aware long short term memory networks. Different from previous multisense word embeddings models that assign multiple independent and sense-specific embeddings to each word, our proposed models are lightweight and have flexible frameworks that regard word sense as the composition of two parts: a general sense derived from a large corpus and a topic-specific sense derived from a task-specific corpus. In addition, our proposed models focus on semantic composition instead of word understanding. With the help of topic models, we can integrate the topic-specific sense at word-level before the composition and sentence-level after the composition. Comprehensive experiments on five public sentence classification datasets are conducted and the results show that our proposed topic-aware deep compositional models produce competitive or better performance than other text representation learning methods.", "classified_sentences": [ { "sentence": "In recent years, deep compositional models have emerged as a popular technique for representation learning of sentence in computational linguistic and natural language processing.", "category": "background" }, { "sentence": "These models normally train various forms of neural networks on top of pretrained word embeddings using a task-specific corpus.", "category": "background" }, { "sentence": "However, most of these works neglect the multisense nature of words in the pretrained word embeddings.", "category": "background" }, { "sentence": "In this paper we introduce topic models to enrich the word embeddings for multisenses of words.", "category": "method" }, { "sentence": "The integration of the topic model with various semantic compositional processes leads to topic-aware convolutional neural network and topic-aware long short term memory networks.", "category": "method" }, { "sentence": "Different from previous multisense word embeddings models that assign multiple independent and sense-specific embeddings to each word, our proposed models are lightweight and have flexible frameworks that regard word sense as the composition of two parts: a general sense derived from a large corpus and a topic-specific sense derived from a task-specific corpus.", "category": "method" }, { "sentence": "In addition, our proposed models focus on semantic composition instead of word understanding.", "category": "method" }, { "sentence": "With the help of topic models, we can integrate the topic-specific sense at word-level before the composition and sentence-level after the composition.", "category": "method" }, { "sentence": "Comprehensive experiments on five public sentence classification datasets are conducted and the results show that our proposed topic-aware deep compositional models produce competitive or better performance than other text representation learning methods.", "category": "result" } ] }, { "paper_id": "14386920", "title": "Evaluating Probabilistic Forecasts Using Information Theory", "abstract": "The problem of assessing the quality of an operational forecasting system that produces probabilistic forecasts is addressed using information theory. A measure of the quality of the forecasting scheme, based on the amount of a data compression it allows, is outlined. This measure, called ignorance, is a logarithmic scoring rule that is a modified version of relative entropy and can be calculated for real forecasts and realizations. It is equivalent to the expected returns that would be obtained by placing bets proportional to the forecast probabilities. Like the cost–loss score, ignorance is not equivalent to the Brier score, but, unlike cost–loss scores, ignorance easily generalizes beyond binary decision scenarios. The use of the skill score is illustrated by evaluating the ECMWF ensemble forecasts for temperature at London’s Heathrow airport.", "classified_sentences": [ { "sentence": "The problem of assessing the quality of an operational forecasting system that produces probabilistic forecasts is addressed using information theory.", "category": "background" }, { "sentence": "A measure of the quality of the forecasting scheme, based on the amount of a data compression it allows, is outlined.", "category": "method" }, { "sentence": "This measure, called ignorance, is a logarithmic scoring rule that is a modified version of relative entropy and can be calculated for real forecasts and realizations.", "category": "method" }, { "sentence": "It is equivalent to the expected returns that would be obtained by placing bets proportional to the forecast probabilities.", "category": "method" }, { "sentence": "Like the cost–loss score, ignorance is not equivalent to the Brier score, but, unlike cost–loss scores, ignorance easily generalizes beyond binary decision scenarios.", "category": "method" }, { "sentence": "The use of the skill score is illustrated by evaluating the ECMWF ensemble forecasts for temperature at London’s Heathrow airport.", "category": "result" } ] }, { "paper_id": "14455617", "title": "Efficient Salient Region Detection with Soft Image Abstraction", "abstract": "Detecting visually salient regions in images is one of the fundamental problems in computer vision. We propose a novel method to decompose an image into large scale perceptually homogeneous elements for efficient salient region detection, using a soft image abstraction representation. By considering both appearance similarity and spatial distribution of image pixels, the proposed representation abstracts out unnecessary image details, allowing the assignment of comparable saliency values across similar regions, and producing perceptually accurate salient region detection. We evaluate our salient region detection approach on the largest publicly available dataset with pixel accurate annotations. The experimental results show that the proposed method outperforms 18 alternate methods, reducing the mean absolute error by 25.2% compared to the previous best result, while being computationally more efficient.", "classified_sentences": [ { "sentence": "Detecting visually salient regions in images is one of the fundamental problems in computer vision.", "category": "background" }, { "sentence": "We propose a novel method to decompose an image into large scale perceptually homogeneous elements for efficient salient region detection, using a soft image abstraction representation.", "category": "method" }, { "sentence": "By considering both appearance similarity and spatial distribution of image pixels, the proposed representation abstracts out unnecessary image details, allowing the assignment of comparable saliency values across similar regions, and producing perceptually accurate salient region detection.", "category": "method" }, { "sentence": "We evaluate our salient region detection approach on the largest publicly available dataset with pixel accurate annotations.", "category": "method" }, { "sentence": "The experimental results show that the proposed method outperforms 18 alternate methods, reducing the mean absolute error by 25.2% compared to the previous best result, while being computationally more efficient.", "category": "result" } ] }, { "paper_id": "14712921", "title": "Predictive Uncertainty Estimation of Hydrological Multi-Model Ensembles Using Pair-Copula Construction", "abstract": "Predictive uncertainty (PU) is defined as the probability of occurrence of an observed variable of interest, conditional on all available information. In this context, hydrological model predictions and forecasts are considered to be accessible but yet uncertain information. To estimate the PU of hydrological multi-model ensembles, we apply a method based on the use of copulas which enables modelling the dependency structures between variates independently of their marginal distributions. Given that the option to employ copula functions imposes certain limitations in the multivariate case, we model the multivariate distribution as a cascade of bivariate copulas by using the pair-copula construction. We apply a mixture of probability distributions to estimate the marginal densities and distributions of daily flow rates for various meteorological and hydrological situations. The proposed method is applied to a multi-model ensemble involving two hydrological and one statistical flow models at two gauge stations in the Moselle river basin. Verification and inter-comparison with other PU assessment methods show that copulas are well-suited for this scope and constitute a valid approach for predictive uncertainty estimation of hydrological multi-model predictions.", "classified_sentences": [ { "sentence": "Predictive uncertainty (PU) is defined as the probability of occurrence of an observed variable of interest, conditional on all available information.", "category": "background" }, { "sentence": "In this context, hydrological model predictions and forecasts are considered to be accessible but yet uncertain information.", "category": "background" }, { "sentence": "To estimate the PU of hydrological multi-model ensembles, we apply a method based on the use of copulas which enables modelling the dependency structures between variates independently of their marginal distributions.", "category": "method" }, { "sentence": "Given that the option to employ copula functions imposes certain limitations in the multivariate case, we model the multivariate distribution as a cascade of bivariate copulas by using the pair-copula construction.", "category": "method" }, { "sentence": "We apply a mixture of probability distributions to estimate the marginal densities and distributions of daily flow rates for various meteorological and hydrological situations.", "category": "method" }, { "sentence": "The proposed method is applied to a multi-model ensemble involving two hydrological and one statistical flow models at two gauge stations in the Moselle river basin.", "category": "method" }, { "sentence": "Verification and inter-comparison with other PU assessment methods show that copulas are well-suited for this scope and constitute a valid approach for predictive uncertainty estimation of hydrological multi-model predictions.", "category": "result" } ] }, { "paper_id": "14828437", "title": "Compact Global Descriptors for Visual Search", "abstract": "The first step in an image retrieval pipeline consists of comparing global descriptors from a large database to find a short list of candidate matching images. The more compact the global descriptor, the faster the descriptors can be compared for matching. State-of-the-art global descriptors based on Fisher Vectors are represented with tens of thousands of floating point numbers. While there is significant work on compression of local descriptors, there is relatively little work on compression of high dimensional Fisher Vectors. We study the problem of global descriptor compression in the context of image retrieval, focusing on extremely compact binary representations: 64-1024 bits. Motivated by the remarkable success of deep neural networks in recent literature, we propose a compression scheme based on deeply stacked Restricted Boltzmann Machines (SRBM), which learn lower dimensional non-linear subspaces on which the data lie. We provide a thorough evaluation of several state-of-the-art compression schemes based on PCA, Locality Sensitive Hashing, Product Quantization and greedy bit selection, and show that the proposed compression scheme outperforms all existing schemes.", "classified_sentences": [ { "sentence": "The first step in an image retrieval pipeline consists of comparing global descriptors from a large database to find a short list of candidate matching images.", "category": "background" }, { "sentence": "The more compact the global descriptor, the faster the descriptors can be compared for matching.", "category": "background" }, { "sentence": "State-of-the-art global descriptors based on Fisher Vectors are represented with tens of thousands of floating point numbers.", "category": "background" }, { "sentence": "While there is significant work on compression of local descriptors, there is relatively little work on compression of high dimensional Fisher Vectors.", "category": "background" }, { "sentence": "We study the problem of global descriptor compression in the context of image retrieval, focusing on extremely compact binary representations: 64-1024 bits.", "category": "method" }, { "sentence": "Motivated by the remarkable success of deep neural networks in recent literature, we propose a compression scheme based on deeply stacked Restricted Boltzmann Machines (SRBM), which learn lower dimensional non-linear subspaces on which the data lie.", "category": "method" }, { "sentence": "We provide a thorough evaluation of several state-of-the-art compression schemes based on PCA, Locality Sensitive Hashing, Product Quantization and greedy bit selection, and show that the proposed compression scheme outperforms all existing schemes.", "category": "result" } ] }, { "paper_id": "15976968", "title": "Probabilistic Quantitative Precipitation Forecasts Based on Reforecast Analogs: Theory and Application", "abstract": "A general theory is proposed for the statistical correction of weather forecasts based on observed analogs. An estimate is sought for the probability density function (pdf) of the observed state, given today’s numerical forecast. Assume that an infinite set of reforecasts (hindcasts) and associated observations are available and that the climate is stable. Assume that it is possible to find a set of past model forecast states that are nearly identical to the current forecast state. With the dates of these past forecasts, the asymptotically correct probabilistic forecast can be formed from the distribution of observed states on those dates. Unfortunately, this general theory of analogs is not useful for estimating the global pdf with a limited set of reforecasts, for the chance of finding even one effectively identical forecast analog in that limited set is vanishingly small, and the climate is not stable. Nonetheless, approximations can be made to this theory to make it useful for statistically correcting weather forecasts. For instance, when estimating the state in a local region, choose the dates of analogs based on a pattern match of the local weather forecast; with a few decades of reforecasts, there are usually many close analogs. Several approximate analog techniques are then tested for their ability to skillfully calibrate probabilistic forecasts of 24-h precipitation amount. A 25-yr set of reforecasts from a reduced-resolution global forecast model is used. The analog techniques find past ensemble-mean forecasts in a local region that are similar to today’s ensemble-mean forecasts in that region. Probabilistic forecasts are formed from the analyzed weather on the dates of the past analogs. All of the analog techniques provide dramatic improvements in the Brier skill score relative to basing probabilities on the raw ensemble counts or the counts corrected for bias. However, the analog techniques did not produce guidance that was much more skillful than that produced by a logistic regression technique. Among the analog techniques tested, it was determined that small improvements to the baseline analog technique that matches ensemble-mean precipitation forecasts are possible. Forecast skill can be improved slightly by matching the ranks of the mean forecasts rather than the raw mean forecasts by using highly localized search regions for shorter-term forecasts and larger search regions for longer forecasts, by matching precipitable water in addition to precipitation amount, and by spatially smoothing the probabilities.", "classified_sentences": [ { "sentence": "A general theory is proposed for the statistical correction of weather forecasts based on observed analogs.", "category": "method" }, { "sentence": "An estimate is sought for the probability density function (pdf) of the observed state, given today’s numerical forecast.", "category": "method" }, { "sentence": "Assume that an infinite set of reforecasts (hindcasts) and associated observations are available and that the climate is stable.", "category": "background" }, { "sentence": "Assume that it is possible to find a set of past model forecast states that are nearly identical to the current forecast state.", "category": "background" }, { "sentence": "With the dates of these past forecasts, the asymptotically correct probabilistic forecast can be formed from the distribution of observed states on those dates.", "category": "method" }, { "sentence": "Unfortunately, this general theory of analogs is not useful for estimating the global pdf with a limited set of reforecasts, for the chance of finding even one effectively identical forecast analog in that limited set is vanishingly small, and the climate is not stable.", "category": "background" }, { "sentence": "Nonetheless, approximations can be made to this theory to make it useful for statistically correcting weather forecasts.", "category": "method" }, { "sentence": "For instance, when estimating the state in a local region, choose the dates of analogs based on a pattern match of the local weather forecast; with a few decades of reforecasts, there are usually many close analogs.", "category": "method" }, { "sentence": "Several approximate analog techniques are then tested for their ability to skillfully calibrate probabilistic forecasts of 24-h precipitation amount.", "category": "method" }, { "sentence": "A 25-yr set of reforecasts from a reduced-resolution global forecast model is used.", "category": "method" }, { "sentence": "The analog techniques find past ensemble-mean forecasts in a local region that are similar to today’s ensemble-mean forecasts in that region.", "category": "result" }, { "sentence": "Probabilistic forecasts are formed from the analyzed weather on the dates of the past analogs.", "category": "method" }, { "sentence": "All of the analog techniques provide dramatic improvements in the Brier skill score relative to basing probabilities on the raw ensemble counts or the counts corrected for bias.", "category": "result" }, { "sentence": "However, the analog techniques did not produce guidance that was much more skillful than that produced by a logistic regression technique.", "category": "result" }, { "sentence": "Among the analog techniques tested, it was determined that small improvements to the baseline analog technique that matches ensemble-mean precipitation forecasts are possible.", "category": "result" }, { "sentence": "Forecast skill can be improved slightly by matching the ranks of the mean forecasts rather than the raw mean forecasts by using highly localized search regions for shorter-term forecasts and larger search regions for longer forecasts, by matching precipitable water in addition to precipitation amount, and by spatially smoothing the probabilities.", "category": "method" } ] }, { "paper_id": "16679628", "title": "Improving Efficiency in Training of Artificial Neural Networks using Information-rich Data", "abstract": "Artificial Neural Networks (ANNs) are classified as a data-driven technique, which implies that their learning improves as more and more training data are presented. This observation is based on the premise that a longer time series of training samples will contain more events of different types, and hence, the generalization ability of the ANN will improve. However, a longer time series need not necessarily contain more information. If there is considerable repetition of the same type of information, the ANN may not become “wiser”, and one may be just wasting computational effort and time. This study assumes that there are segments in a long time series that contain a large quantum of information. The reason behind this assumption is that the information contained in any hydrological series is not uniformly distributed, and it may be cyclic in nature. If an ANN is trained using these segments rather than the whole series, the training would be the same or better based on the information contained in the series. A pre-processing can be used to select information-rich data for training. However, most of the conventional pre-processing methods do not perform well due to large variation in magnitude, scale and many zeros in the data series. Therefore, it is not very easy to identify these information-rich segments in long time series with large variation in magnitude and many zeros. In this study, the data depth function was used as a tool for the identification of critical (information) segments in a time series, which does not depend on large variation in magnitude, scale or the presence of many zeros in data. Data from two gauging sites were used to compare the performance of ANN trained on the whole data set and just the data from critical events. Selection of data for critical events was done by two methods, using the depth function (identification of critical events (ICE) algorithm) and using random selection. Inter-comparison of the performance of the ANNs trained using the complete data sets and the pruned data sets shows that the ANN trained using the data from critical events, i.e., information-rich data (whose length could be one third to half of the series), gave similar results as the ANN trained using the complete data set. However, if the data set is pruned randomly, the performance of the ANN degrades significantly. The concept of this paper may be very useful for training data-driven models where the training time series is incomplete.", "classified_sentences": [ { "sentence": "Artificial Neural Networks (ANNs) are classified as a data-driven technique, which implies that their learning improves as more and more training data are presented.", "category": "background" }, { "sentence": "This observation is based on the premise that a longer time series of training samples will contain more events of different types, and hence, the generalization ability of the ANN will improve.", "category": "background" }, { "sentence": "However, a longer time series need not necessarily contain more information.", "category": "background" }, { "sentence": "If there is considerable repetition of the same type of information, the ANN may not become “wiser”, and one may be just wasting computational effort and time.", "category": "background" }, { "sentence": "This study assumes that there are segments in a long time series that contain a large quantum of information.", "category": "background" }, { "sentence": "The reason behind this assumption is that the information contained in any hydrological series is not uniformly distributed, and it may be cyclic in nature.", "category": "background" }, { "sentence": "If an ANN is trained using these segments rather than the whole series, the training would be the same or better based on the information contained in the series.", "category": "method" }, { "sentence": "A pre-processing can be used to select information-rich data for training.", "category": "method" }, { "sentence": "However, most of the conventional pre-processing methods do not perform well due to large variation in magnitude, scale and many zeros in the data series.", "category": "background" }, { "sentence": "Therefore, it is not very easy to identify these information-rich segments in long time series with large variation in magnitude and many zeros.", "category": "background" }, { "sentence": "In this study, the data depth function was used as a tool for the identification of critical (information) segments in a time series, which does not depend on large variation in magnitude, scale or the presence of many zeros in data.", "category": "method" }, { "sentence": "Data from two gauging sites were used to compare the performance of ANN trained on the whole data set and just the data from critical events.", "category": "method" }, { "sentence": "Selection of data for critical events was done by two methods, using the depth function (identification of critical events (ICE) algorithm) and using random selection.", "category": "method" }, { "sentence": "Inter-comparison of the performance of the ANNs trained using the complete data sets and the pruned data sets shows that the ANN trained using the data from critical events, i.e., information-rich data (whose length could be one third to half of the series), gave similar results as the ANN trained using the complete data set.", "category": "result" }, { "sentence": "However, if the data set is pruned randomly, the performance of the ANN degrades significantly.", "category": "result" }, { "sentence": "The concept of this paper may be very useful for training data-driven models where the training time series is incomplete.", "category": "background" } ] }, { "paper_id": "19172367", "title": "Word translation with Wikipedia", "abstract": "Wikipedia contains a large collection of articles which are connected across languages. In this research two models are proposed which use this structure to obtain word translations. The first model solely uses the link structure within and between two language projects. The second model uses cosine similarity between word embeddings and uses the first model as seed model. The second model uses two monolingual corpora and is based on the assumption that the most similar words in the source and goal language of a word are the same. The first model’s performance lies between 30-55% correct translations, the second model has a performance between 1-22% correct translations. The poor performance of the second model shows that the assumption that translation is possible based solely on similar words might be too naive.", "classified_sentences": [ { "sentence": "Wikipedia contains a large collection of articles which are connected across languages.", "category": "background" }, { "sentence": "In this research two models are proposed which use this structure to obtain word translations.", "category": "method" }, { "sentence": "The first model solely uses the link structure within and between two language projects.", "category": "method" }, { "sentence": "The second model uses cosine similarity between word embeddings and uses the first model as seed model.", "category": "method" }, { "sentence": "The second model uses two monolingual corpora and is based on the assumption that the most similar words in the source and goal language of a word are the same.", "category": "method" }, { "sentence": "The first model’s performance lies between 30-55% correct translations, the second model has a performance between 1-22% correct translations.", "category": "result" }, { "sentence": "The poor performance of the second model shows that the assumption that translation is possible based solely on similar words might be too naive.", "category": "result" } ] }, { "paper_id": "162577031", "title": "Hurricane Katrina and the Forgotten Coast of Mississippi", "abstract": "HURRICANE KATRINA AND THE FORGOTTEN COAST OF MISSISSIPPI. By Susan L. Cutter, Christopher T. Emrich, Jerry T. Mitchell, Walter W. Piegorsch, Mark M. Smith, and Lynn Weber, xiv and 194 pp.; maps, diagrs. , ills. , notes, index; New York: Cambridge University Press, 2014. $120.00 (cloth), ISBN 9781107023949. As the book title indicates, Hurricane Katrina and the Forgotten Coast of Mississippi examines the social and economic consequences of Hurricane Katrina upon Mississippi coastal communities to understand different outcomes to the recovery effort. The authors provide a well-written and thorough assessment of recovery through geographic data on social, built-environment, and hazard vulnerabilities; the history of the Mississippi Gulf Coast as it relates to disasters and social difference; statistical analysis of settlement and demographic change on the coast; and trajectory predictions of settlement and demographic change. Mississippi, despite the significant damage that the coast received from Hurricane Katrina, did not receive adequate attention from the media nor academic researchers. As the authors argue, the Mississippi coast has a unique historical background and social and economic trajectory. Blanket theories of impact, recovery, or forecasted trajectories are not sufficient for understanding \"the forgotten coast of Mississippi. \" The authors address this gap in the literature. The book starts at a very logical point for disaster literature: defining recovery. The authors set out to define recovery and, more importantly, they seek to explain differences in recovery as they relate to the physical landscape of the Mississippi Gulf Coast, as well as different rates of social and economic recovery. They present a multidisciplinary assessment of the social and economic processes producing unequal recovery. After recovery and its measures are addressed, the authors examine the history of hurricanes and their impact on the Mississippi coast while addressing the disparate levels of damage. The authors introduce \"Katrina fatigue\" as the exhaustion of frontline recovery workers, including nonprofits, which occurred within two to three years of the storm (p. 106). Using qualitative and quantitative data on past and current recovery efforts, the authors forecast the trajectory that recovery from Hurricane Katrina that will continue on the Mississippi coast, as well as the ability of the coast to absorb and respond to inevitable future disasters. Recovery is not easily defined or measured. As disaster impacts fall into areas addressed by multiple disciplines, single variable measures of recovery are simply inadequate. The authors are to be applauded on their efforts to take a multidisciplinary, as well as a normative, approach to recovery. Disaster resilience and recovery literature is often satisfied with answering questions about the way things were, the way things are, and the delta between, while ignoring how things ought to be. The authors address recovery for whom, recovery of what, and recovery to what end in their assessment of Mississippi's Hurricane Katrina story. They consider the interaction of the differential impact of the hurricane, as well as the social vulnerabilities. Recovery inequities are handled not just at the county level as the authors seek to provide insight into why certain communities may be lagging or leading the way in recovery. The emphasis that the authors place on power and its role in processes producing unequal recovery is essential to understanding Mississippi and it's response to disasters. …", "classified_sentences": [ { "sentence": "HURRICANE KATRINA AND THE FORGOTTEN COAST OF MISSISSIPPI.", "category": "background" }, { "sentence": "By Susan L.", "category": "background" }, { "sentence": "Cutter, Christopher T.", "category": "background" }, { "sentence": "Emrich, Jerry T.", "category": "background" }, { "sentence": "Mitchell, Walter W.", "category": "background" }, { "sentence": "Piegorsch, Mark M.", "category": "background" }, { "sentence": "Smith, and Lynn Weber, xiv and 194 pp.; maps, diagrs.", "category": "background" }, { "sentence": ", ills.", "category": "background" }, { "sentence": ", notes, index; New York: Cambridge University Press, 2014.", "category": "background" }, { "sentence": "$120.00 (cloth), ISBN 9781107023949.", "category": "background" }, { "sentence": "As the book title indicates, Hurricane Katrina and the Forgotten Coast of Mississippi examines the social and economic consequences of Hurricane Katrina upon Mississippi coastal communities to understand different outcomes to the recovery effort.", "category": "background" }, { "sentence": "The authors provide a well-written and thorough assessment of recovery through geographic data on social, built-environment, and hazard vulnerabilities; the history of the Mississippi Gulf Coast as it relates to disasters and social difference; statistical analysis of settlement and demographic change on the coast; and trajectory predictions of settlement and demographic change.", "category": "method" }, { "sentence": "Mississippi, despite the significant damage that the coast received from Hurricane Katrina, did not receive adequate attention from the media nor academic researchers.", "category": "background" }, { "sentence": "As the authors argue, the Mississippi coast has a unique historical background and social and economic trajectory.", "category": "background" }, { "sentence": "Blanket theories of impact, recovery, or forecasted trajectories are not sufficient for understanding \"the forgotten coast of Mississippi.", "category": "background" }, { "sentence": "\" The authors address this gap in the literature.", "category": "background" }, { "sentence": "The book starts at a very logical point for disaster literature: defining recovery.", "category": "method" }, { "sentence": "The authors set out to define recovery and, more importantly, they seek to explain differences in recovery as they relate to the physical landscape of the Mississippi Gulf Coast, as well as different rates of social and economic recovery.", "category": "method" }, { "sentence": "They present a multidisciplinary assessment of the social and economic processes producing unequal recovery.", "category": "method" }, { "sentence": "After recovery and its measures are addressed, the authors examine the history of hurricanes and their impact on the Mississippi coast while addressing the disparate levels of damage.", "category": "method" }, { "sentence": "The authors introduce \"Katrina fatigue\" as the exhaustion of frontline recovery workers, including nonprofits, which occurred within two to three years of the storm (p. 106).", "category": "method" }, { "sentence": "Using qualitative and quantitative data on past and current recovery efforts, the authors forecast the trajectory that recovery from Hurricane Katrina that will continue on the Mississippi coast, as well as the ability of the coast to absorb and respond to inevitable future disasters.", "category": "method" }, { "sentence": "Recovery is not easily defined or measured.", "category": "background" }, { "sentence": "As disaster impacts fall into areas addressed by multiple disciplines, single variable measures of recovery are simply inadequate.", "category": "background" }, { "sentence": "The authors are to be applauded on their efforts to take a multidisciplinary, as well as a normative, approach to recovery.", "category": "method" }, { "sentence": "Disaster resilience and recovery literature is often satisfied with answering questions about the way things were, the way things are, and the delta between, while ignoring how things ought to be.", "category": "background" }, { "sentence": "The authors address recovery for whom, recovery of what, and recovery to what end in their assessment of Mississippi's Hurricane Katrina story.", "category": "method" }, { "sentence": "They consider the interaction of the differential impact of the hurricane, as well as the social vulnerabilities.", "category": "method" }, { "sentence": "Recovery inequities are handled not just at the county level as the authors seek to provide insight into why certain communities may be lagging or leading the way in recovery.", "category": "method" }, { "sentence": "The emphasis that the authors place on power and its role in processes producing unequal recovery is essential to understanding Mississippi and it's response to disasters.", "category": "method" } ] }, { "paper_id": "86617491", "title": "Machine Learning Tool Development in Fire Safety Design Review", "abstract": "The feasibility of object detection technique to recognize and localize the less-featured building elements on the architectural drawings was tested. An object detection engine using Faster R-CNN deep learning techniques was trained that machine can learn from existing architectural drawings what the different variations of the building components look like. Then machine is capable to tag these elements on a 2D architectural drawing, and engineers can apply some rule-based checks to flag up any design that violates the building code of practice. It is proved that machine can go through a large amount of design drawings quickly, recognize where the important building components are located, and report whether the design conforms to the building codes.", "classified_sentences": [ { "sentence": "The feasibility of object detection technique to recognize and localize the less-featured building elements on the architectural drawings was tested.", "category": "background" }, { "sentence": "An object detection engine using Faster R-CNN deep learning techniques was trained that machine can learn from existing architectural drawings what the different variations of the building components look like.", "category": "method" }, { "sentence": "Then machine is capable to tag these elements on a 2D architectural drawing, and engineers can apply some rule-based checks to flag up any design that violates the building code of practice.", "category": "method" }, { "sentence": "It is proved that machine can go through a large amount of design drawings quickly, recognize where the important building components are located, and report whether the design conforms to the building codes.", "category": "result" } ] }, { "paper_id": "39643440", "title": "Survey on Malicious Web Pages Detection Techniques", "abstract": "The World Wide Web has become an inseparable part of millions of people who use online services e.g. online banking, online shopping, social networking, e-commerce, and store and manage user sensitive information, etc. In fact, it is a popular tool for any class of user over the Internet. Rich Web based applications are available over the World Wide Web to provide such types of services. At the same time, the Web has become an important means for people to interact with each other and do business. This is the positive side of this technology. Unfortunately, the Web has also become a more dangerous place. The popularity of World Wide Web has also attracted intruders and attackers. These intruders abuse the Internet and users by performing illegitimate activity for financial profit. The Web pages that contain such types of attacks or malicious code are called as malicious Web pages. While the existing approaches are good indicators in detecting malicious Web pages, there are still open issues in Web page features selection and detection techniques. In this paper, we are giving an extensive survey of existing malicious Web pages detection approaches and features they have used.", "classified_sentences": [ { "sentence": "The World Wide Web has become an inseparable part of millions of people who use online services e.g. online banking, online shopping, social networking, e-commerce, and store and manage user sensitive information, etc. In fact, it is a popular tool for any class of user over the Internet.", "category": "background" }, { "sentence": "Rich Web based applications are available over the World Wide Web to provide such types of services.", "category": "background" }, { "sentence": "At the same time, the Web has become an important means for people to interact with each other and do business.", "category": "background" }, { "sentence": "This is the positive side of this technology.", "category": "background" }, { "sentence": "Unfortunately, the Web has also become a more dangerous place.", "category": "background" }, { "sentence": "The popularity of World Wide Web has also attracted intruders and attackers.", "category": "background" }, { "sentence": "These intruders abuse the Internet and users by performing illegitimate activity for financial profit.", "category": "background" }, { "sentence": "The Web pages that contain such types of attacks or malicious code are called as malicious Web pages.", "category": "background" }, { "sentence": "While the existing approaches are good indicators in detecting malicious Web pages, there are still open issues in Web page features selection and detection techniques.", "category": "background" }, { "sentence": "In this paper, we are giving an extensive survey of existing malicious Web pages detection approaches and features they have used.", "category": "method" } ] }, { "paper_id": "41532174", "title": "A new graph ranking model for image saliency detection problem", "abstract": "Saliency detection is an important problem in many computer vision applications. As a kind of popular method, graph based manifold ranking (GMR) has been successfully used in saliency detection problem. In traditional GMR saliency detection, it involves two main stages, i.e., ranking with background queries and ranking with foreground queries. However, in GMR method, these two stages are conducted separately, which ignores the correlation between background and foreground cues. In this paper, we propose a new graph ranking model, which aims to perform background and foreground ranking simultaneously by exploiting the correlation between background and foreground cues. We derive a closed-form solution for it. Experimental results on four benchmark datasets demonstrate that the proposed method performs better than some other state-of-art methods.", "classified_sentences": [ { "sentence": "Saliency detection is an important problem in many computer vision applications.", "category": "background" }, { "sentence": "As a kind of popular method, graph based manifold ranking (GMR) has been successfully used in saliency detection problem.", "category": "method" }, { "sentence": "In traditional GMR saliency detection, it involves two main stages, i.e., ranking with background queries and ranking with foreground queries.", "category": "method" }, { "sentence": "However, in GMR method, these two stages are conducted separately, which ignores the correlation between background and foreground cues.", "category": "background" }, { "sentence": "In this paper, we propose a new graph ranking model, which aims to perform background and foreground ranking simultaneously by exploiting the correlation between background and foreground cues.", "category": "method" }, { "sentence": "We derive a closed-form solution for it.", "category": "method" }, { "sentence": "Experimental results on four benchmark datasets demonstrate that the proposed method performs better than some other state-of-art methods.", "category": "result" } ] }, { "paper_id": "44872246", "title": "An Image Quality Assessment Index Based on Visual Saliency and Gradient Amplitude for Telemedicine Application", "abstract": "This paper proposes an image quality assessment method based on visual saliency and gradient amplitude. Multi-scale decomposition is used, then with the analysis of the image's phase spectrum in frequency domain, the visual saliency map can be obtained. Meanwhile, the gradient amplitude map is obtained through analyzing the image's contrast feature. The weighted products of the visual saliency map and the gradient amplitude map of all scales are synthesized to get the objective quality evaluation value. Many experiments are made on three image quality assessment (IQA) databases which are available to the public, including TID2008, IVC and A57. The experimental results demonstrate that the method proposed in this paper can obtain greater consistency with human subjective evaluation than that of other related image quality assessment indexes.", "classified_sentences": [ { "sentence": "This paper proposes an image quality assessment method based on visual saliency and gradient amplitude.", "category": "method" }, { "sentence": "Multi-scale decomposition is used, then with the analysis of the image's phase spectrum in frequency domain, the visual saliency map can be obtained.", "category": "method" }, { "sentence": "Meanwhile, the gradient amplitude map is obtained through analyzing the image's contrast feature.", "category": "method" }, { "sentence": "The weighted products of the visual saliency map and the gradient amplitude map of all scales are synthesized to get the objective quality evaluation value.", "category": "method" }, { "sentence": "Many experiments are made on three image quality assessment (IQA) databases which are available to the public, including TID2008, IVC and A57.", "category": "method" }, { "sentence": "The experimental results demonstrate that the method proposed in this paper can obtain greater consistency with human subjective evaluation than that of other related image quality assessment indexes.", "category": "result" } ] }, { "paper_id": "46922267", "title": "Deep belief network based detection and categorization of malicious URLs", "abstract": "ABSTRACT The Internet, web consumers and computing systems have become more vulnerable to cyber-attacks. Malicious uniform resource locator (URL) is a prominent cyber-attack broadly used with the intention of data, money or personal information stealing. Malicious URLs comprise phishing URLs, spamming URLs, and malware URLs. Detection of malicious URL and identification of their attack type are important to thwart such attacks and to adopt required countermeasures. The proposed methodology for detection and categorization of malicious URLs uses stacked restricted Boltzmann machine for feature selection with deep neural network for binary classification. For multiple classes, IBK-kNN, Binary Relevance, and Label Powerset with SVM are used for classification. The approach is tested with 27700 URL samples and the results demonstrate that the deep learning-based feature selection and classification techniques are able to quickly train the network and detect with reduced false positives.", "classified_sentences": [ { "sentence": "The Internet, web consumers and computing systems have become more vulnerable to cyber-attacks.", "category": "background" }, { "sentence": "Malicious uniform resource locator (URL) is a prominent cyber-attack broadly used with the intention of data, money or personal information stealing.", "category": "background" }, { "sentence": "Malicious URLs comprise phishing URLs, spamming URLs, and malware URLs.", "category": "background" }, { "sentence": "Detection of malicious URL and identification of their attack type are important to thwart such attacks and to adopt required countermeasures.", "category": "background" }, { "sentence": "The proposed methodology for detection and categorization of malicious URLs uses stacked restricted Boltzmann machine for feature selection with deep neural network for binary classification.", "category": "method" }, { "sentence": "For multiple classes, IBK-kNN, Binary Relevance, and Label Powerset with SVM are used for classification.", "category": "method" }, { "sentence": "The approach is tested with 27700 URL samples and the results demonstrate that the deep learning-based feature selection and classification techniques are able to quickly train the network and detect with reduced false positives.", "category": "result" } ] }, { "paper_id": "46957798", "title": "Training Convolutional Networks on Truncated Text", "abstract": "Classifiers trained using deep neural networks have been shown to be effective for a wide variety of classification tasks including text sentiment. One such approach is to use convolutional neural networks to learn from character-level representations of documents. This approach is appealing as all feature engineering, extraction or selection is performed by the neural network, automatically generating high level abstract representations of the data. With character-level learning, network topology is dependent on document length as this determines the input shape for the network. In this paper, we investigate how limiting the number of characters used from a document impacts performance and compare neural network performance against a Multinomial Naive Bayes baseline. Our results show the required number of characters is linked to the document domain and neural network performance exceeds that of Multinomial Naive Bayes; however, too low of a number results in performance degradation.", "classified_sentences": [ { "sentence": "Classifiers trained using deep neural networks have been shown to be effective for a wide variety of classification tasks including text sentiment.", "category": "background" }, { "sentence": "One such approach is to use convolutional neural networks to learn from character-level representations of documents.", "category": "method" }, { "sentence": "This approach is appealing as all feature engineering, extraction or selection is performed by the neural network, automatically generating high level abstract representations of the data.", "category": "background" }, { "sentence": "With character-level learning, network topology is dependent on document length as this determines the input shape for the network.", "category": "method" }, { "sentence": "In this paper, we investigate how limiting the number of characters used from a document impacts performance and compare neural network performance against a Multinomial Naive Bayes baseline.", "category": "method" }, { "sentence": "Our results show the required number of characters is linked to the document domain and neural network performance exceeds that of Multinomial Naive Bayes; however, too low of a number results in performance degradation.", "category": "result" } ] }, { "paper_id": "52338594", "title": "Predicting Argumenthood of English Preposition Phrases", "abstract": "Distinguishing between core and non-core dependents (i.e., arguments and adjuncts) of a verb is a longstanding, nontrivial problem. In natural language processing, argumenthood information is important in tasks such as semantic role labeling (SRL) and preposition phrase (PP) attachment disambiguation. In theoretical linguistics, many diagnostic tests for argumenthood exist but they often yield conflicting and potentially gradient results. This is especially the case for syntactically oblique items such as PPs. We propose two PP argumenthood prediction tasks branching from these two motivations: (1) binary argument/adjunct classification of PPs in VerbNet, and (2) gradient argumenthood prediction using human judgments as gold standard, and report results from prediction models that use pretrained word embeddings and other linguistically informed features. Our best results on each task are (1) $acc. =0.955$, $F_1=0.954$ (ELMo+BiLSTM) and (2) Pearson's $r=0.624$ (word2vec+MLP). Furthermore, we demonstrate the utility of argumenthood prediction in improving sentence representations via performance gains on SRL when a sentence encoder is pretrained with our tasks.", "classified_sentences": [ { "sentence": "Distinguishing between core and non-core dependents (i.e., arguments and adjuncts) of a verb is a longstanding, nontrivial problem.", "category": "background" }, { "sentence": "In natural language processing, argumenthood information is important in tasks such as semantic role labeling (SRL) and preposition phrase (PP) attachment disambiguation.", "category": "background" }, { "sentence": "In theoretical linguistics, many diagnostic tests for argumenthood exist but they often yield conflicting and potentially gradient results.", "category": "background" }, { "sentence": "This is especially the case for syntactically oblique items such as PPs.", "category": "background" }, { "sentence": "We propose two PP argumenthood prediction tasks branching from these two motivations: (1) binary argument/adjunct classification of PPs in VerbNet, and (2) gradient argumenthood prediction using human judgments as gold standard, and report results from prediction models that use pretrained word embeddings and other linguistically informed features.", "category": "method" }, { "sentence": "Our best results on each task are (1) $acc.", "category": "result" }, { "sentence": "=0.955$, $F_1=0.954$ (ELMo+BiLSTM) and (2) Pearson's $r=0.624$ (word2vec+MLP).", "category": "result" }, { "sentence": "Furthermore, we demonstrate the utility of argumenthood prediction in improving sentence representations via performance gains on SRL when a sentence encoder is pretrained with our tasks.", "category": "result" } ] }, { "paper_id": "52941969", "title": "Unifying Sum and Weighted Aggregations for Efficient Yet Effective Image Representation Computation", "abstract": "Embedding and aggregating a set of local descriptors (e.g., SIFT) into a single vector is normally used to represent images in image search. Standard aggregation operations include sum and weighted aggregations. While showing high efficiency, sum aggregation lacks discriminative power. In contrast, weighted aggregation shows promising retrieval performance but suffers extremely high time cost. In this paper, we present a general mixed aggregation method that unifies sum and weighted aggregation methods. Owing to its general formulation, our method is able to balance the trade-off between retrieval quality and image representation efficiency. Additionally, to improve query performance, we propose computing multiple weighting coefficients rather than one for each to be aggregated vector by partitioning them into several components with negligible computational cost. Extensive experimental results on standard public image retrieval benchmarks demonstrate that our aggregation method achieves state-of-the-art performance while showing over ten times speedup over baselines.", "classified_sentences": [ { "sentence": "Embedding and aggregating a set of local descriptors (e.g., SIFT) into a single vector is normally used to represent images in image search.", "category": "background" }, { "sentence": "Standard aggregation operations include sum and weighted aggregations.", "category": "background" }, { "sentence": "While showing high efficiency, sum aggregation lacks discriminative power.", "category": "background" }, { "sentence": "In contrast, weighted aggregation shows promising retrieval performance but suffers extremely high time cost.", "category": "background" }, { "sentence": "In this paper, we present a general mixed aggregation method that unifies sum and weighted aggregation methods.", "category": "method" }, { "sentence": "Owing to its general formulation, our method is able to balance the trade-off between retrieval quality and image representation efficiency.", "category": "method" }, { "sentence": "Additionally, to improve query performance, we propose computing multiple weighting coefficients rather than one for each to be aggregated vector by partitioning them into several components with negligible computational cost.", "category": "method" }, { "sentence": "Extensive experimental results on standard public image retrieval benchmarks demonstrate that our aggregation method achieves state-of-the-art performance while showing over ten times speedup over baselines.", "category": "result" } ] }, { "paper_id": "53074742", "title": "SALGAN 360 : VISUAL SALIENCY PREDICTION ON 360 DEGREE IMAGES WITH GENERATIVE ADVERSARIAL NETWORKS", "abstract": "Understanding visual attention of observers on 360◦ images gains interest along with the booming trend of Virtual Reality applications. Extending existing saliency prediction methods from traditional 2D images to 360◦ images is not a direct approach due to the lack of a sufficient large 360◦ image saliency database. In this paper, we propose to extend the SalGAN, a 2D saliency model based on the generative adversarial network, to SalGAN360 by fine tuning the SalGAN with our new loss function to predict both global and local saliency maps. Our experiments show that the SalGAN360 outperforms the tested state-of-the-art methods.", "classified_sentences": [ { "sentence": "Understanding visual attention of observers on 360◦ images gains interest along with the booming trend of Virtual Reality applications.", "category": "background" }, { "sentence": "Extending existing saliency prediction methods from traditional 2D images to 360◦ images is not a direct approach due to the lack of a sufficient large 360◦ image saliency database.", "category": "background" }, { "sentence": "In this paper, we propose to extend the SalGAN, a 2D saliency model based on the generative adversarial network, to SalGAN360 by fine tuning the SalGAN with our new loss function to predict both global and local saliency maps.", "category": "method" }, { "sentence": "Our experiments show that the SalGAN360 outperforms the tested state-of-the-art methods.", "category": "result" } ] }, { "paper_id": "53965832", "title": "Greedy Hash: Towards Fast Optimization for Accurate Hash Coding in CNN", "abstract": "To convert the input into binary code, hashing algorithm has been widely used for approximate nearest neighbor search on large-scale image sets due to its computation and storage efficiency. Deep hashing further improves the retrieval quality by combining the hash coding with deep neural network. However, a major difficulty in deep hashing lies in the discrete constraints imposed on the network output, which generally makes the optimization NP hard. In this work, we adopt the greedy principle to tackle this NP hard problem by iteratively updating the network toward the probable optimal discrete solution in each iteration. A hash coding layer is designed to implement our approach which strictly uses the sign function in forward propagation to maintain the discrete constraints, while in back propagation the gradients are transmitted intactly to the front layer to avoid the vanishing gradients. In addition to the theoretical derivation, we provide a new perspective to visualize and understand the effectiveness and efficiency of our algorithm. Experiments on benchmark datasets show that our scheme outperforms state-of-the-art hashing methods in both supervised and unsupervised tasks.", "classified_sentences": [ { "sentence": "To convert the input into binary code, hashing algorithm has been widely used for approximate nearest neighbor search on large-scale image sets due to its computation and storage efficiency.", "category": "background" }, { "sentence": "Deep hashing further improves the retrieval quality by combining the hash coding with deep neural network.", "category": "background" }, { "sentence": "However, a major difficulty in deep hashing lies in the discrete constraints imposed on the network output, which generally makes the optimization NP hard.", "category": "background" }, { "sentence": "In this work, we adopt the greedy principle to tackle this NP hard problem by iteratively updating the network toward the probable optimal discrete solution in each iteration.", "category": "method" }, { "sentence": "A hash coding layer is designed to implement our approach which strictly uses the sign function in forward propagation to maintain the discrete constraints, while in back propagation the gradients are transmitted intactly to the front layer to avoid the vanishing gradients.", "category": "method" }, { "sentence": "In addition to the theoretical derivation, we provide a new perspective to visualize and understand the effectiveness and efficiency of our algorithm.", "category": "method" }, { "sentence": "Experiments on benchmark datasets show that our scheme outperforms state-of-the-art hashing methods in both supervised and unsupervised tasks.", "category": "result" } ] }, { "paper_id": "56170365", "title": "Region of Interest Detection based on Local Entropy Feature for Disaster Victim Detection System", "abstract": "Region of interest (ROI) detection plays an important role in object detection. It needs to be accurate and fast in some applications like real time disaster victim detection systems. ROI can reduce time and search space in detecting objects. In this paper visual saliency map is used for ROI detection. In most literature, most of ROI detection models only concentrate on reducing false positive (detecting wrong objects as intended ones) rate rather than false negative (missing intended object). In disaster victim detection, missing disaster victims is more important than detecting other objects like victim. So, the proposed method also focuses on reducing false negative error rate in object detection. In the proposed system, local entropy feature is added in Graph Based Visual Saliency (GBVS) map in addition to colour, orientation and shape feature maps.", "classified_sentences": [ { "sentence": "Region of interest (ROI) detection plays an important role in object detection.", "category": "background" }, { "sentence": "It needs to be accurate and fast in some applications like real time disaster victim detection systems.", "category": "background" }, { "sentence": "ROI can reduce time and search space in detecting objects.", "category": "background" }, { "sentence": "In this paper visual saliency map is used for ROI detection.", "category": "method" }, { "sentence": "In most literature, most of ROI detection models only concentrate on reducing false positive (detecting wrong objects as intended ones) rate rather than false negative (missing intended object).", "category": "background" }, { "sentence": "In disaster victim detection, missing disaster victims is more important than detecting other objects like victim.", "category": "background" }, { "sentence": "So, the proposed method also focuses on reducing false negative error rate in object detection.", "category": "method" }, { "sentence": "In the proposed system, local entropy feature is added in Graph Based Visual Saliency (GBVS) map in addition to colour, orientation and shape feature maps.", "category": "method" } ] }, { "paper_id": "56248749", "title": "Ensemble forecasting: status and perspectives", "abstract": "One of the main challenges for Numerical Weather Prediction is the Quantitative Precipitation Forecasting (QPF). The accurate forecast of high-impact weather still remains difficult beyond day 2 and many limited-area ensemble prediction systems have been recently developed so as to provide more reliable forecasts than achievable with a single deterministic forecast. As a consequence the calibration of ensemble precipitation forecasts has become a very demanding task, for improving the QPF, especially as an input to hydrological models. Different calibration techniques are compared: cumulative distribution function, linear regression and analogues method.", "classified_sentences": [ { "sentence": "One of the main challenges for Numerical Weather Prediction is the Quantitative Precipitation Forecasting (QPF).", "category": "background" }, { "sentence": "The accurate forecast of high-impact weather still remains difficult beyond day 2 and many limited-area ensemble prediction systems have been recently developed so as to provide more reliable forecasts than achievable with a single deterministic forecast.", "category": "background" }, { "sentence": "As a consequence the calibration of ensemble precipitation forecasts has become a very demanding task, for improving the QPF, especially as an input to hydrological models.", "category": "background" }, { "sentence": "Different calibration techniques are compared: cumulative distribution function, linear regression and analogues method.", "category": "method" } ] }, { "paper_id": "56256046", "title": "An Improved Region Contrast and Global Distribution Saliency Detection algorithm", "abstract": "Keywords: Saliency detection, region contrast, global distribution Abstract. According to the local contrast and global distribution of an image, this paper detecting salient images through bottom-up data driven . First, this paper using adaptive segmentation method divided image into non-overlapping images, improved Block and Chessboar distance from a linear combination to replace the Euclidean distance method to calculate the regional features of contrast functions, then calculate the global distribution of feature functions, finally fusion of the above features for computing saliency map. The algorithm taking into account local features and global features to get more accurate saliency map. Test our method on the international public data sets MSRA-1000,the experimental result proves that the images extracted by this method are more accurate and more clearly, while reducing the calculation time of regional characteristics, having strong noise and high texture regions resistance , and can basically ignore the complex background.", "classified_sentences": [ { "sentence": "Keywords: Saliency detection, region contrast, global distribution Abstract.", "category": "background" }, { "sentence": "According to the local contrast and global distribution of an image, this paper detecting salient images through bottom-up data driven .", "category": "method" }, { "sentence": "First, this paper using adaptive segmentation method divided image into non-overlapping images, improved Block and Chessboar distance from a linear combination to replace the Euclidean distance method to calculate the regional features of contrast functions, then calculate the global distribution of feature functions, finally fusion of the above features for computing saliency map.", "category": "method" }, { "sentence": "The algorithm taking into account local features and global features to get more accurate saliency map.", "category": "method" }, { "sentence": "Test our method on the international public data sets MSRA-1000,the experimental result proves that the images extracted by this method are more accurate and more clearly, while reducing the calculation time of regional characteristics, having strong noise and high texture regions resistance , and can basically ignore the complex background.", "category": "result" } ] }, { "paper_id": "56393852", "title": "Application of the NCEP Ensemble Prediction System to Medium-Range Forecasting in South Africa: New Products, Benefits, and Challenges", "abstract": "Abstract The National Centers for Environmental Prediction (NCEP) Ensemble Forecasting System (EFS) is used operationally in South Africa for medium-range forecasts up to 14 days ahead. The use of model-generated probability forecasts has a clear benefit in the skill of the 1–7-day forecasts. This is seen in the forecast probability distribution being more successful in spanning the observed space than a single deterministic forecast and, thus, substantially reducing the instances of missed events in the forecast. In addition, the probability forecasts generated using the EFS are particularly useful in estimating confidence in forecasts. During the second week of the forecast the EFS is used as a heads-up for possible synoptic-scale events and also for predicting average weather conditions and probability density distributions of some elements such as maximum temperature and wind. This paper assesses the medium-range forecast process and the application of the NCEP EFS at the South African Weather Service.", "classified_sentences": [ { "sentence": "Abstract The National Centers for Environmental Prediction (NCEP) Ensemble Forecasting System (EFS) is used operationally in South Africa for medium-range forecasts up to 14 days ahead.", "category": "background" }, { "sentence": "The use of model-generated probability forecasts has a clear benefit in the skill of the 1–7-day forecasts.", "category": "background" }, { "sentence": "This is seen in the forecast probability distribution being more successful in spanning the observed space than a single deterministic forecast and, thus, substantially reducing the instances of missed events in the forecast.", "category": "result" }, { "sentence": "In addition, the probability forecasts generated using the EFS are particularly useful in estimating confidence in forecasts.", "category": "result" }, { "sentence": "During the second week of the forecast the EFS is used as a heads-up for possible synoptic-scale events and also for predicting average weather conditions and probability density distributions of some elements such as maximum temperature and wind.", "category": "method" }, { "sentence": "This paper assesses the medium-range forecast process and the application of the NCEP EFS at the South African Weather Service.", "category": "method" } ] }, { "paper_id": "193718462", "title": "arTenTen: a new, vast corpus for Arabic", "abstract": "We present arTenTen, a web crawled corpus of Arabic, gathered in 2012, and a member of the TenTen Corpus Family (Jakubicek et al 2013). arTenTen comprises 5.8 billion words. It has been carefully cleaned, including duplicate removal, using the JusText and Onion tools (Pomikalek 2011). We are currently (May 2013) in the process of tokenising, lemmatising and part-of-speech tagging arTenTen with the leading MADA tool version 3.2 (Habash and Rambow 2005; Habash et al. 2009). Once arTenTen is fully encoded, we will compare it with Arabic Gigaword and an earlier web-crawled corpus (Sharoff 2006). We also plan to explore arTenTen’s composition in relation to Modern Standard Arabic and the dialects, using, amongst other things, Buckwalter and Parkinson’s Frequency Dictionary (2011) and the keywords method presented in (Kilgarriff 2012).", "classified_sentences": [ { "sentence": "We present arTenTen, a web crawled corpus of Arabic, gathered in 2012, and a member of the TenTen Corpus Family (Jakubicek et al 2013).", "category": "background" }, { "sentence": "arTenTen comprises 5.8 billion words.", "category": "background" }, { "sentence": "It has been carefully cleaned, including duplicate removal, using the JusText and Onion tools (Pomikalek 2011).", "category": "method" }, { "sentence": "We are currently (May 2013) in the process of tokenising, lemmatising and part-of-speech tagging arTenTen with the leading MADA tool version 3.2 (Habash and Rambow 2005; Habash et al. 2009).", "category": "method" }, { "sentence": "Once arTenTen is fully encoded, we will compare it with Arabic Gigaword and an earlier web-crawled corpus (Sharoff 2006).", "category": "method" }, { "sentence": "We also plan to explore arTenTen’s composition in relation to Modern Standard Arabic and the dialects, using, amongst other things, Buckwalter and Parkinson’s Frequency Dictionary (2011) and the keywords method presented in (Kilgarriff 2012).", "category": "method" } ] }, { "paper_id": "195970621", "title": "Using Machine Learning to Detect Malicious URLs", "abstract": "on the opinion of the heads of departments, without taking into account their relationship. It is noted that due to the lack of automation and correctly conducted analysis, program projects are formed with errors and inaccuracies. To improve the situation, it is proposed to use an integrated information system based on data array analytics and the use of AI. The system consists of three subsystems: \"LifeControl\" for automatic analysis of the needs of the population, the subsystem of the balanced scorecard, which sets the task of determining key indicators that allow assessing the development of the municipality, and the subsystem for supporting the development and analysis of program projects.", "classified_sentences": [ { "sentence": "on the opinion of the heads of departments, without taking into account their relationship.", "category": "background" }, { "sentence": "It is noted that due to the lack of automation and correctly conducted analysis, program projects are formed with errors and inaccuracies.", "category": "background" }, { "sentence": "To improve the situation, it is proposed to use an integrated information system based on data array analytics and the use of AI.", "category": "method" }, { "sentence": "The system consists of three subsystems: \"LifeControl\" for automatic analysis of the needs of the population, the subsystem of the balanced scorecard, which sets the task of determining key indicators that allow assessing the development of the municipality, and the subsystem for supporting the development and analysis of program projects.", "category": "method" } ] }, { "paper_id": "198905491", "title": "Tweet Classification without the Tweet: An Empirical Examination of User versus Document Attributes", "abstract": "NLP naturally puts a primary focus on leveraging document language, occasionally considering user attributes as supplemental. However, as we tackle more social scientific tasks, it is possible user attributes might be of primary importance and the document supplemental. Here, we systematically investigate the predictive power of user-level features alone versus document-level features for document-level tasks. We first show user attributes can sometimes carry more task-related information than the document itself. For example, a tweet-level stance detection model using only 13 user-level attributes (i.e. features that did not depend on the specific tweet) was able to obtain a higher F1 than the top-performing SemEval participant. We then consider multiple tasks and a wider range of user attributes, showing the performance of strong document-only models can often be improved (as in stance, sentiment, and sarcasm) with user attributes, particularly benefiting tasks with stable “trait-like” outcomes (e.g. stance) most relative to frequently changing “state-like” outcomes (e.g. sentiment). These results not only support the growing work on integrating user factors into predictive systems, but that some of our NLP tasks might be better cast primarily as user-level (or human) tasks.", "classified_sentences": [ { "sentence": "NLP naturally puts a primary focus on leveraging document language, occasionally considering user attributes as supplemental.", "category": "background" }, { "sentence": "However, as we tackle more social scientific tasks, it is possible user attributes might be of primary importance and the document supplemental.", "category": "background" }, { "sentence": "Here, we systematically investigate the predictive power of user-level features alone versus document-level features for document-level tasks.", "category": "method" }, { "sentence": "We first show user attributes can sometimes carry more task-related information than the document itself.", "category": "result" }, { "sentence": "For example, a tweet-level stance detection model using only 13 user-level attributes (i.e. features that did not depend on the specific tweet) was able to obtain a higher F1 than the top-performing SemEval participant.", "category": "result" }, { "sentence": "We then consider multiple tasks and a wider range of user attributes, showing the performance of strong document-only models can often be improved (as in stance, sentiment, and sarcasm) with user attributes, particularly benefiting tasks with stable “trait-like” outcomes (e.g. stance) most relative to frequently changing “state-like” outcomes (e.g. sentiment).", "category": "result" }, { "sentence": "These results not only support the growing work on integrating user factors into predictive systems, but that some of our NLP tasks might be better cast primarily as user-level (or human) tasks.", "category": "result" } ] }, { "paper_id": "65083087", "title": "Structured learning with inexact search: advances in shift-reduce CCG parsing", "abstract": "Statistical shift-reduce parsing involves the interplay of representation learning, structured learning, and inexact search. This dissertation considers approaches that tightly integrate these three elements and explores three novel models for shift-reduce CCG parsing. First, I develop a dependency model, in which the selection of shift-reduce action sequences producing a dependency structure is treated as a hidden variable; the key components of the model are a dependency oracle and a learning algorithm that integrates the dependency oracle, the structured perceptron, and beam search. Second, I present expected F-measure training and show how to derive a globally normalized RNN model, in which beam search is naturally incorporated and used in conjunction with the objective to learn shift-reduce action sequences optimized for the final evaluation metric. Finally, I describe an LSTM model that is able to construct parser state representations incrementally by following the shift-reduce syntactic derivation process; I show expected F-measure training, which is agnostic to the underlying neural network, can be applied in this setting to obtain globally normalized greedy and beam-search LSTM shift-reduce parsers.", "classified_sentences": [ { "sentence": "Statistical shift-reduce parsing involves the interplay of representation learning, structured learning, and inexact search.", "category": "background" }, { "sentence": "This dissertation considers approaches that tightly integrate these three elements and explores three novel models for shift-reduce CCG parsing.", "category": "method" }, { "sentence": "First, I develop a dependency model, in which the selection of shift-reduce action sequences producing a dependency structure is treated as a hidden variable; the key components of the model are a dependency oracle and a learning algorithm that integrates the dependency oracle, the structured perceptron, and beam search.", "category": "method" }, { "sentence": "Second, I present expected F-measure training and show how to derive a globally normalized RNN model, in which beam search is naturally incorporated and used in conjunction with the objective to learn shift-reduce action sequences optimized for the final evaluation metric.", "category": "method" }, { "sentence": "Finally, I describe an LSTM model that is able to construct parser state representations incrementally by following the shift-reduce syntactic derivation process; I show expected F-measure training, which is agnostic to the underlying neural network, can be applied in this setting to obtain globally normalized greedy and beam-search LSTM shift-reduce parsers.", "category": "method" } ] }, { "paper_id": "201636367", "title": "Analogies Explained: Towards Understanding Word Embeddings", "abstract": "Word embeddings generated by neural network methods such as word2vec (W2V) are well known to exhibit seemingly linear behaviour, e.g. the embeddings of analogy “woman is to queen as man is to king” approximately describe a parallelogram. This property is particularly intriguing since the embeddings are not trained to achieve it. Several explanations have been proposed, but each introduces assumptions that do not hold in practice. We derive a probabilistically grounded definition of paraphrasing that we re-interpret as word transformation, a mathematical description of “wx is to wy”. From these concepts we prove existence of linear relationships between W2V-type embeddings that underlie the analogical phenomenon, identifying explicit error terms.", "classified_sentences": [ { "sentence": "Word embeddings generated by neural network methods such as word2vec (W2V) are well known to exhibit seemingly linear behaviour, e.g. the embeddings of analogy “woman is to queen as man is to king” approximately describe a parallelogram.", "category": "background" }, { "sentence": "This property is particularly intriguing since the embeddings are not trained to achieve it.", "category": "background" }, { "sentence": "Several explanations have been proposed, but each introduces assumptions that do not hold in practice.", "category": "background" }, { "sentence": "We derive a probabilistically grounded definition of paraphrasing that we re-interpret as word transformation, a mathematical description of “wx is to wy”.", "category": "method" }, { "sentence": "From these concepts we prove existence of linear relationships between W2V-type embeddings that underlie the analogical phenomenon, identifying explicit error terms.", "category": "result" } ] }, { "paper_id": "201809837", "title": "Deep Architectures for Crowd Flow Prediction", "abstract": "Crowd flow prediction is significant in crowd management and public security. However, accurate crowd flow prediction is challenging, for it is influenced by numerous complicated factors, such as traffic accidents and weather impact. In this treatise, we recommend two deep crowd flow prediction architectures: P-GRU and P-DBT by introducing a gated recurrent unit network/a deep Bi-LSTM model, regression layer, precipitation record, dropout training method, and residual network. The proposed models possess a nice capacity to dig up the deeply hidden information of crowd flow. Moreover, they are able to make efficient use of crowd flow data and precipitation recordings. The forecast architectures are assessed on taxi trajectory data and bike trajectory data in Chongqing with additional precipitation recordings collected from China Meteorological Data Service Center. Experiments based on two kinds of datasets demonstrate that the proposed models have a more ideal performance in comparison with other model architectures.", "classified_sentences": [ { "sentence": "Crowd flow prediction is significant in crowd management and public security.", "category": "background" }, { "sentence": "However, accurate crowd flow prediction is challenging, for it is influenced by numerous complicated factors, such as traffic accidents and weather impact.", "category": "background" }, { "sentence": "In this treatise, we recommend two deep crowd flow prediction architectures: P-GRU and P-DBT by introducing a gated recurrent unit network/a deep Bi-LSTM model, regression layer, precipitation record, dropout training method, and residual network.", "category": "method" }, { "sentence": "The proposed models possess a nice capacity to dig up the deeply hidden information of crowd flow.", "category": "method" }, { "sentence": "Moreover, they are able to make efficient use of crowd flow data and precipitation recordings.", "category": "method" }, { "sentence": "The forecast architectures are assessed on taxi trajectory data and bike trajectory data in Chongqing with additional precipitation recordings collected from China Meteorological Data Service Center.", "category": "method" }, { "sentence": "Experiments based on two kinds of datasets demonstrate that the proposed models have a more ideal performance in comparison with other model architectures.", "category": "result" } ] }, { "paper_id": "202127341", "title": "Multi-Layer Abstraction Saliency for Airport Detection in SAR Images", "abstract": "The detection of airports using synthetic aperture radar (SAR) images has attracted considerable attention. Traditional methods easily result in inaccurate detection due to the complex scenes and multiplicative speckle noise. Therefore, airport detection from SAR images is still a challenging task. In order to limit the influence of unnecessary and attractive details and noise, we propose a multi-layer abstraction saliency model for airport detection in SAR images in this paper. Specifically, we first obtain airport support regions and superpixels in the first layer. According to the dis-similarity between foreground and background superpixels, airport components are explored by iterative refinement for each airport support region in the second layer. In the third layer, airport adobes are produced by clustering. Based on the characteristics of an airport in SAR images, we propose three saliency cues, including local contrast (LC), adobe deformation (AD), and global uniqueness (GU), to obtain adobe-level saliency. Furthermore, we assign saliency to each pixel by Bayesian inference. Finally, we can explore airport location using integrated saliency map. The proposed approach is tested on an airport data set collected from Gaofen-3, TerraSAR, and RadarSat. Our method achieves 88.89% detection rate. The experimental results demonstrate that the proposed algorithm is effective and outperforms the previously airport detection methods. The code will be available at https://github.com/NengyuanLiu/MyAirportSaliency.", "classified_sentences": [ { "sentence": "The detection of airports using synthetic aperture radar (SAR) images has attracted considerable attention.", "category": "background" }, { "sentence": "Traditional methods easily result in inaccurate detection due to the complex scenes and multiplicative speckle noise.", "category": "background" }, { "sentence": "Therefore, airport detection from SAR images is still a challenging task.", "category": "background" }, { "sentence": "In order to limit the influence of unnecessary and attractive details and noise, we propose a multi-layer abstraction saliency model for airport detection in SAR images in this paper.", "category": "method" }, { "sentence": "Specifically, we first obtain airport support regions and superpixels in the first layer.", "category": "method" }, { "sentence": "According to the dis-similarity between foreground and background superpixels, airport components are explored by iterative refinement for each airport support region in the second layer.", "category": "method" }, { "sentence": "In the third layer, airport adobes are produced by clustering.", "category": "method" }, { "sentence": "Based on the characteristics of an airport in SAR images, we propose three saliency cues, including local contrast (LC), adobe deformation (AD), and global uniqueness (GU), to obtain adobe-level saliency.", "category": "method" }, { "sentence": "Furthermore, we assign saliency to each pixel by Bayesian inference.", "category": "method" }, { "sentence": "Finally, we can explore airport location using integrated saliency map.", "category": "method" }, { "sentence": "The proposed approach is tested on an airport data set collected from Gaofen-3, TerraSAR, and RadarSat.", "category": "method" }, { "sentence": "Our method achieves 88.89% detection rate.", "category": "result" }, { "sentence": "The experimental results demonstrate that the proposed algorithm is effective and outperforms the previously airport detection methods.", "category": "result" }, { "sentence": "The code will be available at https://github.com/NengyuanLiu/MyAirportSaliency.", "category": "method" } ] }, { "paper_id": "69833386", "title": "Predicting the Argumenthood of English Prepositional Phrases", "abstract": "Distinguishing between arguments and adjuncts of a verb is a longstanding, nontrivial problem. In natural language processing, argumenthood information is important in tasks such as semantic role labeling (SRL) and prepositional phrase (PP) attachment disambiguation. In theoretical linguistics, many diagnostic tests for argumenthood exist but they often yield conflicting and potentially gradient results. This is especially the case for syntactically oblique items such as PPs. We propose two PP argumenthood prediction tasks branching from these two motivations: (1) binary argumentadjunct classification of PPs in VerbNet, and (2) gradient argumenthood prediction using human judgments as gold standard, and report results from prediction models that use pretrained word embeddings and other linguistically informed features. Our best results on each task are (1) acc. = 0.955, F1 = 0.954 (ELMo+BiLSTM) and (2) Pearson’s r = 0.624 (word2vec+MLP). Furthermore, we demonstrate the utility of argumenthood prediction in improving sentence representations via performance gains on SRL when a sentence encoder is pretrained with our tasks.", "classified_sentences": [ { "sentence": "Distinguishing between arguments and adjuncts of a verb is a longstanding, nontrivial problem.", "category": "background" }, { "sentence": "In natural language processing, argumenthood information is important in tasks such as semantic role labeling (SRL) and prepositional phrase (PP) attachment disambiguation.", "category": "background" }, { "sentence": "In theoretical linguistics, many diagnostic tests for argumenthood exist but they often yield conflicting and potentially gradient results.", "category": "background" }, { "sentence": "This is especially the case for syntactically oblique items such as PPs.", "category": "background" }, { "sentence": "We propose two PP argumenthood prediction tasks branching from these two motivations: (1) binary argument-adjunct classification of PPs in VerbNet, and (2) gradient argumenthood prediction using human judgments as gold standard, and report results from prediction models that use pretrained word embeddings and other linguistically informed features.", "category": "method" }, { "sentence": "Our best results on each task are (1) acc. = 0.955, F1 = 0.954 (ELMo+BiLSTM) and (2) Pearson’s r = 0.624 (word2vec+MLP).", "category": "result" }, { "sentence": "Furthermore, we demonstrate the utility of argumenthood prediction in improving sentence representations via performance gains on SRL when a sentence encoder is pretrained with our tasks.", "category": "result" } ] }, { "paper_id": "71149533", "title": "Digging Deeper Into Egocentric Gaze Prediction", "abstract": "This paper digs deeper into factors that influence egocentric gaze. Instead of training deep models for this purpose in a blind manner, we propose to inspect factors that contribute to gaze guidance during daily tasks. Bottom-up saliency and optical flow are assessed versus strong spatial prior baselines. Task-specific cues such as vanishing point, manipulation point, and hand regions are analyzed as representatives of top-down information. We also look into the contribution of these factors by investigating a simple recurrent neural model for ego-centric gaze prediction. First, deep features are extracted for all input video frames. Then, a gated recurrent unit is employed to integrate information over time and to predict the next fixation. We propose an integrated model that combines the recurrent model with several top-down and bottom-up cues. Extensive experiments over multiple datasets reveal that (1) spatial biases are strong in egocentric videos, (2) bottom-up attention models perform poorly in predicting gaze and underperform spatial biases, (3) deep features perform better compared to traditional features, (4) as opposed to hand regions, the manipulation point is a strong influential cue for gaze prediction, (5) combining the proposed recurrent model with bottom-up cues, vanishing points and, in particular, manipulation point results in the best gaze prediction accuracy over egocentric videos, (6) the knowledge transfer works best for cases where the tasks or sequences are similar, and (7) task and activity recognition can benefit from gaze prediction. Our findings suggest that (1) there should be more emphasis on hand-object interaction and (2) the egocentric vision community should consider larger datasets including diverse stimuli and more subjects.", "classified_sentences": [ { "sentence": "This paper digs deeper into factors that influence egocentric gaze.", "category": "background" }, { "sentence": "Instead of training deep models for this purpose in a blind manner, we propose to inspect factors that contribute to gaze guidance during daily tasks.", "category": "method" }, { "sentence": "Bottom-up saliency and optical flow are assessed versus strong spatial prior baselines.", "category": "method" }, { "sentence": "Task-specific cues such as vanishing point, manipulation point, and hand regions are analyzed as representatives of top-down information.", "category": "method" }, { "sentence": "We also look into the contribution of these factors by investigating a simple recurrent neural model for ego-centric gaze prediction.", "category": "method" }, { "sentence": "First, deep features are extracted for all input video frames.", "category": "method" }, { "sentence": "Then, a gated recurrent unit is employed to integrate information over time and to predict the next fixation.", "category": "method" }, { "sentence": "We propose an integrated model that combines the recurrent model with several top-down and bottom-up cues.", "category": "method" }, { "sentence": "Extensive experiments over multiple datasets reveal that (1) spatial biases are strong in egocentric videos, (2) bottom-up attention models perform poorly in predicting gaze and underperform spatial biases, (3) deep features perform better compared to traditional features, (4) as opposed to hand regions, the manipulation point is a strong influential cue for gaze prediction, (5) combining the proposed recurrent model with bottom-up cues, vanishing points and, in particular, manipulation point results in the best gaze prediction accuracy over egocentric videos, (6) the knowledge transfer works best for cases where the tasks or sequences are similar, and (7) task and activity recognition can benefit from gaze prediction.", "category": "result" }, { "sentence": "Our findings suggest that (1) there should be more emphasis on hand-object interaction and (2) the egocentric vision community should consider larger datasets including diverse stimuli and more subjects.", "category": "result" } ] }, { "paper_id": "207960074", "title": "PDRCNN: Precise Phishing Detection with Recurrent Convolutional Neural Networks", "abstract": "Through well-designed counterfeit websites, phishing induces online users to visit forged web pages to obtain their private sensitive information, e.g., account number and password. Existing antiphishing approaches are mostly based on page-related features, which require to crawl content of web pages as well as accessing third-party search engines or DNS services. This not only leads to their low efficiency in detecting phishing but also makes them rely on network environment and third-party services heavily. In this paper, we propose a fast phishing website detection approach called PDRCNN that relies only on the URL of the website. PDRCNN neither needs to retrieve content of the target website nor uses any third-party services as previous approaches do. It encodes the information of an URL into a two-dimensional tensor and feeds the tensor into a novelly designed deep learning neural network to classify the original URL. We first use a bidirectional LSTM network to extract global features of the constructed tensor and give all string information to each character in the URL. After that, we use a CNN to automatically judge which characters play key roles in phishing detection, capture the key components of the URL, and compress the extracted features into a fixed length vector space. By combining the two types of networks, PDRCNN achieves better performance than just using either one of them. We built a dataset containing nearly 500,000 URLs which are obtained through Alexa and PhishTank. Experimental results show that PDRCNN achieves a detection accuracy of 97% and an AUC value of 99%, which is much better than state-of-the-art approaches. Furthermore, the recognition process is very fast: on the trained PDRCNN model, the average per URL detection time only cost 0.4 ms.", "classified_sentences": [ { "sentence": "Through well-designed counterfeit websites, phishing induces online users to visit forged web pages to obtain their private sensitive information, e.g., account number and password.", "category": "background" }, { "sentence": "Existing antiphishing approaches are mostly based on page-related features, which require to crawl content of web pages as well as accessing third-party search engines or DNS services.", "category": "background" }, { "sentence": "This not only leads to their low efficiency in detecting phishing but also makes them rely on network environment and third-party services heavily.", "category": "background" }, { "sentence": "In this paper, we propose a fast phishing website detection approach called PDRCNN that relies only on the URL of the website.", "category": "method" }, { "sentence": "PDRCNN neither needs to retrieve content of the target website nor uses any third-party services as previous approaches do.", "category": "method" }, { "sentence": "It encodes the information of an URL into a two-dimensional tensor and feeds the tensor into a novelly designed deep learning neural network to classify the original URL.", "category": "method" }, { "sentence": "We first use a bidirectional LSTM network to extract global features of the constructed tensor and give all string information to each character in the URL.", "category": "method" }, { "sentence": "After that, we use a CNN to automatically judge which characters play key roles in phishing detection, capture the key components of the URL, and compress the extracted features into a fixed length vector space.", "category": "method" }, { "sentence": "By combining the two types of networks, PDRCNN achieves better performance than just using either one of them.", "category": "method" }, { "sentence": "We built a dataset containing nearly 500,000 URLs which are obtained through Alexa and PhishTank.", "category": "method" }, { "sentence": "Experimental results show that PDRCNN achieves a detection accuracy of 97% and an AUC value of 99%, which is much better than state-of-the-art approaches.", "category": "result" }, { "sentence": "Furthermore, the recognition process is very fast: on the trained PDRCNN model, the average per URL detection time only cost 0.4 ms.", "category": "result" } ] }, { "paper_id": "207997986", "title": "Attention-based bidirectional gated recurrent unit neural networks for sentiment analysis", "abstract": "Sentiment analysis is an important research direction of natural language processing. In-depth exploration of online textual emotional information has great social significance social and commercial value for market research, online public opinion discovery and early warning. In this paper, the gated recurrent unit neural network and the attention mechanism are combined to propose a text sentiment analysis model---Attention-BGRU. The attention mechanism was added to the gated recurrent unit neural network, and the model was implemented under the Keras deep learning framework. According to the experimental results, the comparison with the existing models shows that the proposed model has obvious advantages over the general deep learning method.", "classified_sentences": [ { "sentence": "Sentiment analysis is an important research direction of natural language processing.", "category": "background" }, { "sentence": "In-depth exploration of online textual emotional information has great social significance social and commercial value for market research, online public opinion discovery and early warning.", "category": "background" }, { "sentence": "In this paper, the gated recurrent unit neural network and the attention mechanism are combined to propose a text sentiment analysis model---Attention-BGRU.", "category": "method" }, { "sentence": "The attention mechanism was added to the gated recurrent unit neural network, and the model was implemented under the Keras deep learning framework.", "category": "method" }, { "sentence": "According to the experimental results, the comparison with the existing models shows that the proposed model has obvious advantages over the general deep learning method.", "category": "result" } ] }, { "paper_id": "77392571", "title": "Exploring Efficiency of Character-level Convolution Neuron Network and Long Short Term Memory on Malicious URL Detection", "abstract": "Machine learning techniques, especially deep learning neuron networks have been increasingly applied to solve the problems relating to information security and cybersecurity. Malicious URL (Uniform Resource Locator) detection is one of these. It is considered as a binary classification in machine learning, in which a URL or website address is classed as malign or benign. In this work, we implement the experiments on two different datasets to explore the efficiency of three proposed character-level deep neuron networks: (1) CNN (Convolution Neuron Network) based on VGG-16 architecture (Visual Geometry Group), (2) LSTM (Long Short Term Memory), and a fusion of CNN and LSTM for malicious URL detection. The experimental results are promising, especially for the fusion scheme of LSTM and CNN, with above 96% for precision and 98% for recall.", "classified_sentences": [ { "sentence": "Machine learning techniques, especially deep learning neuron networks have been increasingly applied to solve the problems relating to information security and cybersecurity.", "category": "background" }, { "sentence": "Malicious URL (Uniform Resource Locator) detection is one of these.", "category": "background" }, { "sentence": "It is considered as a binary classification in machine learning, in which a URL or website address is classed as malign or benign.", "category": "background" }, { "sentence": "In this work, we implement the experiments on two different datasets to explore the efficiency of three proposed character-level deep neuron networks: (1) CNN (Convolution Neuron Network) based on VGG-16 architecture (Visual Geometry Group), (2) LSTM (Long Short Term Memory), and a fusion of CNN and LSTM for malicious URL detection.", "category": "method" }, { "sentence": "The experimental results are promising, especially for the fusion scheme of LSTM and CNN, with above 96% for precision and 98% for recall.", "category": "result" } ] }, { "paper_id": "212771741", "title": "Fusion of Handcrafted and Deep Features for Medical Image Classification", "abstract": "Medical image classification has recently attracted increased attention. Effective feature extraction and learning are key means to improve classification performance. However, in the current study, handcrafted feature are mostly designed with intuitive mode, while the deep feature depends on a large amount of training samples and has weak interpretability. To capture more discriminative features for medical image, a novel feature fusion approach, termed multi-layer visual feature fusion (MLVSF), has been proposed on the basis of low-level, mid-level and deep features. More specifically, by fusing the handcrafted and deep features generated by local binary pattern variant, bag-of-visual-words, convolutional neural network, respectively, MLVSF can effectively enhance the discriminating power of features for medical image recognition. Experimental results on two medical image datasets show that MLVSF can improve convolutional neural networks, and achieve a better classification accuracy in comparison with some state-of-the-art methods.", "classified_sentences": [ { "sentence": "Medical image classification has recently attracted increased attention.", "category": "background" }, { "sentence": "Effective feature extraction and learning are key means to improve classification performance.", "category": "background" }, { "sentence": "However, in the current study, handcrafted feature are mostly designed with intuitive mode, while the deep feature depends on a large amount of training samples and has weak interpretability.", "category": "background" }, { "sentence": "To capture more discriminative features for medical image, a novel feature fusion approach, termed multi-layer visual feature fusion (MLVSF), has been proposed on the basis of low-level, mid-level and deep features.", "category": "method" }, { "sentence": "More specifically, by fusing the handcrafted and deep features generated by local binary pattern variant, bag-of-visual-words, convolutional neural network, respectively, MLVSF can effectively enhance the discriminating power of features for medical image recognition.", "category": "method" }, { "sentence": "Experimental results on two medical image datasets show that MLVSF can improve convolutional neural networks, and achieve a better classification accuracy in comparison with some state-of-the-art methods.", "category": "result" } ] }, { "paper_id": "213275296", "title": "Malicious Url Recognition and Detection Using Attention-based Cnn-lstm", "abstract": "A malicious Uniform Resource Locator (URL) recognition and detection method based on the combination of Attention mechanism with Convolutional Neural Network and Long Short-Term Memory Network (Attention-Based CNN-LSTM), is proposed. Firstly, the WHOIS check method is used to extract and filter features, including the URL texture information, the URL string statistical information of attributes and the WHOIS information, and the features are subsequently encoded and pre-processed followed by inputting them to the constructed Convolutional Neural Network (CNN) convolution layer to extract local features. Secondly, in accordance with the weights from the Attention mechanism, the generated local features are input into the Long-Short Term Memory (LSTM) model, and subsequently pooled to calculate the global features of the URLs. Finally, the URLs are detected and classified by the SoftMax function using global features. The results demonstrate that compared with the existing methods, the Attention-based CNN-LSTM mechanism has higher accuracy for malicious URL detection.", "classified_sentences": [ { "sentence": "A malicious Uniform Resource Locator (URL) recognition and detection method based on the combination of Attention mechanism with Convolutional Neural Network and Long Short-Term Memory Network (Attention-Based CNN-LSTM), is proposed.", "category": "method" }, { "sentence": "Firstly, the WHOIS check method is used to extract and filter features, including the URL texture information, the URL string statistical information of attributes and the WHOIS information, and the features are subsequently encoded and pre-processed followed by inputting them to the constructed Convolutional Neural Network (CNN) convolution layer to extract local features.", "category": "method" }, { "sentence": "Secondly, in accordance with the weights from the Attention mechanism, the generated local features are input into the Long-Short Term Memory (LSTM) model, and subsequently pooled to calculate the global features of the URLs.", "category": "method" }, { "sentence": "Finally, the URLs are detected and classified by the SoftMax function using global features.", "category": "method" }, { "sentence": "The results demonstrate that compared with the existing methods, the Attention-based CNN-LSTM mechanism has higher accuracy for malicious URL detection.", "category": "result" } ] }, { "paper_id": "225213610", "title": "An Effective Phishing Detection Model Based on Character Level Convolutional Neural Network from URL", "abstract": "Phishing is the easiest way to use cybercrime with the aim of enticing people to give accurate information such as account IDs, bank details, and passwords. This type of cyberattack is usually triggered by emails, instant messages, or phone calls. The existing anti-phishing techniques are mainly based on source code features, which require to scrape the content of web pages, and on third-party services which retard the classification process of phishing URLs. Although the machine learning techniques have lately been used to detect phishing, they require essential manual feature engineering and are not an expert at detecting emerging phishing offenses. Due to the recent rapid development of deep learning techniques, many deep learning-based methods have also been introduced to enhance the classification performance. In this paper, a fast deep learning-based solution model, which uses character-level convolutional neural network (CNN) for phishing detection based on the URL of the website, is proposed. The proposed model does not require the retrieval of target website content or the use of any third-party services. It captures information and sequential patterns of URL strings without requiring a prior knowledge about phishing, and then uses the sequential pattern features for fast classification of the actual URL. For evaluations, comparisons are provided between different traditional machine learning models and deep learning models using various feature sets such as hand-crafted, character embedding, character level TF-IDF, and character level count vectors features. According to the experiments, the proposed model achieved an accuracy of 95.02% on our dataset and an accuracy of 98.58%, 95.46%, and 95.22% on benchmark datasets which outperform the existing phishing URL models.", "classified_sentences": [ { "sentence": "Phishing is the easiest way to use cybercrime with the aim of enticing people to give accurate information such as account IDs, bank details, and passwords.", "category": "background" }, { "sentence": "This type of cyberattack is usually triggered by emails, instant messages, or phone calls.", "category": "background" }, { "sentence": "The existing anti-phishing techniques are mainly based on source code features, which require to scrape the content of web pages, and on third-party services which retard the classification process of phishing URLs.", "category": "background" }, { "sentence": "Although the machine learning techniques have lately been used to detect phishing, they require essential manual feature engineering and are not an expert at detecting emerging phishing offenses.", "category": "background" }, { "sentence": "Due to the recent rapid development of deep learning techniques, many deep learning-based methods have also been introduced to enhance the classification performance.", "category": "background" }, { "sentence": "In this paper, a fast deep learning-based solution model, which uses character-level convolutional neural network (CNN) for phishing detection based on the URL of the website, is proposed.", "category": "method" }, { "sentence": "The proposed model does not require the retrieval of target website content or the use of any third-party services.", "category": "method" }, { "sentence": "It captures information and sequential patterns of URL strings without requiring a prior knowledge about phishing, and then uses the sequential pattern features for fast classification of the actual URL.", "category": "method" }, { "sentence": "For evaluations, comparisons are provided between different traditional machine learning models and deep learning models using various feature sets such as hand-crafted, character embedding, character level TF-IDF, and character level count vectors features.", "category": "method" }, { "sentence": "According to the experiments, the proposed model achieved an accuracy of 95.02% on our dataset and an accuracy of 98.58%, 95.46%, and 95.22% on benchmark datasets which outperform the existing phishing URL models.", "category": "result" } ] }, { "paper_id": "231740694", "title": "Revisiting the Prepositional-Phrase Attachment Problem Using Explicit Commonsense Knowledge", "abstract": "We revisit the challenging problem of resolving prepositional-phrase (PP) attachment ambiguity. To date, proposed solutions are either rule-based, where explicit grammar rules direct how to resolve ambiguities; or statistical, where the decision is learned from a corpus of labeled examples. We argue that explicit commonsense knowledge bases can provide an essential ingredient for making good attachment decisions. We implemented a module, named Patch-Comm, that can be used by a variety of conventional parsers, to make attachment decisions. Where the commonsense KB does not provide direct answers, we fall back on a more general system that infers\"out-of-knowledge-base\"assertions in a manner similar to the way some NLP systems handle out-of-vocabulary words. Our results suggest that the commonsense knowledge-based approach can provide the best of both worlds, integrating rule-based and statistical techniques. As the field is increasingly coming to recognize the importance of explainability in AI, a commonsense approach can enable NLP developers to better understand the behavior of systems, and facilitate natural dialogues with end users.", "classified_sentences": [ { "sentence": "We revisit the challenging problem of resolving prepositional-phrase (PP) attachment ambiguity.", "category": "background" }, { "sentence": "To date, proposed solutions are either rule-based, where explicit grammar rules direct how to resolve ambiguities; or statistical, where the decision is learned from a corpus of labeled examples.", "category": "background" }, { "sentence": "We argue that explicit commonsense knowledge bases can provide an essential ingredient for making good attachment decisions.", "category": "method" }, { "sentence": "We implemented a module, named Patch-Comm, that can be used by a variety of conventional parsers, to make attachment decisions.", "category": "method" }, { "sentence": "Where the commonsense KB does not provide direct answers, we fall back on a more general system that infers\"out-of-knowledge-base\"assertions in a manner similar to the way some NLP systems handle out-of-vocabulary words.", "category": "method" }, { "sentence": "Our results suggest that the commonsense knowledge-based approach can provide the best of both worlds, integrating rule-based and statistical techniques.", "category": "result" }, { "sentence": "As the field is increasingly coming to recognize the importance of explainability in AI, a commonsense approach can enable NLP developers to better understand the behavior of systems, and facilitate natural dialogues with end users.", "category": "result" } ] }, { "paper_id": "231846372", "title": "SLUA: A Super Lightweight Unsupervised Word Alignment Model via Cross-Lingual Contrastive Learning", "abstract": "Word alignment is essential for the down-streaming cross-lingual language understanding and generation tasks. Recently, the performance of the neural word alignment models [Zenkel et al. , 2020; Garg et al. , 2019; Ding et al. , 2019 ] has exceeded that of statistical models. However, they heavily rely on sophisticated translation models. In this study, we propose a Super Lightweight Unsupervised word Alignment ( SLUA ) model, in which a bidirectional symmetric attention trained with a contrastive learning objective is introduced, and an agreement loss is employed to bind the attention maps, such that the alignments follow mirror-like symmetry hypothesis. Experimental re-sults on several public benchmarks demonstrate that our model achieves competitive, if not better, performance compared to the state of the art in word alignment while significantly reducing the training and decoding time on average. Further ablation analysis and case studies show the superiority of our proposed SLUA. Notably, we recognize our model as a pioneer attempt to unify bilingual word embedding and word alignments. En-couragingly, our approach achieves 16.4 × speedup against GIZA++, and 50 × parameter compression compared with the Transformer-based alignment methods. We will release our code to facilitate the community.", "classified_sentences": [ { "sentence": "Word alignment is essential for the down-streaming cross-lingual language understanding and generation tasks.", "category": "background" }, { "sentence": "Recently, the performance of the neural word alignment models [Zenkel et al. , 2020; Garg et al. , 2019; Ding et al. , 2019 ] has exceeded that of statistical models.", "category": "background" }, { "sentence": "However, they heavily rely on sophisticated translation models.", "category": "background" }, { "sentence": "In this study, we propose a Super Lightweight Unsupervised word Alignment ( SLUA ) model, in which a bidirectional symmetric attention trained with a contrastive learning objective is introduced, and an agreement loss is employed to bind the attention maps, such that the alignments follow mirror-like symmetry hypothesis.", "category": "method" }, { "sentence": "Experimental re-sults on several public benchmarks demonstrate that our model achieves competitive, if not better, performance compared to the state of the art in word alignment while significantly reducing the training and decoding time on average.", "category": "result" }, { "sentence": "Further ablation analysis and case studies show the superiority of our proposed SLUA.", "category": "result" }, { "sentence": "Notably, we recognize our model as a pioneer attempt to unify bilingual word embedding and word alignments.", "category": "method" }, { "sentence": "En-couragingly, our approach achieves 16.4 × speedup against GIZA++, and 50 × parameter compression compared with the Transformer-based alignment methods.", "category": "result" }, { "sentence": "We will release our code to facilitate the community.", "category": "method" } ] }, { "paper_id": "244908167", "title": "PhishMatch: A Layered Approach for Effective Detection of Phishing URLs", "abstract": "Phishing attacks continue to be a significant threat on the Internet. Prior studies show that it is possible to determine whether a website is phishing or not just by analyzing its URL more carefully. A major advantage of the URL based approach is that it can identify a phishing website even before the web page is rendered in the browser, thus avoiding other potential problems such as cryptojacking and drive-by downloads. However, traditional URL based approaches have their limitations. Blacklist based approaches are prone to zero-hour phishing attacks, advanced machine learning based approaches consume high resources, and other approaches send the URL to a remote server which compromises user's privacy. In this paper, we present a layered anti-phishing defense, PhishMatch, which is robust, accurate, inexpensive, and client-side. We design a space-time efficient Aho-Corasick algorithm for exact string matching and n-gram based indexing technique for approximate string matching to detect various cybersquatting techniques in the phishing URL. To reduce false positives, we use a global whitelist and personalized user whitelists. We also determine the context in which the URL is visited and use that information to classify the input URL more accurately. The last component of PhishMatch involves a machine learning model and controlled search engine queries to classify the URL. A prototype plugin of PhishMatch, developed for the Chrome browser, was found to be fast and lightweight. Our evaluation shows that PhishMatch is both efficient and effective.", "classified_sentences": [ { "sentence": "Phishing attacks continue to be a significant threat on the Internet.", "category": "background" }, { "sentence": "Prior studies show that it is possible to determine whether a website is phishing or not just by analyzing its URL more carefully.", "category": "background" }, { "sentence": "A major advantage of the URL based approach is that it can identify a phishing website even before the web page is rendered in the browser, thus avoiding other potential problems such as cryptojacking and drive-by downloads.", "category": "background" }, { "sentence": "However, traditional URL based approaches have their limitations.", "category": "background" }, { "sentence": "Blacklist based approaches are prone to zero-hour phishing attacks, advanced machine learning based approaches consume high resources, and other approaches send the URL to a remote server which compromises user's privacy.", "category": "background" }, { "sentence": "In this paper, we present a layered anti-phishing defense, PhishMatch, which is robust, accurate, inexpensive, and client-side.", "category": "method" }, { "sentence": "We design a space-time efficient Aho-Corasick algorithm for exact string matching and n-gram based indexing technique for approximate string matching to detect various cybersquatting techniques in the phishing URL.", "category": "method" }, { "sentence": "To reduce false positives, we use a global whitelist and personalized user whitelists.", "category": "method" }, { "sentence": "We also determine the context in which the URL is visited and use that information to classify the input URL more accurately.", "category": "method" }, { "sentence": "The last component of PhishMatch involves a machine learning model and controlled search engine queries to classify the URL.", "category": "method" }, { "sentence": "A prototype plugin of PhishMatch, developed for the Chrome browser, was found to be fast and lightweight.", "category": "result" }, { "sentence": "Our evaluation shows that PhishMatch is both efficient and effective.", "category": "result" } ] }, { "paper_id": "247604055", "title": "Detection of Malicious Cyber Fraud using Machine Learning Techniques", "abstract": "As the technology and internet have come to their dawn, the rate of cyber-crimes has also increased. This increases the risk of information insecurity and the spread of crimes such as spam, farming and phishing, financial fraud, etc. Particularly, the attackers/hackers spread malicious uniform resource locators (URLs) to exploit vulnerabilities of the system and gain the personal information of the users. Thus, a study on malicious URL detection is necessary to prevent such attacks. Several studies exist which show numerous ways to determine malicious URLs based on machine learning (ML) and deep learning (DL), but there are some problems, for example, malicious features cannot be extracted efficiently. In this research, a model is proposed to ascertain malicious URLs, which is formulated on random forest, support vector machine (SVM), deep neural network (DNN), convolutional neural network (CNN). The several datasets are considered containing malicious and benign URLs to train the model to detect URL behaviour and attributes. The empirical results show that the suggested method can detect malicious URLs efficiently, based on URL behaviour and attributes. Thus, the solution may be advised as an efficient and reliable solution for the problem of malicious URL detection.", "classified_sentences": [ { "sentence": "As the technology and internet have come to their dawn, the rate of cyber-crimes has also increased.", "category": "background" }, { "sentence": "This increases the risk of information insecurity and the spread of crimes such as spam, farming and phishing, financial fraud, etc. Particularly, the attackers/hackers spread malicious uniform resource locators (URLs) to exploit vulnerabilities of the system and gain the personal information of the users.", "category": "background" }, { "sentence": "Thus, a study on malicious URL detection is necessary to prevent such attacks.", "category": "background" }, { "sentence": "Several studies exist which show numerous ways to determine malicious URLs based on machine learning (ML) and deep learning (DL), but there are some problems, for example, malicious features cannot be extracted efficiently.", "category": "background" }, { "sentence": "In this research, a model is proposed to ascertain malicious URLs, which is formulated on random forest, support vector machine (SVM), deep neural network (DNN), convolutional neural network (CNN).", "category": "method" }, { "sentence": "The several datasets are considered containing malicious and benign URLs to train the model to detect URL behaviour and attributes.", "category": "method" }, { "sentence": "The empirical results show that the suggested method can detect malicious URLs efficiently, based on URL behaviour and attributes.", "category": "result" }, { "sentence": "Thus, the solution may be advised as an efficient and reliable solution for the problem of malicious URL detection.", "category": "result" } ] }, { "paper_id": "247940179", "title": "Towards Web Phishing Detection Limitations and Mitigation", "abstract": "Web phishing remains a serious cyber threat responsible for most data breaches. Machine Learning (ML)-based anti-phishing detectors are seen as an effective countermeasure, and are increasingly adopted by web-browsers and software products. However, with an average of 10K phishing links reported per hour to platforms such as PhishTank and VirusTotal (VT), the deficiencies of such ML-based solutions are laid bare. We first explore how phishing sites bypass ML-based detection with a deep dive into 13K phishing pages targeting major brands such as Facebook. Results show successful evasion is caused by: (1) use of benign services to obscure phishing URLs; (2) high similarity between the HTML structures of phishing and benign pages; (3) hiding the ultimate phishing content within Javascript and running such scripts only on the client; (4) looking beyond typical credentials and credit cards for new content such as IDs and documents; (5) hiding phishing content until after human interaction. We attribute the root cause to the dependency of ML-based models on the vertical feature space (webpage content). These solutions rely only on what phishers present within the page itself. Thus, we propose Anti-SubtlePhish, a more resilient model based on logistic regression. The key augmentation is the inclusion of a horizontal feature space, which examines correlation variables between the final render of suspicious pages against what trusted services have recorded (e.g., PageRank). To defeat (1) and (2), we correlate information between WHOIS, PageRank, and page analytics. To combat (3), (4) and (5), we correlate features after rendering the page. Experiments with 100K phishing/benign sites show promising accuracy (98.8%). We also obtained 100% accuracy against 0-day phishing pages that were manually crafted, comparing well to the 0% recorded by VT vendors over the first four days.", "classified_sentences": [ { "sentence": "Web phishing remains a serious cyber threat responsible for most data breaches.", "category": "background" }, { "sentence": "Machine Learning (ML)-based anti-phishing detectors are seen as an effective countermeasure, and are increasingly adopted by web-browsers and software products.", "category": "background" }, { "sentence": "However, with an average of 10K phishing links reported per hour to platforms such as PhishTank and VirusTotal (VT), the deficiencies of such ML-based solutions are laid bare.", "category": "background" }, { "sentence": "We first explore how phishing sites bypass ML-based detection with a deep dive into 13K phishing pages targeting major brands such as Facebook.", "category": "method" }, { "sentence": "Results show successful evasion is caused by: (1) use of benign services to obscure phishing URLs; (2) high similarity between the HTML structures of phishing and benign pages; (3) hiding the ultimate phishing content within Javascript and running such scripts only on the client; (4) looking beyond typical credentials and credit cards for new content such as IDs and documents; (5) hiding phishing content until after human interaction.", "category": "result" }, { "sentence": "We attribute the root cause to the dependency of ML-based models on the vertical feature space (webpage content).", "category": "result" }, { "sentence": "These solutions rely only on what phishers present within the page itself.", "category": "result" }, { "sentence": "Thus, we propose Anti-SubtlePhish, a more resilient model based on logistic regression.", "category": "method" }, { "sentence": "The key augmentation is the inclusion of a horizontal feature space, which examines correlation variables between the final render of suspicious pages against what trusted services have recorded (e.g., PageRank).", "category": "method" }, { "sentence": "To defeat (1) and (2), we correlate information between WHOIS, PageRank, and page analytics.", "category": "method" }, { "sentence": "To combat (3), (4) and (5), we correlate features after rendering the page.", "category": "method" }, { "sentence": "Experiments with 100K phishing/benign sites show promising accuracy (98.8%).", "category": "result" }, { "sentence": "We also obtained 100% accuracy against 0-day phishing pages that were manually crafted, comparing well to the 0% recorded by VT vendors over the first four days.", "category": "result" } ] }, { "paper_id": "251877507", "title": "Research on Named Entity Recognition Based on Multi-Task Learning and Biaffine Mechanism", "abstract": "Commonly used nested entity recognition methods are span-based entity recognition methods, which focus on learning the head and tail representations of entities. This method lacks obvious boundary supervision, which leads to the failure of the correct candidate entities to be predicted, resulting in the problem of high precision and low recall. To solve the above problems, this paper proposes a named entity recognition method based on multi-task learning and biaffine mechanism, introduces the idea of multi-task learning, and divides the task into two subtasks, entity span classification and boundary detection. The entity span classification task uses biaffine mechanism to score the resulting spans and select the most likely entity class. The boundary detection task mainly solves the problem of low recall caused by the lack of boundary supervision in span classification. It captures the relationship between adjacent words in the input text according to the context, indicates the boundary range of entities, and enhances the span representation through additional boundary supervision. The experimental results show that the named entity recognition method based on multi-task learning and biaffine mechanism can improve the F1 value by up to 7.05%, 12.63%, and 14.68% on the GENIA, ACE2004, and ACE2005 nested datasets compared with other methods, which verifies that this method has better performance on the nested entity recognition task.", "classified_sentences": [ { "sentence": "Commonly used nested entity recognition methods are span-based entity recognition methods, which focus on learning the head and tail representations of entities.", "category": "background" }, { "sentence": "This method lacks obvious boundary supervision, which leads to the failure of the correct candidate entities to be predicted, resulting in the problem of high precision and low recall.", "category": "background" }, { "sentence": "To solve the above problems, this paper proposes a named entity recognition method based on multi-task learning and biaffine mechanism, introduces the idea of multi-task learning, and divides the task into two subtasks, entity span classification and boundary detection.", "category": "method" }, { "sentence": "The entity span classification task uses biaffine mechanism to score the resulting spans and select the most likely entity class.", "category": "method" }, { "sentence": "The boundary detection task mainly solves the problem of low recall caused by the lack of boundary supervision in span classification.", "category": "method" }, { "sentence": "It captures the relationship between adjacent words in the input text according to the context, indicates the boundary range of entities, and enhances the span representation through additional boundary supervision.", "category": "method" }, { "sentence": "The experimental results show that the named entity recognition method based on multi-task learning and biaffine mechanism can improve the F1 value by up to 7.05%, 12.63%, and 14.68% on the GENIA, ACE2004, and ACE2005 nested datasets compared with other methods, which verifies that this method has better performance on the nested entity recognition task.", "category": "result" } ] }, { "paper_id": "120980818", "title": "Atmospheric Circulation Associated with Persistent Generalized Frosts in Central-Southern South America", "abstract": "Abstract This paper describes the large-scale atmospheric circulation associated with persistent generalized frosts (GFs; at least 75% of the stations report frosts) in the east-central region of Argentina known as the Wet Pampa. The GF events are grouped according to their persistence, and NCEP–NCAR reanalysis data are used to create daily composites of mass and wind field anomalies during the 1961–90 winters. The GFs are caused by an anticyclonic anomaly that enters South America, generating southerly wind anomalies and cold air advection that are strengthened by the meridional layout of a cyclonic anomaly over the South Atlantic Ocean. In the case of the more persistent events the wind anomaly grows during the previous days and becomes quasi-stationary. Also, the study identifies at 250 hPa a double train of eastward-moving Rossby waves along the subtropical and subpolar latitudes, respectively, of the Southern Hemisphere. The layout of both wave trains favors the development of an intense southerly wi.", "classified_sentences": [ { "sentence": "Abstract This paper describes the large-scale atmospheric circulation associated with persistent generalized frosts (GFs; at least 75% of the stations report frosts) in the east-central region of Argentina known as the Wet Pampa.", "category": "background" }, { "sentence": "The GF events are grouped according to their persistence, and NCEP–NCAR reanalysis data are used to create daily composites of mass and wind field anomalies during the 1961–90 winters.", "category": "method" }, { "sentence": "The GFs are caused by an anticyclonic anomaly that enters South America, generating southerly wind anomalies and cold air advection that are strengthened by the meridional layout of a cyclonic anomaly over the South Atlantic Ocean.", "category": "result" }, { "sentence": "In the case of the more persistent events the wind anomaly grows during the previous days and becomes quasi-stationary.", "category": "result" }, { "sentence": "Also, the study identifies at 250 hPa a double train of eastward-moving Rossby waves along the subtropical and subpolar latitudes, respectively, of the Southern Hemisphere.", "category": "result" }, { "sentence": "The layout of both wave trains favors the development of an intense southerly wi.", "category": "result" } ] }, { "paper_id": "121674874", "title": "Ensemble Forecasting at NMC: The Generation of Perturbations", "abstract": "Abstract On 7 December 1992, The National Meteorological Center (NMC) started operational ensemble forecasting. The ensemble forecast configuration implemented provides 14 independent forecasts every day verifying on days 1–10. In this paper we briefly review existing methods for creating perturbations for ensemble forecasting. We point out that a regular analysis cycle is a “breeding ground” for fast-growing modes. Based on this observation, we devise a simple and inexpensive method to generate growing modes of the atmosphere. The new method, “breeding of growing modes”, or BGM, consists of one additional, perturbed short-range forecast, introduced on top of the regular analysis in an analysis cycle. The difference between the control and perturbed six-hour (first guess) forecast is scaled back to the size of the initial perturbation and then reintroduced onto the new atmospheric analysis. Thus, the perturbation evolves along with the time dependent analysis fields, ensuring that after a few days of cycl.", "classified_sentences": [ { "sentence": "Abstract On 7 December 1992, The National Meteorological Center (NMC) started operational ensemble forecasting.", "category": "background" }, { "sentence": "The ensemble forecast configuration implemented provides 14 independent forecasts every day verifying on days 1–10.", "category": "background" }, { "sentence": "In this paper we briefly review existing methods for creating perturbations for ensemble forecasting.", "category": "method" }, { "sentence": "We point out that a regular analysis cycle is a “breeding ground” for fast-growing modes.", "category": "method" }, { "sentence": "Based on this observation, we devise a simple and inexpensive method to generate growing modes of the atmosphere.", "category": "method" }, { "sentence": "The new method, “breeding of growing modes”, or BGM, consists of one additional, perturbed short-range forecast, introduced on top of the regular analysis in an analysis cycle.", "category": "method" }, { "sentence": "The difference between the control and perturbed six-hour (first guess) forecast is scaled back to the size of the initial perturbation and then reintroduced onto the new atmospheric analysis.", "category": "method" }, { "sentence": "Thus, the perturbation evolves along with the time dependent analysis fields, ensuring that after a few days of cycl.", "category": "result" } ] }, { "paper_id": "122701338", "title": "Smoothing forecast ensembles with fitted probability distributions", "abstract": "Forecast ensembles from the European Centre for Medium‐Range Weather Forecasts Ensemble Prediction System for surface‐weather variables are smoothed by fitting Gaussian distributions. The possibility of bifurcation or other non‐Gaussian behaviours in an ensemble is allowed for by including probability mixtures of two Gaussian distributions when justified by the data. Variables that are clearly non‐Gaussian (wind speed and cloud cover) are transformed before fitting, and multivariate data with dimensions as high as four are considered. The smoothed ensembles provide more‐accurate quantile and probability estimates in a perfect‐model setting, particularly for small ensemble sizes and more‐extreme events. This advantage increases in the presence of random errors in the ensemble means, but diminishes for underdispersed ensembles. Allowing representation of ensembles as Gaussian mixtures also leads to a sharper ‘spread–skill’ relationship in the data considered. Copyright © 2002 Royal Meteorological Society.", "classified_sentences": [ { "sentence": "Forecast ensembles from the European Centre for Medium‐Range Weather Forecasts Ensemble Prediction System for surface‐weather variables are smoothed by fitting Gaussian distributions.", "category": "method" }, { "sentence": "The possibility of bifurcation or other non‐Gaussian behaviours in an ensemble is allowed for by including probability mixtures of two Gaussian distributions when justified by the data.", "category": "method" }, { "sentence": "Variables that are clearly non‐Gaussian (wind speed and cloud cover) are transformed before fitting, and multivariate data with dimensions as high as four are considered.", "category": "method" }, { "sentence": "The smoothed ensembles provide more‐accurate quantile and probability estimates in a perfect‐model setting, particularly for small ensemble sizes and more‐extreme events.", "category": "result" }, { "sentence": "This advantage increases in the presence of random errors in the ensemble means, but diminishes for underdispersed ensembles.", "category": "result" }, { "sentence": "Allowing representation of ensembles as Gaussian mixtures also leads to a sharper ‘spread–skill’ relationship in the data considered.", "category": "result" } ] }, { "paper_id": "123435390", "title": "Using Varied Microphysics to Account for Uncertainty in Warm-Season QPF in a Convection-Allowing Ensemble", "abstract": "AbstractTwo approaches for accounting for errors in quantitative precipitation forecasts (QPFs) due to uncertainty in the microphysics (MP) parameterization in a convection-allowing ensemble are examined. They include mixed MP (MMP) composed mostly of double-moment schemes and perturbing parameters within the Weather Research and Forecasting single-moment 6-class microphysics scheme (WSM6) MP scheme (PPMP). Thirty-five cases of real-time storm-scale ensemble forecasts produced by the Center for Analysis and Prediction of Storms during the NOAA Hazardous Weather Testbed 2011 Spring Experiment were examined.The MMP ensemble had better fractions Brier scores (FBSs) for most lead times and thresholds, but the PPMP ensemble had better relative operating characteristic (ROC) scores for higher precipitation thresholds. The pooled ensemble formed by randomly drawing five members from the MMP and PPMP ensembles was no more skillful than the more accurate of the MMP and PPMP ensembles. Significant positive impact w.", "classified_sentences": [ { "sentence": "AbstractTwo approaches for accounting for errors in quantitative precipitation forecasts (QPFs) due to uncertainty in the microphysics (MP) parameterization in a convection-allowing ensemble are examined.", "category": "background" }, { "sentence": "They include mixed MP (MMP) composed mostly of double-moment schemes and perturbing parameters within the Weather Research and Forecasting single-moment 6-class microphysics scheme (WSM6) MP scheme (PPMP).", "category": "method" }, { "sentence": "Thirty-five cases of real-time storm-scale ensemble forecasts produced by the Center for Analysis and Prediction of Storms during the NOAA Hazardous Weather Testbed 2011 Spring Experiment were examined.", "category": "method" }, { "sentence": "The MMP ensemble had better fractions Brier scores (FBSs) for most lead times and thresholds, but the PPMP ensemble had better relative operating characteristic (ROC) scores for higher precipitation thresholds.", "category": "result" }, { "sentence": "The pooled ensemble formed by randomly drawing five members from the MMP and PPMP ensembles was no more skillful than the more accurate of the MMP and PPMP ensembles.", "category": "result" }, { "sentence": "Significant positive impact w.", "category": "result" } ] }, { "paper_id": "123477672", "title": "Skill and relative economic value of the ECMWF ensemble prediction system", "abstract": "The economic value of the European Centre for Medium‐Range Weather Forecasts (ECMWF) operational ensemble prediction system (EPS) is assessed relative to the value of a perfect deterministic forecast. The EPS has substantial relative value throughout the medium range. Probability forecasts derived from the EPS are of greater benefit than a deterministic forecast produced by the same model. Indeed, for many users, the probability forecasts have more value than a shorter‐range deterministic forecast. Based on the measures used here, the additional information in the EPS (reflecting the uncertainty in the initial conditions) provides a benefit to users equivalent to many years' development of the forecast model and assimilation system.", "classified_sentences": [ { "sentence": "The economic value of the European Centre for Medium‐Range Weather Forecasts (ECMWF) operational ensemble prediction system (EPS) is assessed relative to the value of a perfect deterministic forecast.", "category": "background" }, { "sentence": "The EPS has substantial relative value throughout the medium range.", "category": "result" }, { "sentence": "Probability forecasts derived from the EPS are of greater benefit than a deterministic forecast produced by the same model.", "category": "result" }, { "sentence": "Indeed, for many users, the probability forecasts have more value than a shorter‐range deterministic forecast.", "category": "result" }, { "sentence": "Based on the measures used here, the additional information in the EPS (reflecting the uncertainty in the initial conditions) provides a benefit to users equivalent to many years' development of the forecast model and assimilation system.", "category": "result" } ] }, { "paper_id": "124417581", "title": "Improvement of ensemble reliability with a new dressing kernel", "abstract": "A new method of combining dynamical and statistical ensembles for the purpose of improving ensemble reliability for underdispersive ensembles is introduced. The method involves adding independent sets of N random four‐dimensional ‘dressing’ perturbations to each of the K members of a dynamical ensemble forecast to obtain an N×K dressed ensemble. The new method mathematically constrains the stochastic process used to generate the statistical dressing perturbations so that it removes seasonally averaged errors in the second moment measures for originally underdispersive ensembles. A random‐number generator experiment and an experiment with the ensemble transform Kalman filter (ETKF) ensemble generation scheme show that the previously proposed ‘best‐member’ dressing method fails to reliably predict the second moment of the distribution of forecast errors, whereas the new dressing method reliably predicts this second moment. After being dressed with the second moment constraint method, the ETKF ensemble is more skilful than the undressed ensemble. The ETKF ensemble postprocessed with the new dressing method is applied for probabilistic forecasts of cooling degree‐days (CDD) for Boston. It is shown that the new kernel's ability to account for temporally correlated forecast errors results in ensemble forecasts of CDDs with reliable spread, whereas the best‐member method leads to an underdispersive ensemble of CDD forecasts. Copyright © 2005 Royal Meteorological Society", "classified_sentences": [ { "sentence": "A new method of combining dynamical and statistical ensembles for the purpose of improving ensemble reliability for underdispersive ensembles is introduced.", "category": "method" }, { "sentence": "The method involves adding independent sets of N random four‐dimensional ‘dressing’ perturbations to each of the K members of a dynamical ensemble forecast to obtain an N×K dressed ensemble.", "category": "method" }, { "sentence": "The new method mathematically constrains the stochastic process used to generate the statistical dressing perturbations so that it removes seasonally averaged errors in the second moment measures for originally underdispersive ensembles.", "category": "method" }, { "sentence": "A random‐number generator experiment and an experiment with the ensemble transform Kalman filter (ETKF) ensemble generation scheme show that the previously proposed ‘best‐member’ dressing method fails to reliably predict the second moment of the distribution of forecast errors, whereas the new dressing method reliably predicts this second moment.", "category": "result" }, { "sentence": "After being dressed with the second moment constraint method, the ETKF ensemble is more skilful than the undressed ensemble.", "category": "result" }, { "sentence": "The ETKF ensemble postprocessed with the new dressing method is applied for probabilistic forecasts of cooling degree‐days (CDD) for Boston.", "category": "result" }, { "sentence": "It is shown that the new kernel's ability to account for temporally correlated forecast errors results in ensemble forecasts of CDDs with reliable spread, whereas the best‐member method leads to an underdispersive ensemble of CDD forecasts.", "category": "result" } ] }, { "paper_id": "259661505", "title": "Classification on Unsupervised Deep Hashing With Pseudo Labels Using Support Vector Machine for Scalable Image Retrieval", "abstract": "The content-based image retrieval (CBIR) method operates on the low-level visual features of the user input query object, which makes it difficult for users to formulate the query and also does not provide adequate retrieval results. In the past, image annotation was suggested as the best possible framework for CBIR, which works on automatically signing keywords to images that support image retrieval. The recent successes of deep learning techniques, especially Convolutional Neural Networks (CNN), in solving computer vision applications have inspired me to work on this paper to solve the problem of CBIR using a dataset of annotated images", "classified_sentences": [ { "sentence": "The content-based image retrieval (CBIR) method operates on the low-level visual features of the user input query object, which makes it difficult for users to formulate the query and also does not provide adequate retrieval results.", "category": "background" }, { "sentence": "In the past, image annotation was suggested as the best possible framework for CBIR, which works on automatically signing keywords to images that support image retrieval.", "category": "background" }, { "sentence": "The recent successes of deep learning techniques, especially Convolutional Neural Networks (CNN), in solving computer vision applications have inspired me to work on this paper to solve the problem of CBIR using a dataset of annotated images", "category": "method" } ] }, { "paper_id": "259922125", "title": "Added value of seasonal hindcasts for UK hydrological drought outlook", "abstract": "The UK has experienced recurring periods of hydrological droughts in the past, including the latest drought declared in summer 10 2022. Seasonal hindcasts, consisting of a large sample of plausible weather sequences, can be used to add value to existing approaches to water resources planning. In this study, the drivers of winter rainfall in the Greater Anglia region are investigated using the ECMWF SEAS5 hindcast dataset, which includes 2850 plausible winters across 25 ensemble members and 3 lead times. Four winter clusters were defined using the hindcast winters based on possible combinations of various atmospheric circulation indices (such as North Atlantic Oscillation, East Atlantic Pattern and the El-Niño Southern Oscillation). Using the 15 2022 drought as a case study, we demonstrate how storylines of the event could be created in autumn 2022 to provide an outlook of drought conditions and to explore plausible worst cases over winter 2022/23 and beyond. The winter clusters span a range of temperature and rainfall response in the Anglian region and represent circulation storylines that could have happened over winter 2022/23. Although winter 2022/23 has now passed, we aim to demonstrate the added value of this approach to provide outlooks during an ongoing event with a brief retrospective of how winter 2022/23 transpired. Storylines created from 20 the hindcast winters were simulated using the GR6J catchment hydrological model and the groundwater level model Aquimod at selected catchments and boreholes in the Anglian region. Results show that drier than average winters characterised by predominantly NAO-/EA- and NAO+/EA- circulation patterns would result in the continuation of the drought with a high likelihood of below normal to low river flows across all selected catchments and boreholes by spring and summer 2023. Catchments in Norfolk are particularly vulnerable to a dry summer in 2023", "classified_sentences": [ { "sentence": "The UK has experienced recurring periods of hydrological droughts in the past, including the latest drought declared in summer 10 2022.", "category": "background" }, { "sentence": "Seasonal hindcasts, consisting of a large sample of plausible weather sequences, can be used to add value to existing approaches to water resources planning.", "category": "background" }, { "sentence": "In this study, the drivers of winter rainfall in the Greater Anglia region are investigated using the ECMWF SEAS5 hindcast dataset, which includes 2850 plausible winters across 25 ensemble members and 3 lead times.", "category": "method" }, { "sentence": "Four winter clusters were defined using the hindcast winters based on possible combinations of various atmospheric circulation indices (such as North Atlantic Oscillation, East Atlantic Pattern and the El-Niño Southern Oscillation).", "category": "method" }, { "sentence": "Using the 15 2022 drought as a case study, we demonstrate how storylines of the event could be created in autumn 2022 to provide an outlook of drought conditions and to explore plausible worst cases over winter 2022/23 and beyond.", "category": "method" }, { "sentence": "The winter clusters span a range of temperature and rainfall response in the Anglian region and represent circulation storylines that could have happened over winter 2022/23.", "category": "method" }, { "sentence": "Although winter 2022/23 has now passed, we aim to demonstrate the added value of this approach to provide outlooks during an ongoing event with a brief retrospective of how winter 2022/23 transpired.", "category": "method" }, { "sentence": "Storylines created from 20 the hindcast winters were simulated using the GR6J catchment hydrological model and the groundwater level model Aquimod at selected catchments and boreholes in the Anglian region.", "category": "method" }, { "sentence": "Results show that drier than average winters characterised by predominantly NAO-/EA- and NAO+/EA- circulation patterns would result in the continuation of the drought with a high likelihood of below normal to low river flows across all selected catchments and boreholes by spring and summer 2023.", "category": "result" }, { "sentence": "Catchments in Norfolk are particularly vulnerable to a dry summer in 2023", "category": "result" } ] }, { "paper_id": "259937082", "title": "Snapshot Spectral Clustering - a costless approach to deep clustering ensembles generation", "abstract": "Despite tremendous advancements in Artificial Intelligence, learning from large sets of data in an unsupervised manner remains a significant challenge. Classical clustering algorithms often fail to discover complex dependencies in large datasets, especially considering sparse, high-dimensional spaces. However, deep learning techniques proved to be successful when dealing with large quantities of data, efficiently reducing their dimensionality without losing track of underlying information. Several interesting advancements have already been made to combine deep learning and clustering. Still, the idea of enhancing the clustering results by combining multiple views of the data generated by deep neural networks appears to be insufficiently explored yet. This paper aims to investigate this direction and bridge the gap between deep neural networks, clustering techniques and ensemble learning methods. To achieve this goal, we propose a novel deep clustering ensemble method - Snapshot Spectral Clustering, designed to maximize the gain from combining multiple data views while minimizing the computational costs of creating the ensemble. Comparative analysis and experiments described in this paper prove the proposed concept, while the conducted hyperparameter study provides a valuable intuition to follow when selecting proper values.", "classified_sentences": [ { "sentence": "Despite tremendous advancements in Artificial Intelligence, learning from large sets of data in an unsupervised manner remains a significant challenge.", "category": "background" }, { "sentence": "Classical clustering algorithms often fail to discover complex dependencies in large datasets, especially considering sparse, high-dimensional spaces.", "category": "background" }, { "sentence": "However, deep learning techniques proved to be successful when dealing with large quantities of data, efficiently reducing their dimensionality without losing track of underlying information.", "category": "background" }, { "sentence": "Several interesting advancements have already been made to combine deep learning and clustering.", "category": "background" }, { "sentence": "Still, the idea of enhancing the clustering results by combining multiple views of the data generated by deep neural networks appears to be insufficiently explored yet.", "category": "background" }, { "sentence": "This paper aims to investigate this direction and bridge the gap between deep neural networks, clustering techniques and ensemble learning methods.", "category": "method" }, { "sentence": "To achieve this goal, we propose a novel deep clustering ensemble method - Snapshot Spectral Clustering, designed to maximize the gain from combining multiple data views while minimizing the computational costs of creating the ensemble.", "category": "method" }, { "sentence": "Comparative analysis and experiments described in this paper prove the proposed concept, while the conducted hyperparameter study provides a valuable intuition to follow when selecting proper values.", "category": "result" } ] }, { "paper_id": "128810012", "title": "Numerical simulations of snowfall events: Sensitivity analysis of physical parameterizations", "abstract": "Accurate estimation of snowfall episodes several hours or even days in advance is essential to minimize risks to transport and other human activities. Every year, these episodes cause severe traffic problems on the northwestern Iberian Peninsula. In order to analyze the influence of different parameterization schemes, 15 snowfall days were analyzed with the Weather Research and Forecasting (WRF) model, defining three nested domains with resolutions of 27, 9, and 3 km. We implemented four microphysical parameterizations (WRF Single‐Moment 6‐class scheme, Goddard, Thompson, and Morrison) and two planetary boundary layer schemes (Yonsei University and Mellor‐Yamada‐Janjic), yielding eight distinct combinations. To validate model estimates, a network of 97 precipitation gauges was used, together with dichotomous data of snowfall presence/absence from snowplow requests to the emergency service of Spain and observatories of the Spanish Meteorological Agency. The results indicate that the most accurate setting of WRF for the study area was that using the Thompson microphysical parameterization and Mellor‐Yamada‐Janjic scheme, although the Thompson and Yonsei University combination had greater accuracy in determining the temporal distribution of precipitation over 1 day. Combining the eight deterministic members in an ensemble average improved results considerably. Further, the root mean square difference decreased markedly using a multiple linear regression as postprocessing. In addition, our method was able to provide mean ensemble precipitation and maximum expected precipitation,which can be very useful in the management of water resources. Finally, we developed an application that allows determination of the risk of snowfall above a certain threshold.", "classified_sentences": [ { "sentence": "Accurate estimation of snowfall episodes several hours or even days in advance is essential to minimize risks to transport and other human activities.", "category": "background" }, { "sentence": "Every year, these episodes cause severe traffic problems on the northwestern Iberian Peninsula.", "category": "background" }, { "sentence": "In order to analyze the influence of different parameterization schemes, 15 snowfall days were analyzed with the Weather Research and Forecasting (WRF) model, defining three nested domains with resolutions of 27, 9, and 3 km.", "category": "method" }, { "sentence": "We implemented four microphysical parameterizations (WRF Single‐Moment 6‐class scheme, Goddard, Thompson, and Morrison) and two planetary boundary layer schemes (Yonsei University and Mellor‐Yamada‐Janjic), yielding eight distinct combinations.", "category": "method" }, { "sentence": "To validate model estimates, a network of 97 precipitation gauges was used, together with dichotomous data of snowfall presence/absence from snowplow requests to the emergency service of Spain and observatories of the Spanish Meteorological Agency.", "category": "method" }, { "sentence": "The results indicate that the most accurate setting of WRF for the study area was that using the Thompson microphysical parameterization and Mellor‐Yamada‐Janjic scheme, although the Thompson and Yonsei University combination had greater accuracy in determining the temporal distribution of precipitation over 1 day.", "category": "result" }, { "sentence": "Combining the eight deterministic members in an ensemble average improved results considerably.", "category": "result" }, { "sentence": "Further, the root mean square difference decreased markedly using a multiple linear regression as postprocessing.", "category": "result" }, { "sentence": "In addition, our method was able to provide mean ensemble precipitation and maximum expected precipitation,which can be very useful in the management of water resources.", "category": "result" }, { "sentence": "Finally, we developed an application that allows determination of the risk of snowfall above a certain threshold.", "category": "result" } ] }, { "paper_id": "128949776", "title": "Scale issues in verification of precipitation forecasts", "abstract": "Precipitation forecasts from numerical weather prediction models are often compared to rain gauge observations to make inferences as to model performance and the “best” resolution needed to accurately capture the structure of observed precipitation. A common approach to quantitative precipitation forecast (QPF) verification is to interpolate the model-predicted areal averages (typically assigned to the center point of the model grid boxes) to the observation sites and compare observed and predicted point values using statistical scores such as bias and RMSE. In such an approach, the fact that the interpolated values and their uncertainty depend on the scale (model resolution) of the values from which the interpolation was done is typically ignored. This interpolation error, which comes from scale effects, is referred to here as the “representativeness error. ” It is a nonzero scale-dependent error even for the case of a perfect model and thus can be seen as independent of model performance. The scale dependency of the representativeness error can have a significant effect on model verification, especially when model performance is judged as a function of grid resolution. An alternative method is to upscale the gauge observations to areal averages and compare at the scale of the model output. Issues of scale arise here too, with a different scale dependency in the representativeness error. This paper examines the merits and limitations of both verification methods (area-to-point and point-to-area) in view of the pronounced spatial variability of precipitation fields and the inherent scale dependency of the representativeness error in each of the verification procedures. A composite method combining the two procedures is introduced and shown to diminish the scale dependency of the representativeness error.", "classified_sentences": [ { "sentence": "Precipitation forecasts from numerical weather prediction models are often compared to rain gauge observations to make inferences as to model performance and the “best” resolution needed to accurately capture the structure of observed precipitation.", "category": "background" }, { "sentence": "A common approach to quantitative precipitation forecast (QPF) verification is to interpolate the model-predicted areal averages (typically assigned to the center point of the model grid boxes) to the observation sites and compare observed and predicted point values using statistical scores such as bias and RMSE.", "category": "method" }, { "sentence": "In such an approach, the fact that the interpolated values and their uncertainty depend on the scale (model resolution) of the values from which the interpolation was done is typically ignored.", "category": "background" }, { "sentence": "This interpolation error, which comes from scale effects, is referred to here as the “representativeness error.”", "category": "background" }, { "sentence": "” It is a nonzero scale-dependent error even for the case of a perfect model and thus can be seen as independent of model performance.", "category": "background" }, { "sentence": "The scale dependency of the representativeness error can have a significant effect on model verification, especially when model performance is judged as a function of grid resolution.", "category": "background" }, { "sentence": "An alternative method is to upscale the gauge observations to areal averages and compare at the scale of the model output.", "category": "method" }, { "sentence": "Issues of scale arise here too, with a different scale dependency in the representativeness error.", "category": "background" }, { "sentence": "This paper examines the merits and limitations of both verification methods (area-to-point and point-to-area) in view of the pronounced spatial variability of precipitation fields and the inherent scale dependency of the representativeness error in each of the verification procedures.", "category": "method" }, { "sentence": "A composite method combining the two procedures is introduced and shown to diminish the scale dependency of the representativeness error.", "category": "result" } ] }, { "paper_id": "265693688", "title": "Rethinking Object Saliency Ranking: A Novel Whole-Flow Processing Paradigm", "abstract": "Existing salient object detection methods are capable of predicting binary maps that highlight visually salient regions. However, these methods are limited in their ability to differentiate the relative importance of multiple objects and the relationships among them, which can lead to errors and reduced accuracy in downstream tasks that depend on the relative importance of multiple objects. To conquer, this paper proposes a new paradigm for saliency ranking, which aims to completely focus on ranking salient objects by their “importance order”. While previous works have shown promising performance, they still face ill-posed problems. First, the saliency ranking ground truth (GT) orders generation methods are unreasonable since determining the correct ranking order is not well-defined, resulting in false alarms. Second, training a ranking model remains challenging because most saliency ranking methods follow the multi-task paradigm, leading to conflicts and trade-offs among different tasks. Third, existing regression-based saliency ranking methods are complex for saliency ranking models due to their reliance on instance mask-based saliency ranking orders. These methods require a significant amount of data to perform accurately and can be challenging to implement effectively. To solve these problems, this paper conducts an in-depth analysis of the causes and proposes a whole-flow processing paradigm of saliency ranking task from the perspective of “GT data generation”, “network structure design” and “training protocol”. The proposed approach outperforms existing state-of-the-art methods on the widely-used SALICON set, as demonstrated by extensive experiments with fair and reasonable comparisons. The saliency ranking task is still in its infancy, and our proposed unified framework can serve as a fundamental strategy to guide future work. The code and data will be available at https://github.com/MengkeSong/Saliency-Ranking-Paradigm.", "classified_sentences": [ { "sentence": "Existing salient object detection methods are capable of predicting binary maps that highlight visually salient regions.", "category": "background" }, { "sentence": "However, these methods are limited in their ability to differentiate the relative importance of multiple objects and the relationships among them, which can lead to errors and reduced accuracy in downstream tasks that depend on the relative importance of multiple objects.", "category": "background" }, { "sentence": "To conquer, this paper proposes a new paradigm for saliency ranking, which aims to completely focus on ranking salient objects by their “importance order”.", "category": "method" }, { "sentence": "While previous works have shown promising performance, they still face ill-posed problems.", "category": "background" }, { "sentence": "First, the saliency ranking ground truth (GT) orders generation methods are unreasonable since determining the correct ranking order is not well-defined, resulting in false alarms.", "category": "background" }, { "sentence": "Second, training a ranking model remains challenging because most saliency ranking methods follow the multi-task paradigm, leading to conflicts and trade-offs among different tasks.", "category": "background" }, { "sentence": "Third, existing regression-based saliency ranking methods are complex for saliency ranking models due to their reliance on instance mask-based saliency ranking orders.", "category": "background" }, { "sentence": "These methods require a significant amount of data to perform accurately and can be challenging to implement effectively.", "category": "background" }, { "sentence": "To solve these problems, this paper conducts an in-depth analysis of the causes and proposes a whole-flow processing paradigm of saliency ranking task from the perspective of “GT data generation”, “network structure design” and “training protocol”.", "category": "method" }, { "sentence": "The proposed approach outperforms existing state-of-the-art methods on the widely-used SALICON set, as demonstrated by extensive experiments with fair and reasonable comparisons.", "category": "result" }, { "sentence": "The saliency ranking task is still in its infancy, and our proposed unified framework can serve as a fundamental strategy to guide future work.", "category": "result" }, { "sentence": "The code and data will be available at https://github.com/MengkeSong/Saliency-Ranking-Paradigm.", "category": "result" } ] }, { "paper_id": "133647636", "title": "Ensemble air quality forecasts over the Lower Fraser Valley, British Columbia: a summer 2004 case study", "abstract": "The ensemble-averaging approach is potentially a technique for improving the performance of real-time photochemical airquality modeling. Ensemble photochemical airquality forecasts are tested extensively using the Community Multiscale Air Quality (CMAQ) model-system with mesonet observations from the Emergency Weather Net (EmWxNet) and the Quality-Controlled AQ Data Set over the Lower Fraser Valley (LFV). The CMAQ model is run daily over a 12 km resolution domain (Figure 1 top) covering southern British Columbia, Washington State, and the northern portion of Oregon State. A 4 km resolution grid (Figure 1 bottom) is nested within the 12km grid, and it covers the southern tip of British Columbia (including the LFV) and the northern part of Washington State (including the Seattle area). CMAQ is driven by two different meteorological models: the Mesoscale Compressible Community Model (MC2), and the Fifth-Generation NCAR / Penn State Mesoscale Model (MM5).", "classified_sentences": [ { "sentence": "The ensemble-averaging approach is potentially a technique for improving the performance of real-time photochemical airquality modeling.", "category": "method" }, { "sentence": "Ensemble photochemical airquality forecasts are tested extensively using the Community Multiscale Air Quality (CMAQ) model-system with mesonet observations from the Emergency Weather Net (EmWxNet) and the Quality-Controlled AQ Data Set over the Lower Fraser Valley (LFV).", "category": "method" }, { "sentence": "The CMAQ model is run daily over a 12 km resolution domain (Figure 1 top) covering southern British Columbia, Washington State, and the northern portion of Oregon State.", "category": "method" }, { "sentence": "A 4 km resolution grid (Figure 1 bottom) is nested within the 12km grid, and it covers the southern tip of British Columbia (including the LFV) and the northern part of Washington State (including the Seattle area).", "category": "method" }, { "sentence": "CMAQ is driven by two different meteorological models: the Mesoscale Compressible Community Model (MC2), and the Fifth-Generation NCAR / Penn State Mesoscale Model (MM5).", "category": "method" } ] }, { "paper_id": "368524", "title": "Squibs: Prepositional Phrase Attachment without Oracles", "abstract": "Work on prepositional phrase (PP) attachment resolution generally assumes that there is an oracle that provides the two hypothesized structures that we want to choose between. The information that there are two possible attachment sites and the information about the lexical heads of those phrases is usually extracted from gold-standard parse trees. We show that the performance of reattachment methods is higher with such an oracle than without. Because oracles are not available in NLP applications, this indicates that the current evaluation methodology for PP attachment does not produce realistic performance numbers. We argue that PP attachment should not be evaluated in isolation, but instead as an integral component of a parsing system, without using information from the gold-standard oracle.", "classified_sentences": [ { "sentence": "Work on prepositional phrase (PP) attachment resolution generally assumes that there is an oracle that provides the two hypothesized structures that we want to choose between.", "category": "background" }, { "sentence": "The information that there are two possible attachment sites and the information about the lexical heads of those phrases is usually extracted from gold-standard parse trees.", "category": "background" }, { "sentence": "We show that the performance of reattachment methods is higher with such an oracle than without.", "category": "result" }, { "sentence": "Because oracles are not available in NLP applications, this indicates that the current evaluation methodology for PP attachment does not produce realistic performance numbers.", "category": "result" }, { "sentence": "We argue that PP attachment should not be evaluated in isolation, but instead as an integral component of a parsing system, without using information from the gold-standard oracle.", "category": "method" } ] }, { "paper_id": "1213698", "title": "Bag-of-Words Against Nearest-Neighbor Search for Visual Object Retrieval", "abstract": "We compare the Bag-of-Words (BoW) framework with the Approximate Nearest-Neighbor (ANN) based system in the context of visual object retrieval. This comparison is motivated by the implicit connection between these two methods: generally speaking, the BoW framework can be regarded as a quantization-guided ANN voting system. The value of establishing such comparison lies in: first, by comparing with other quantization-free ANN system, the performance loss caused by the quantization error in the BoW framework can be estimated quantitatively. Second, this comparison completely inspects the pros and cons of both ANN and BoW methods, thus to facilitate new algorithm design. In this study, by taking an independent dataset as the reference to validate matches, we design an ANN voting system that outperforms all other methods. Comprehensive and computationally intensive experiments are conducted on two Oxford datasets and two TrecVid instance search datasets, and the new state-of-the-art is achieved.", "classified_sentences": [ { "sentence": "We compare the Bag-of-Words (BoW) framework with the Approximate Nearest-Neighbor (ANN) based system in the context of visual object retrieval.", "category": "method" }, { "sentence": "This comparison is motivated by the implicit connection between these two methods: generally speaking, the BoW framework can be regarded as a quantization-guided ANN voting system.", "category": "background" }, { "sentence": "The value of establishing such comparison lies in: first, by comparing with other quantization-free ANN system, the performance loss caused by the quantization error in the BoW framework can be estimated quantitatively.", "category": "background" }, { "sentence": "Second, this comparison completely inspects the pros and cons of both ANN and BoW methods, thus to facilitate new algorithm design.", "category": "background" }, { "sentence": "In this study, by taking an independent dataset as the reference to validate matches, we design an ANN voting system that outperforms all other methods.", "category": "method" }, { "sentence": "Comprehensive and computationally intensive experiments are conducted on two Oxford datasets and two TrecVid instance search datasets, and the new state-of-the-art is achieved.", "category": "result" } ] }, { "paper_id": "2256138", "title": "Isocentric color saliency in images", "abstract": "In this paper we propose a novel computational method to infer visual saliency in images. The computational method is based on the idea that salient objects should have local characteristics that are different than the rest of the scene, being edges, color or shape, and that these characteristics can be combined to infer global information. The proposed approach is fast, does not require any learning and the experimentation shows that it can enhance interesting objects in images, improving the state of the art performance on a public dataset.", "classified_sentences": [ { "sentence": "In this paper we propose a novel computational method to infer visual saliency in images.", "category": "method" }, { "sentence": "The computational method is based on the idea that salient objects should have local characteristics that are different than the rest of the scene, being edges, color or shape, and that these characteristics can be combined to infer global information.", "category": "method" }, { "sentence": "The proposed approach is fast, does not require any learning and the experimentation shows that it can enhance interesting objects in images, improving the state of the art performance on a public dataset.", "category": "result" } ] }, { "paper_id": "5060957", "title": "Task-oriented Evaluation of Syntactic Parsers and Their Representations", "abstract": "This paper presents a comparative evaluation of several state-of-the-art English parsers based on different frameworks. Our approach is to measure the impact of each parser when it is used as a component of an information extraction system that performs protein-protein interaction (PPI) identification in biomedical papers. We evaluate eight parsers (based on dependency parsing, phrase structure parsing, or deep parsing) using five different parse representations. We run a PPI system with several combinations of parser and parse representation, and examine their impact on PPI identification accuracy. Our experiments show that the levels of accuracy obtained with these different parsers are similar, but that accuracy improvements vary when the parsers are retrained with domain-specific data.", "classified_sentences": [ { "sentence": "This paper presents a comparative evaluation of several state-of-the-art English parsers based on different frameworks.", "category": "background" }, { "sentence": "Our approach is to measure the impact of each parser when it is used as a component of an information extraction system that performs protein-protein interaction (PPI) identification in biomedical papers.", "category": "method" }, { "sentence": "We evaluate eight parsers (based on dependency parsing, phrase structure parsing, or deep parsing) using five different parse representations.", "category": "method" }, { "sentence": "We run a PPI system with several combinations of parser and parse representation, and examine their impact on PPI identification accuracy.", "category": "method" }, { "sentence": "Our experiments show that the levels of accuracy obtained with these different parsers are similar, but that accuracy improvements vary when the parsers are retrained with domain-specific data.", "category": "result" } ] }, { "paper_id": "9826412", "title": "Saliency histogram equalisation and its application to image resizing", "abstract": "This study presents a novel method for content-aware image resizing based on the normalised saliency map of an image termed saliency histogram (SH), which provides the probability of a target's presence in the spatial location domain. Motivated by the idea of spreading the histogram of an image on the grey scale known as image histogram equalisation for contrast enhancement, fast SH equalisation (SHE) has been proposed to distribute the content of an image in the resized image domain similarly. Experimental results show that SHE has a tendency to protect salient foreground objects, and it takes only fractions of a second to resize an 1024 × 768 image.", "classified_sentences": [ { "sentence": "This study presents a novel method for content-aware image resizing based on the normalised saliency map of an image termed saliency histogram (SH), which provides the probability of a target's presence in the spatial location domain.", "category": "method" }, { "sentence": "Motivated by the idea of spreading the histogram of an image on the grey scale known as image histogram equalisation for contrast enhancement, fast SH equalisation (SHE) has been proposed to distribute the content of an image in the resized image domain similarly.", "category": "method" }, { "sentence": "Experimental results show that SHE has a tendency to protect salient foreground objects, and it takes only fractions of a second to resize an 1024 × 768 image.", "category": "result" } ] }, { "paper_id": "10530060", "title": "Probability models for Visual Search Literature Survey", "abstract": "The development of efficient artificial machine vision systems depends on the ability to mimic aspects of the human visual system. Humans scan the world using a highresolution central region called the fovea and a low resolution surrounding area to guide the search. A direct consequence of this non-uniform sampling is the active nature in which human visual system gathers data in the real world using fixations and saccades. In this report, we look at a few techniques that attempt to mimic the visual search strategies of the human visual system. 1.INTRODUCTION The human visual system uses a dynamic process of actively scanning the visual environment. The active nature of scanning is reflected in the eye scanpath pattern. These sequence of fixations and saccades (constituting the scanpaths) are attributed to the distribution of the photoreceptors on the retina. The photoreceptors are packed densely at the point of focus on the retina fovea and the sampling rate drops almost exponentially from the fovea. Fig 1 shows a typical retinal sampling grid. As a result, humans see with very high resolution at the fixation point and the resolution falls away from the fixation point. Fig 2 shows a typical image on the retina the fixation point being the middle of the image. In order to build a detailed representation of the image, the eye scans the scene with a series of fixations and jumps (saccades) to new fixation points. Information is gathered by the eye during fixations while no information is gathered during the saccades. The fixation duration is about 200ms. Fig 3 shows a typical scan path of the human eye while looking at the image [1]. The active nature of looking has its advantages in terms of speed and reduced storage requirements (due to the non-uniform resolution across the image) in building artificial vision systems. It also have significant applications in the area of video compression where the region around the fixation point in the video sequence is transmitted with high resolution and regions away from the fixation point are blurred. The development of foveation based artificial vision systems and video compression schemes depends on the ability to determine the fixation points/area of interest regions in the image. However, in general, we cannot predict a person’s scanpath while viewing a scene in a realistic way. One common solution to determine the eye scan path is the use of eye trackers. An alternative solution is to develop models for the fixation problem. Since deterministic solution to the fixation point prediction problem is impossible (different people look at the same image using different scan paths based on the motive), I propose to investigate the possibility of building a probabilistic model for eye fixations in a visual search environment. Besides the applications already mentioned, the development of such a fixation model has significant applications in computer vision applications such as pictorial image database query and image understanding. 2. Previous models for fixation point selection The primary goal of many machine vision systems has been the development of algorithms that interpret visual data from cameras to help computers to see. Most of the active vision systems developed are developed for a specific task and hence perform only in constrained scenarios. In this section we will briefly go over three such techniques. 1. Image features and fixations Privetera and Stark [2] propose a computational model for human scanpaths based on intelligent image processing of digital images. The basic idea is to define algorithmic regions of interest (aROI) generated by the image processing algorithms and compare the result with human regions of interest (hROI). The comparison of the aROI and hROI is accomplished by analyzing their spatial/structural binding (location similarity) and temporal/sequential binding (order of fixations). Based on their experiments, the aROI generated by wavelet decomposition of the image (which is inherently multi-resolution) seemed to match the hROIs well. Symmetry and contrast also seemed to be strongly correlated with fixation. Their results also indicate that the fixation point prediction can be no better 50% i.e. only half the predictions made are accurate. While the results of this paper are definitely promising, the techniques to determine fixation points do not seem to account for the fact that the next fixation point selection is dependent on the current fixation point. Further, a weighted result of using multiple image processing algorithms might produce better prediction of aROIs. 2. Probability models Klarquist and Bovik [3] propose an alternative technique for fixation point selection in 3D space. The fixation point selection was developed for FOVEA \"an active vision system platform with capabilities similar to sophisticated biological vision systems\"[3]. FOVEA uses a probabilistic approach to fixation point selection and hence makes the selection of the fixation point less rigid and also contingent on the features around the current fixation point. The probability model is developed using a number of criteria. The fixation point selection process is independent of the criteria and hence creates a clear dichotomy between the selection criterion and the selection process. The selection criterion is based on local information content (gradient information), proximity of the candidate fixation point to the current fixation point and the surface map in the vicinity of the current fixation point. However no indication of the performance of their system with human scanpaths is provided. 3. Saliency models for image understanding Henderson [4] proposes a more robust method towards fixation point selection in images. The model incorporates the cognition factor involved in fixation point selection. The initial fixation map is derived by analyzing low-level features (contrast, edges) in the image. Based on the task at hand (search for a target), the model is trained to \"understand\" the image. Incorporating cognition into a model is a difficult task since cognition is task specific. However the proposed model facilitates both the prediction of the fixation points and the duration of fixations. 3. Probability models for Visual Search For my project, I propose to investigate probability models in a constrained visual search environment. Human subjects will be presented with a target that he/she will need to search in a scene. The eye movements will be recorded using the Model 504 remote eye tracker from Applied Science Laboratories (ASL). The scanpaths will be analyzed for extracting fixation points and saccades. The region around the fixation points will then be analyzed to understand the similarity between the target and region the eye fixated on. The analysis of the similarity will be based on correlation, edge co-occurance and eigen decomposition techniques to name a few. This data will be analyzed to determine \"attractors\" features that force the eye to fixate around a point as opposed to other points. This result will then be extended to determine a probability fixation model based on density of \"attractors\" in the image. These results will then be tested with human scan path data. The experiments will be conducted using the MATLAB psychophysics toolbox and ASL provided software for scanpath analysis. 4.Conclusion Determination of fixation points is an important step in applications that use foveation strategies. Development of a probability model in a constrained \"visual search\" environment is proposed. This work will be extended in future to the \"visual surveillance\" which differs from visual search in that there is not target to be found and hence generalizes to the case of finding fixation points in an arbitrary scene. Figure 1:Sampling structure in the retina Figure 2A: Original uniform resolution image of Lena Figure 2B:Foveated image of Lena 4. References 1. A. L. Yarbus, Eye movements and Vision New York:Plenum Press, 1967 2. C. M. Privitera and L. W. Stark, \"Algorithms for Defining Visual Regions of Interest: Comparison with Eye Fixations,\" IEEE Tran. On Pattern Analysis and Machine Intelligence, September 2000, Vol 22, No 9, Pg. 970-982 3. W. N. Klarquist and A. C. Bovik, \"FOVEA: A Foveated Vergent Active Stereo Vision System for Dynamic Three-Dimensional Scene Recovery,\" IEEE Trans on Robotics and Automation, October 1998,Vol 14,No 5, Pg 755-770 4. J. M. Henderson, \"Eye movement control during visual object processing: Effects of intial fixation position and semantic costraint\" ,Journal of Experimental Psychology, 1993, Vol 47, Pg 79-98 Figure 3: Fixation and saccades in viewing an image [1]", "classified_sentences": [ { "sentence": "The development of efficient artificial machine vision systems depends on the ability to mimic aspects of the human visual system.", "category": "background" }, { "sentence": "Humans scan the world using a highresolution central region called the fovea and a low resolution surrounding area to guide the search.", "category": "background" }, { "sentence": "A direct consequence of this non-uniform sampling is the active nature in which human visual system gathers data in the real world using fixations and saccades.", "category": "background" }, { "sentence": "In this report, we look at a few techniques that attempt to mimic the visual search strategies of the human visual system.", "category": "method" }, { "sentence": "1.INTRODUCTION The human visual system uses a dynamic process of actively scanning the visual environment.", "category": "background" }, { "sentence": "The active nature of scanning is reflected in the eye scanpath pattern.", "category": "background" }, { "sentence": "These sequence of fixations and saccades (constituting the scanpaths) are attributed to the distribution of the photoreceptors on the retina.", "category": "background" }, { "sentence": "The photoreceptors are packed densely at the point of focus on the retina fovea and the sampling rate drops almost exponentially from the fovea.", "category": "background" }, { "sentence": "Fig 1 shows a typical retinal sampling grid.", "category": "method" }, { "sentence": "As a result, humans see with very high resolution at the fixation point and the resolution falls away from the fixation point.", "category": "background" }, { "sentence": "Fig 2 shows a typical image on the retina the fixation point being the middle of the image.", "category": "method" }, { "sentence": "In order to build a detailed representation of the image, the eye scans the scene with a series of fixations and jumps (saccades) to new fixation points.", "category": "background" }, { "sentence": "Information is gathered by the eye during fixations while no information is gathered during the saccades.", "category": "background" }, { "sentence": "The fixation duration is about 200ms.", "category": "background" }, { "sentence": "Fig 3 shows a typical scan path of the human eye while looking at the image [1].", "category": "method" }, { "sentence": "The active nature of looking has its advantages in terms of speed and reduced storage requirements (due to the non-uniform resolution across the image) in building artificial vision systems.", "category": "background" }, { "sentence": "It also have significant applications in the area of video compression where the region around the fixation point in the video sequence is transmitted with high resolution and regions away from the fixation point are blurred.", "category": "background" }, { "sentence": "The development of foveation based artificial vision systems and video compression schemes depends on the ability to determine the fixation points/area of interest regions in the image.", "category": "method" }, { "sentence": "However, in general, we cannot predict a person’s scanpath while viewing a scene in a realistic way.", "category": "background" }, { "sentence": "One common solution to determine the eye scan path is the use of eye trackers.", "category": "method" }, { "sentence": "An alternative solution is to develop models for the fixation problem.", "category": "method" }, { "sentence": "Since deterministic solution to the fixation point prediction problem is impossible (different people look at the same image using different scan paths based on the motive), I propose to investigate the possibility of building a probabilistic model for eye fixations in a visual search environment.", "category": "method" }, { "sentence": "Besides the applications already mentioned, the development of such a fixation model has significant applications in computer vision applications such as pictorial image database query and image understanding.", "category": "background" }, { "sentence": "2.", "category": "background" }, { "sentence": "Previous models for fixation point selection The primary goal of many machine vision systems has been the development of algorithms that interpret visual data from cameras to help computers to see.", "category": "background" }, { "sentence": "Most of the active vision systems developed are developed for a specific task and hence perform only in constrained scenarios.", "category": "background" }, { "sentence": "In this section we will briefly go over three such techniques.", "category": "method" }, { "sentence": "1.", "category": "background" }, { "sentence": "Image features and fixations Privitera and Stark [2] propose a computational model for human scanpaths based on intelligent image processing of digital images.", "category": "method" }, { "sentence": "The basic idea is to define algorithmic regions of interest (aROI) generated by the image processing algorithms and compare the result with human regions of interest (hROI).", "category": "method" }, { "sentence": "The comparison of the aROI and hROI is accomplished by analyzing their spatial/structural binding (location similarity) and temporal/sequential binding (order of fixations).", "category": "method" }, { "sentence": "Based on their experiments, the aROI generated by wavelet decomposition of the image (which is inherently multi-resolution) seemed to match the hROIs well.", "category": "method" }, { "sentence": "Symmetry and contrast also seemed to be strongly correlated with fixation.", "category": "method" }, { "sentence": "Their results also indicate that the fixation point prediction can be no better 50% i.e. only half the predictions made are accurate.", "category": "result" }, { "sentence": "While the results of this paper are definitely promising, the techniques to determine fixation points do not seem to account for the fact that the next fixation point selection is dependent on the current fixation point.", "category": "background" }, { "sentence": "Further, a weighted result of using multiple image processing algorithms might produce better prediction of aROIs.", "category": "background" }, { "sentence": "2.", "category": "background" }, { "sentence": "Probability models Klarquist and Bovik [3] propose an alternative technique for fixation point selection in 3D space.", "category": "method" }, { "sentence": "The fixation point selection was developed for FOVEA \"an active vision system platform with capabilities similar to sophisticated biological vision systems\"[3].", "category": "method" }, { "sentence": "FOVEA uses a probabilistic approach to fixation point selection and hence makes the selection of the fixation point less rigid and also contingent on the features around the current fixation point.", "category": "method" }, { "sentence": "The probability model is developed using a number of criteria.", "category": "method" }, { "sentence": "The fixation point selection process is independent of the criteria and hence creates a clear dichotomy between the selection criterion and the selection process.", "category": "method" }, { "sentence": "The selection criterion is based on local information content (gradient information), proximity of the candidate fixation point to the current fixation point and the surface map in the vicinity of the current fixation point.", "category": "method" }, { "sentence": "However no indication of the performance of their system with human scanpaths is provided.", "category": "background" }, { "sentence": "3.", "category": "background" }, { "sentence": "Saliency models for image understanding Henderson [4] proposes a more robust method towards fixation point selection in images.", "category": "method" }, { "sentence": "The model incorporates the cognition factor involved in fixation point selection.", "category": "method" }, { "sentence": "The initial fixation map is derived by analyzing low-level features (contrast, edges) in the image.", "category": "method" }, { "sentence": "Based on the task at hand (search for a target), the model is trained to \"understand\" the image.", "category": "method" }, { "sentence": "Incorporating cognition into a model is a difficult task since cognition is task specific.", "category": "background" }, { "sentence": "However the proposed model facilitates both the prediction of the fixation points and the duration of fixations.", "category": "result" }, { "sentence": "3.", "category": "background" }, { "sentence": "Probability models for Visual Search For my project, I propose to investigate probability models in a constrained visual search environment.", "category": "method" }, { "sentence": "Human subjects will be presented with a target that he/she will need to search in a scene.", "category": "method" }, { "sentence": "The eye movements will be recorded using the Model 504 remote eye tracker from Applied Science Laboratories (ASL).", "category": "method" }, { "sentence": "The scanpaths will be analyzed for extracting fixation points and saccades.", "category": "method" }, { "sentence": "The region around the fixation points will then be analyzed to understand the similarity between the target and region the eye fixated on.", "category": "method" }, { "sentence": "The analysis of the similarity will be based on correlation, edge co-occurance and eigen decomposition techniques to name a few.", "category": "method" }, { "sentence": "This data will be analyzed to determine \"attractors\" features that force the eye to fixate around a point as opposed to other points.", "category": "method" }, { "sentence": "This result will then be extended to determine a probability fixation model based on density of \"attractors\" in the image.", "category": "method" }, { "sentence": "These results will then be tested with human scan path data.", "category": "method" }, { "sentence": "The experiments will be conducted using the MATLAB psychophysics toolbox and ASL provided software for scanpath analysis.", "category": "method" }, { "sentence": "4.Conclusion Determination of fixation points is an important step in applications that use foveation strategies.", "category": "background" }, { "sentence": "Development of a probability model in a constrained \"visual search\" environment is proposed.", "category": "method" }, { "sentence": "This work will be extended in future to the \"visual surveillance\" which differs from visual search in that there is not target to be found and hence generalizes to the case of finding fixation points in an arbitrary scene.", "category": "method" }, { "sentence": "Figure 1:Sampling structure in the retina", "category": "method" }, { "sentence": "Figure 2A: Original uniform resolution image of Lena", "category": "method" }, { "sentence": "Figure 2B:Foveated image of Lena", "category": "method" }, { "sentence": "Figure 3: Fixation and saccades in viewing an image [1]", "category": "method" }, { "sentence": "References 1.", "category": "background" }, { "sentence": "A.", "category": "background" }, { "sentence": "L.", "category": "background" }, { "sentence": "Yarbus, Eye movements and Vision New York:Plenum Press, 1967", "category": "background" }, { "sentence": "2.", "category": "background" }, { "sentence": "C.", "category": "background" }, { "sentence": "M.", "category": "background" }, { "sentence": "Privitera and L.", "category": "background" }, { "sentence": "W.", "category": "background" }, { "sentence": "Stark, \"Algorithms for Defining Visual Regions of Interest: Comparison with Eye Fixations,\" IEEE Tran.", "category": "background" }, { "sentence": "On Pattern Analysis and Machine Intelligence, September 2000, Vol 22, No 9, Pg.", "category": "background" }, { "sentence": "970-982", "category": "background" }, { "sentence": "3.", "category": "background" }, { "sentence": "W.", "category": "background" }, { "sentence": "N.", "category": "background" }, { "sentence": "Klarquist and A.", "category": "background" }, { "sentence": "C.", "category": "background" }, { "sentence": "Bovik, \"FOVEA: A Foveated Vergent Active Stereo Vision System for Dynamic Three-Dimensional Scene Recovery,\" IEEE Trans on Robotics and Automation, October 1998,Vol 14,No 5, Pg 755-770", "category": "background" }, { "sentence": "4.", "category": "background" }, { "sentence": "J. M.", "category": "background" }, { "sentence": "Henderson, \"Eye movement control during visual object processing: Effects of intial fixation position and semantic costraint\" ,Journal of Experimental Psychology, 1993, Vol 47, Pg 79-98", "category": "background" } ] }, { "paper_id": "10576783", "title": "Efficient descriptor learning for large scale localization", "abstract": "Many robotics and Augmented Reality (AR) systems that use sparse keypoint-based visual maps operate in large and highly repetitive environments, where pose tracking and localization are challenging tasks. Additionally, these systems usually face further challenges, such as limited computational power, or insufficient memory for storing large maps of the entire environment. Thus, developing compact map representations and improving retrieval is of considerable interest for enabling large-scale visual place recognition and loop-closure. In this paper, we propose a novel approach to compress descriptors while increasing their discriminability and match-ability, based on recent advances in neural networks. At the same time, we target resource-constrained robotics applications in our design choices. The main contributions of this work are twofold. First, we propose a linear projection from descriptor space to a lower-dimensional Euclidean space, based on a novel supervised learning strategy employing a triplet loss. Second, we show the importance of including contextual appearance information to the visual feature in order to improve matching under strong viewpoint, illumination and scene changes. Through detailed experiments on three challenging datasets, we demonstrate significant gains in performance over state-of-the-art methods.", "classified_sentences": [ { "sentence": "Many robotics and Augmented Reality (AR) systems that use sparse keypoint-based visual maps operate in large and highly repetitive environments, where pose tracking and localization are challenging tasks.", "category": "background" }, { "sentence": "Additionally, these systems usually face further challenges, such as limited computational power, or insufficient memory for storing large maps of the entire environment.", "category": "background" }, { "sentence": "Thus, developing compact map representations and improving retrieval is of considerable interest for enabling large-scale visual place recognition and loop-closure.", "category": "background" }, { "sentence": "In this paper, we propose a novel approach to compress descriptors while increasing their discriminability and match-ability, based on recent advances in neural networks.", "category": "method" }, { "sentence": "At the same time, we target resource-constrained robotics applications in our design choices.", "category": "method" }, { "sentence": "The main contributions of this work are twofold.", "category": "method" }, { "sentence": "First, we propose a linear projection from descriptor space to a lower-dimensional Euclidean space, based on a novel supervised learning strategy employing a triplet loss.", "category": "method" }, { "sentence": "Second, we show the importance of including contextual appearance information to the visual feature in order to improve matching under strong viewpoint, illumination and scene changes.", "category": "method" }, { "sentence": "Through detailed experiments on three challenging datasets, we demonstrate significant gains in performance over state-of-the-art methods.", "category": "result" } ] }, { "paper_id": "16266526", "title": "Visual saliency detection based on topographic independent component analysis", "abstract": "A computational model of visual saliency detection is proposed based on topographic independent component analysis. This model consists of three steps: first training basis functions and extracting features which represents complex cell responses by topographic independent component analysis, then estimating feature distributions, and finally calculating self-information and obtaining saliency maps. It is demonstrated by numerical examples that the proposed model could detect saliency regions in some circumstances when previous related model couldn't, and predict human attention fixations better than other models.", "classified_sentences": [ { "sentence": "A computational model of visual saliency detection is proposed based on topographic independent component analysis.", "category": "method" }, { "sentence": "This model consists of three steps: first training basis functions and extracting features which represents complex cell responses by topographic independent component analysis, then estimating feature distributions, and finally calculating self-information and obtaining saliency maps.", "category": "method" }, { "sentence": "It is demonstrated by numerical examples that the proposed model could detect saliency regions in some circumstances when previous related model couldn't, and predict human attention fixations better than other models.", "category": "result" } ] }, { "paper_id": "18796136", "title": "Attachment : Where do We Stand ?", "abstract": "Prepositional phrase (PP) attachment is a well known challenge to parsing. In this paper, we combine the insights of different works, namely: (1) treating PP attachment as a classification task with an arbitrary number of attachment candidates; (2) using auxiliary distributions to augment the data beyond the hand-annotated training set; (3) using topological fields to get information about the distribution of PP attachment throughout clauses and (4) using state-of-the-art techniques such as word embeddings and neural networks. We show that jointly using these techniques leads to substantial improvements. We also conduct a qualitative analysis to gauge where the ceiling of the task is in a realistic setup.", "classified_sentences": [ { "sentence": "Prepositional phrase (PP) attachment is a well known challenge to parsing.", "category": "background" }, { "sentence": "In this paper, we combine the insights of different works, namely: (1) treating PP attachment as a classification task with an arbitrary number of attachment candidates; (2) using auxiliary distributions to augment the data beyond the hand-annotated training set; (3) using topological fields to get information about the distribution of PP attachment throughout clauses and (4) using state-of-the-art techniques such as word embeddings and neural networks.", "category": "method" }, { "sentence": "We show that jointly using these techniques leads to substantial improvements.", "category": "result" }, { "sentence": "We also conduct a qualitative analysis to gauge where the ceiling of the task is in a realistic setup.", "category": "method" } ] }, { "paper_id": "162168418", "title": "Deep Reinforcement Learning for Detecting Malicious Websites", "abstract": "Phishing is the simplest form of cybercrime with the objective of baiting people into giving away delicate information such as individually recognizable data, banking and credit card details, or even credentials and passwords. This type of simple yet most effective cyber-attack is usually launched through emails, phone calls, or instant messages. The credential or private data stolen are then used to get access to critical records of the victims and can result in extensive fraud and monetary loss. Hence, sending malicious messages to victims is a stepping stone of the phishing procedure. A \\textit{phisher} usually setups a deceptive website, where the victims are conned into entering credentials and sensitive information. It is therefore important to detect these types of malicious websites before causing any harmful damages to victims. Inspired by the evolving nature of the phishing websites, this paper introduces a novel approach based on deep reinforcement learning to model and detect malicious URLs. The proposed model is capable of adapting to the dynamic behavior of the phishing websites and thus learn the features associated with phishing website detection.", "classified_sentences": [ { "sentence": "Phishing is the simplest form of cybercrime with the objective of baiting people into giving away delicate information such as individually recognizable data, banking and credit card details, or even credentials and passwords.", "category": "background" }, { "sentence": "This type of simple yet most effective cyber-attack is usually launched through emails, phone calls, or instant messages.", "category": "background" }, { "sentence": "The credential or private data stolen are then used to get access to critical records of the victims and can result in extensive fraud and monetary loss.", "category": "background" }, { "sentence": "Hence, sending malicious messages to victims is a stepping stone of the phishing procedure.", "category": "background" }, { "sentence": "A \\textit{phisher} usually setups a deceptive website, where the victims are conned into entering credentials and sensitive information.", "category": "background" }, { "sentence": "It is therefore important to detect these types of malicious websites before causing any harmful damages to victims.", "category": "background" }, { "sentence": "Inspired by the evolving nature of the phishing websites, this paper introduces a novel approach based on deep reinforcement learning to model and detect malicious URLs.", "category": "method" }, { "sentence": "The proposed model is capable of adapting to the dynamic behavior of the phishing websites and thus learn the features associated with phishing website detection.", "category": "method" } ] }, { "paper_id": "202539157", "title": "Detecting Deep Neural Network Defects with Data Flow Analysis", "abstract": "Deep neural networks (DNNs) are shown to be promising solutions in many challenging artificial intelligence tasks, including object recognition, natural language processing, and even unmanned driving. A DNN model, generally based on statistical summarization of in-house training data, aims to predict correct output given an input encountered in the wild. In general, 100% precision is therefore impossible due to its probabilistic nature. For DNN practitioners, it is very hard, if not impossible, to figure out whether the low precision of a DNN model is an inevitable result, or caused by defects such as bad network design or improper training process. This paper aims at addressing this challenging problem. We approach with a careful categorization of the root causes of low precision. We find that the internal data flow footprints of a DNN model can provide insights to locate the root cause effectively. We then develop a tool, namely, DeepMorph (DNN Tomography) to analyze the root cause, which can instantly guide a DNN developer to improve the model. Case studies on four popular datasets show the effectiveness of DeepMorph.", "classified_sentences": [ { "sentence": "Deep neural networks (DNNs) are shown to be promising solutions in many challenging artificial intelligence tasks, including object recognition, natural language processing, and even unmanned driving.", "category": "background" }, { "sentence": "A DNN model, generally based on statistical summarization of in-house training data, aims to predict correct output given an input encountered in the wild.", "category": "background" }, { "sentence": "In general, 100% precision is therefore impossible due to its probabilistic nature.", "category": "background" }, { "sentence": "For DNN practitioners, it is very hard, if not impossible, to figure out whether the low precision of a DNN model is an inevitable result, or caused by defects such as bad network design or improper training process.", "category": "background" }, { "sentence": "This paper aims at addressing this challenging problem.", "category": "method" }, { "sentence": "We approach with a careful categorization of the root causes of low precision.", "category": "method" }, { "sentence": "We find that the internal data flow footprints of a DNN model can provide insights to locate the root cause effectively.", "category": "method" }, { "sentence": "We then develop a tool, namely, DeepMorph (DNN Tomography) to analyze the root cause, which can instantly guide a DNN developer to improve the model.", "category": "method" }, { "sentence": "Case studies on four popular datasets show the effectiveness of DeepMorph.", "category": "result" } ] }, { "paper_id": "207831073", "title": "Phishing URL Detection via CNN and Attention-Based Hierarchical RNN", "abstract": "Phishing websites have long been a serious threat to cyber security. For decades, many researchers have been devoted to developing novel techniques to detect phishing websites automatically. While state-of-the-art solutions can achieve superior performances, they require substantial manual feature engineering and are not adept at detecting newly emerging phishing attacks. Therefore, developing techniques that can detect phishing websites automatically and handle zero-day phishing attacks swiftly is still an open challenge in this area. In this work, we propose PhishingNet, a deep learning-based approach for timely detection of phishing Uniform Resource Locators (URLs). Specifically, we use a Convolutional Neural Network (CNN) module to extract character-level spatial feature representations of URLs; meanwhile, we employ an attention-based hierarchical Recurrent Neural Network(RNN) module to extract word-level temporal feature representations of URLs. We then fuse these feature representations via a three-layer CNN to build accurate feature representations of URLs, on which we train a phishing URL classifier. Extensive experiments on a verified dataset collected from the Internet demonstrate that the feature representations extracted automatically are conducive to the improvement of the generalization ability of our approach on newly emerging URLs, which makes our approach achieve competitive performance against other state-of-the-art approaches.", "classified_sentences": [ { "sentence": "Phishing websites have long been a serious threat to cyber security.", "category": "background" }, { "sentence": "For decades, many researchers have been devoted to developing novel techniques to detect phishing websites automatically.", "category": "background" }, { "sentence": "While state-of-the-art solutions can achieve superior performances, they require substantial manual feature engineering and are not adept at detecting newly emerging phishing attacks.", "category": "background" }, { "sentence": "Therefore, developing techniques that can detect phishing websites automatically and handle zero-day phishing attacks swiftly is still an open challenge in this area.", "category": "background" }, { "sentence": "In this work, we propose PhishingNet, a deep learning-based approach for timely detection of phishing Uniform Resource Locators (URLs).", "category": "method" }, { "sentence": "Specifically, we use a Convolutional Neural Network (CNN) module to extract character-level spatial feature representations of URLs; meanwhile, we employ an attention-based hierarchical Recurrent Neural Network(RNN) module to extract word-level temporal feature representations of URLs.", "category": "method" }, { "sentence": "We then fuse these feature representations via a three-layer CNN to build accurate feature representations of URLs, on which we train a phishing URL classifier.", "category": "method" }, { "sentence": "Extensive experiments on a verified dataset collected from the Internet demonstrate that the feature representations extracted automatically are conducive to the improvement of the generalization ability of our approach on newly emerging URLs, which makes our approach achieve competitive performance against other state-of-the-art approaches.", "category": "result" } ] }, { "paper_id": "215822457", "title": "DESEMPENHO DA TÉCNICA DEEP LEARNING NA ANÁLISE E CATEGORIZAÇÃO DE IMAGENS DE DEFEITO DE MADEIRA", "abstract": "A inteligência artificial tem tido grandes avanços em seu campo de pesquisa, contribuindo muito em várias áreas, como a análise e categorização de imagens digitais com o emprego de aprendizado de máquina. Existem várias técnicas específicas para o reconhecimento de imagens e sua categorização, uma dessas técnicas, que utiliza redes neurais artificiais, envolve o estudo específico, a extração das características por meio da análise de dados da imagem do objeto que está sendo analisado e a especificação de qual será o impacto dessas características no modelo neural para cada uma das categorias, o que exige a imersão do pesquisador em uma área ou campo de pesquisa que não é de seu domínio. A Deep Learning, utilizando-se das redes neurais convolucionais artificiais tem a capacidade de aprender e extrair as características durante o treinamento, sem a especificação dessas características no modelo, sendo que, geralmente, apresentam resultados melhores do que aqueles observados por modelos de redes neurais que tiveram as características observadas e programadas por humanos. O objetivo desse trabalho é aplicar a Deep Learning com auxílio da linguagem Python e duas bibliotecas chamadas Keras e NumPy, na categorização de um conjunto de imagens de tábuas de madeira, avaliando seu desempenho. Foram elaborados e avaliados alguns modelos de rede neural convolucional aplicados nesse processo, obtendo resultados promissores, sendo que o melhor deles apresentou erro de categorização na ordem de cinco por cento.", "classified_sentences": [ { "sentence": "A inteligência artificial tem tido grandes avanços em seu campo de pesquisa, contribuindo muito em várias áreas, como a análise e categorização de imagens digitais com o emprego de aprendizado de máquina.", "category": "background" }, { "sentence": "Existem várias técnicas específicas para o reconhecimento de imagens e sua categorização, uma dessas técnicas, que utiliza redes neurais artificiais, envolve o estudo específico, a extração das características por meio da análise de dados da imagem do objeto que está sendo analisado e a especificação de qual será o impacto dessas características no modelo neural para cada uma das categorias, o que exige a imersão do pesquisador em uma área ou campo de pesquisa que não é de seu domínio.", "category": "method" }, { "sentence": "A Deep Learning, utilizando-se das redes neurais convolucionais artificiais tem a capacidade de aprender e extrair as características durante o treinamento, sem a especificação dessas características no modelo, sendo que, geralmente, apresentam resultados melhores do que aqueles observados por modelos de redes neurais que tiveram as características observadas e programadas por humanos.", "category": "background" }, { "sentence": "O objetivo desse trabalho é aplicar a Deep Learning com auxílio da linguagem Python e duas bibliotecas chamadas Keras e NumPy, na categorização de um conjunto de imagens de tábuas de madeira, avaliando seu desempenho.", "category": "method" }, { "sentence": "Foram elaborados e avaliados alguns modelos de rede neural convolucional aplicados nesse processo, obtendo resultados promissores, sendo que o melhor deles apresentou erro de categorização na ordem de cinco por cento.", "category": "result" } ] }, { "paper_id": "218977368", "title": "Subtitles to Segmentation: Improving Low-Resource Speech-to-TextTranslation Pipelines", "abstract": "In this work, we focus on improving ASR output segmentation in the context of low-resource language speech-to-text translation. ASR output segmentation is crucial, as ASR systems segment the input audio using purely acoustic information and are not guaranteed to output sentence-like segments. Since most MT systems expect sentences as input, feeding in longer unsegmented passages can lead to sub-optimal performance. We explore the feasibility of using datasets of subtitles from TV shows and movies to train better ASR segmentation models. We further incorporate part-of-speech (POS) tag and dependency label information (derived from the unsegmented ASR outputs) into our segmentation model. We show that this noisy syntactic information can improve model accuracy. We evaluate our models intrinsically on segmentation quality and extrinsically on downstream MT performance, as well as downstream tasks including cross-lingual information retrieval (CLIR) tasks and human relevance assessments. Our model shows improved performance on downstream tasks for Lithuanian and Bulgarian.", "classified_sentences": [ { "sentence": "In this work, we focus on improving ASR output segmentation in the context of low-resource language speech-to-text translation.", "category": "background" }, { "sentence": "ASR output segmentation is crucial, as ASR systems segment the input audio using purely acoustic information and are not guaranteed to output sentence-like segments.", "category": "background" }, { "sentence": "Since most MT systems expect sentences as input, feeding in longer unsegmented passages can lead to sub-optimal performance.", "category": "background" }, { "sentence": "We explore the feasibility of using datasets of subtitles from TV shows and movies to train better ASR segmentation models.", "category": "method" }, { "sentence": "We further incorporate part-of-speech (POS) tag and dependency label information (derived from the unsegmented ASR outputs) into our segmentation model.", "category": "method" }, { "sentence": "We show that this noisy syntactic information can improve model accuracy.", "category": "result" }, { "sentence": "We evaluate our models intrinsically on segmentation quality and extrinsically on downstream MT performance, as well as downstream tasks including cross-lingual information retrieval (CLIR) tasks and human relevance assessments.", "category": "method" }, { "sentence": "Our model shows improved performance on downstream tasks for Lithuanian and Bulgarian.", "category": "result" } ] }, { "paper_id": "219055856", "title": "A Sensitivity Analysis with COSMO-LM at 1 km Resolution over South Italy", "abstract": "The results of a sensitivity analysis based on COSMO-LM (COnsortium for Small-Scale MOdeling—Lokal Model) simulations driven by ECMWF-IFS (European Centre for Medium-Range Weather Forecasts—Integrated Forecasting System). global data over a domain located in southern Italy are presented. Simulations have been performed at very high resolution (about 1 km). The main aim of this study is to individuate the most sensitive physical and numerical parameters of the model configuration, comparing a set of 18 simulations in terms of temperature and precipitation against ground observations. The parameters that result in having more influence for a proper representation of temperature and precipitation fields are the heat resistance length of laminar layer (which accounts for the high complexity of the interaction of the atmosphere with the underlying surface) and the minimal diffusion coefficient for heat. Temperature values are strongly influenced also by the vertical variation of critical relative humidity. An optimized tuning of these parameters allows COSMO-LM to improve the representation of simulated main features of this area, with significant bias reductions.", "classified_sentences": [ { "sentence": "The results of a sensitivity analysis based on COSMO-LM (COnsortium for Small-Scale MOdeling—Lokal Model) simulations driven by ECMWF-IFS (European Centre for Medium-Range Weather Forecasts—Integrated Forecasting System).", "category": "method" }, { "sentence": "global data over a domain located in southern Italy are presented.", "category": "background" }, { "sentence": "Simulations have been performed at very high resolution (about 1 km).", "category": "method" }, { "sentence": "The main aim of this study is to individuate the most sensitive physical and numerical parameters of the model configuration, comparing a set of 18 simulations in terms of temperature and precipitation against ground observations.", "category": "method" }, { "sentence": "The parameters that result in having more influence for a proper representation of temperature and precipitation fields are the heat resistance length of laminar layer (which accounts for the high complexity of the interaction of the atmosphere with the underlying surface) and the minimal diffusion coefficient for heat.", "category": "result" }, { "sentence": "Temperature values are strongly influenced also by the vertical variation of critical relative humidity.", "category": "result" }, { "sentence": "An optimized tuning of these parameters allows COSMO-LM to improve the representation of simulated main features of this area, with significant bias reductions.", "category": "result" } ] }, { "paper_id": "231853047", "title": "AWAF: AI Enabled Web Contents Authoring Framework", "abstract": "Artificial intelligence (AI) in web development is a new sector that a lot of people are into recently. AI continues to evolve and grow, and plays an increasingly important role in the web app development space [1]. When it comes to developing innovative and more sophisticated web applications, the involved technologies continue to play a bigger role. With the involvement of the internet into our daily lives, particularly businesses are enjoying the aspects of AI. Precisely, companies use AI in proper marketing of their products and enhancing their brand visibility by building their websites and web applications [2]. AI or Machine Learning (ML) models are able to help web app developers to solve problems related to security, user experience, content analysis, quality assurance and much more. This presents the need for a framework or tool that can allow third party developers to seamlessly build an AI based app. In this paper, we present an AI Enabled Web Contents Authoring Framework (AWAF), where AI models can be simply dragged into workspace, provide options to build, train Deep learning models using a simple web visual interface, and ultimately ship the AI features into the web application. Also, we provide an option to connect together smart blocks called AI Nodes, to create our custom deep learning models. These AI Nodes are designed flexible enough to reap the advantages of portability and reusability. And, laterally, we also focus on assigning or distributing the computational AI Nodes to capable IOT-edge devices like high-end TV etc. to leverage their hardware capabilities in order to increase the overall responsiveness of the AI application on low-end devices [3].", "classified_sentences": [ { "sentence": "Artificial intelligence (AI) in web development is a new sector that a lot of people are into recently.", "category": "background" }, { "sentence": "AI continues to evolve and grow, and plays an increasingly important role in the web app development space [1].", "category": "background" }, { "sentence": "When it comes to developing innovative and more sophisticated web applications, the involved technologies continue to play a bigger role.", "category": "background" }, { "sentence": "With the involvement of the internet into our daily lives, particularly businesses are enjoying the aspects of AI.", "category": "background" }, { "sentence": "Precisely, companies use AI in proper marketing of their products and enhancing their brand visibility by building their websites and web applications [2].", "category": "background" }, { "sentence": "AI or Machine Learning (ML) models are able to help web app developers to solve problems related to security, user experience, content analysis, quality assurance and much more.", "category": "background" }, { "sentence": "This presents the need for a framework or tool that can allow third party developers to seamlessly build an AI based app.", "category": "method" }, { "sentence": "In this paper, we present an AI Enabled Web Contents Authoring Framework (AWAF), where AI models can be simply dragged into workspace, provide options to build, train Deep learning models using a simple web visual interface, and ultimately ship the AI features into the web application.", "category": "method" }, { "sentence": "Also, we provide an option to connect together smart blocks called AI Nodes, to create our custom deep learning models.", "category": "method" }, { "sentence": "These AI Nodes are designed flexible enough to reap the advantages of portability and reusability.", "category": "method" }, { "sentence": "And, laterally, we also focus on assigning or distributing the computational AI Nodes to capable IOT-edge devices like high-end TV etc. to leverage their hardware capabilities in order to increase the overall responsiveness of the AI application on low-end devices [3].", "category": "method" } ] }, { "paper_id": "236034267", "title": "Deep Learning to Ternary Hash Codes by Continuation", "abstract": "Recently, it has been observed that $\\{0,\\pm1\\}$-ternary codes which are simply generated from deep features by hard thresholding, tend to outperform $\\{-1, 1\\}$-binary codes in image retrieval. To obtain better ternary codes, we for the first time propose to jointly learn the features with the codes by appending a smoothed function to the networks. During training, the function could evolve into a non-smoothed ternary function by a continuation method, and then generate ternary codes. The method circumvents the difficulty of directly training discrete functions and reduces the quantization errors of ternary codes. Experiments show that the proposed joint learning indeed could produce better ternary codes.", "classified_sentences": [ { "sentence": "Recently, it has been observed that $\\{0,\\pm1\\}$-ternary codes which are simply generated from deep features by hard thresholding, tend to outperform $\\{-1, 1\\}$-binary codes in image retrieval.", "category": "background" }, { "sentence": "To obtain better ternary codes, we for the first time propose to jointly learn the features with the codes by appending a smoothed function to the networks.", "category": "method" }, { "sentence": "During training, the function could evolve into a non-smoothed ternary function by a continuation method, and then generate ternary codes.", "category": "method" }, { "sentence": "The method circumvents the difficulty of directly training discrete functions and reduces the quantization errors of ternary codes.", "category": "method" }, { "sentence": "Experiments show that the proposed joint learning indeed could produce better ternary codes.", "category": "result" } ] }, { "paper_id": "237296976", "title": "Hybrid sequence‐based Android malware detection using natural language processing", "abstract": "Android platform has been the target of attackers due to its openness and increasing popularity. Android malware has explosively increased in recent years, which poses serious threats to Android security. Thus proposing efficient Android malware detection methods is curial in defeating malware. Various features extracted from static or dynamic analysis using machine learning have played an important role in malware detection recently. However, existing code obfuscation, code encryption, and dynamic code loading techniques can be employed to hinder systems that single based on static analysis, purely dynamic analysis systems cannot detect all potential code execution paths. To address these issues, we propose CoDroid, a sequence‐based hybrid Android malware detection method, which utilizes the sequences of static opcode and dynamic system call. We treat one sequence as a sentence in the natural language processing and construct a CNN–BiLSTM–Attention classifier which consists of Convolutional Neural Networks (CNNs), the Bidirectional Long Short‐Term Memory (BiLSTM) with an attention language model. We extensively evaluate CoDroid under a real‐world data set and perform comprehensive analysis against other existing related detection methods. The evaluations show the effectiveness and flexibility of CoDroid across a variety of experimental settings.", "classified_sentences": [ { "sentence": "Android platform has been the target of attackers due to its openness and increasing popularity.", "category": "background" }, { "sentence": "Android malware has explosively increased in recent years, which poses serious threats to Android security.", "category": "background" }, { "sentence": "Thus proposing efficient Android malware detection methods is curial in defeating malware.", "category": "background" }, { "sentence": "Various features extracted from static or dynamic analysis using machine learning have played an important role in malware detection recently.", "category": "background" }, { "sentence": "However, existing code obfuscation, code encryption, and dynamic code loading techniques can be employed to hinder systems that single based on static analysis, purely dynamic analysis systems cannot detect all potential code execution paths.", "category": "background" }, { "sentence": "To address these issues, we propose CoDroid, a sequence‐based hybrid Android malware detection method, which utilizes the sequences of static opcode and dynamic system call.", "category": "method" }, { "sentence": "We treat one sequence as a sentence in the natural language processing and construct a CNN–BiLSTM–Attention classifier which consists of Convolutional Neural Networks (CNNs), the Bidirectional Long Short‐Term Memory (BiLSTM) with an attention language model.", "category": "method" }, { "sentence": "We extensively evaluate CoDroid under a real‐world data set and perform comprehensive analysis against other existing related detection methods.", "category": "method" }, { "sentence": "The evaluations show the effectiveness and flexibility of CoDroid across a variety of experimental settings.", "category": "result" } ] }, { "paper_id": "244129501", "title": "Content Based Image Retrieval using Convolutional neural network and SVM", "abstract": "One of the most interesting and rapidly growing search fields in the last decade has been content-based image retrieval. Image retrieval is the process of obtaining and describing images from a huge database. Images are extracted and then matched to obtain the image. Many feature descriptors have been developed, however their effectiveness is influenced by a variety of factors. This study proposes a new technique for CBIR based on deep learning-based feature extraction and support vector machine-based base similarity matching. In the proposed method, a pre-train convolutional neural network (CNN) is utilised to extract robust features from images, and SVM is used for similarity matching. Publically available Corel dataset is used for the evaluation of performance of the proposed method and experimental analysis of the proposed approach is simulated on MATLAB2017a toolbox which comprises the various functions to simulate it. Propose method gives 97.27 % accuracy, which is significance improvement with compare to existing methods.", "classified_sentences": [ { "sentence": "One of the most interesting and rapidly growing search fields in the last decade has been content-based image retrieval.", "category": "background" }, { "sentence": "Image retrieval is the process of obtaining and describing images from a huge database.", "category": "background" }, { "sentence": "Images are extracted and then matched to obtain the image.", "category": "background" }, { "sentence": "Many feature descriptors have been developed, however their effectiveness is influenced by a variety of factors.", "category": "background" }, { "sentence": "This study proposes a new technique for CBIR based on deep learning-based feature extraction and support vector machine-based base similarity matching.", "category": "method" }, { "sentence": "In the proposed method, a pre-train convolutional neural network (CNN) is utilised to extract robust features from images, and SVM is used for similarity matching.", "category": "method" }, { "sentence": "Publically available Corel dataset is used for the evaluation of performance of the proposed method and experimental analysis of the proposed approach is simulated on MATLAB2017a toolbox which comprises the various functions to simulate it.", "category": "method" }, { "sentence": "Propose method gives 97.27 % accuracy, which is significance improvement with compare to existing methods.", "category": "result" } ] }, { "paper_id": "244199419", "title": "Testing the reliability of forecasting systems", "abstract": "The problem of statistically evaluating forecasting systems is revisited. The forecaster claims the forecasts to exhibit a certain nominal statistical behaviour; for instance, the forecasts provide the expected value (or certain quantiles) of the verification, conditional on the information available at forecast time. Forecasting systems that indeed exhibit the nominal behaviour are referred to as reliable. Statistical tests for reliability are presented (based on an archive of verification–forecast pairs). As noted previously, devising such tests is encumbered by the fact that the dependence structure of the verification–forecast pairs is not known in general. Ignoring this dependence though might lead to incorrect tests and too-frequent rejection of forecasting systems that are actually reliable. On the other hand, reliability typically implies that the forecast provides information about the dependence structure, and using this in conjunction with judicious choices of the test statistic, rigorous results on the asymptotic distribution of the test statistic are obtained. These results are used to test for reliability under minimal additional assumptions on the statistical properties of the verification–forecast pairs. Applications to environmental forecasts are discussed. A python implementation of the discussed methods is available online.", "classified_sentences": [ { "sentence": "The problem of statistically evaluating forecasting systems is revisited.", "category": "background" }, { "sentence": "The forecaster claims the forecasts to exhibit a certain nominal statistical behaviour; for instance, the forecasts provide the expected value (or certain quantiles) of the verification, conditional on the information available at forecast time.", "category": "background" }, { "sentence": "Forecasting systems that indeed exhibit the nominal behaviour are referred to as reliable.", "category": "background" }, { "sentence": "Statistical tests for reliability are presented (based on an archive of verification–forecast pairs).", "category": "method" }, { "sentence": "As noted previously, devising such tests is encumbered by the fact that the dependence structure of the verification–forecast pairs is not known in general.", "category": "background" }, { "sentence": "Ignoring this dependence though might lead to incorrect tests and too-frequent rejection of forecasting systems that are actually reliable.", "category": "background" }, { "sentence": "On the other hand, reliability typically implies that the forecast provides information about the dependence structure, and using this in conjunction with judicious choices of the test statistic, rigorous results on the asymptotic distribution of the test statistic are obtained.", "category": "method" }, { "sentence": "These results are used to test for reliability under minimal additional assumptions on the statistical properties of the verification–forecast pairs.", "category": "method" }, { "sentence": "Applications to environmental forecasts are discussed.", "category": "result" }, { "sentence": "A python implementation of the discussed methods is available online.", "category": "result" } ] }, { "paper_id": "245352600", "title": "Uncertainty with a potential Winter Storm : The East Coast Storm of 14 January 2008", "abstract": "The National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) and the NCEP Global Ensemble Forecast System showed the potential for a major East Winter Storm (ECWS) between 0000 UTC 14-15 January 2008. Initial forecasts showed some run-to-run inconsistencies with the cyclone track. This “model jumpiness” was a clear indication of uncertainty associated with this potential storm.", "classified_sentences": [ { "sentence": "The National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) and the NCEP Global Ensemble Forecast System showed the potential for a major East Winter Storm (ECWS) between 0000 UTC 14-15 January 2008.", "category": "background" }, { "sentence": "Initial forecasts showed some run-to-run inconsistencies with the cyclone track.", "category": "result" }, { "sentence": "This “model jumpiness” was a clear indication of uncertainty associated with this potential storm.", "category": "result" } ] }, { "paper_id": "248941670", "title": "Characterizing Target-absent Human Attention", "abstract": "Human efficiency in finding a target in an image has attracted the attention of machine learning researchers, but what about when no target is there? Knowing how people search in the absence of a target, and when they stop, is important for Human-computer-interaction systems attempting to predict human gaze behavior in the wild. Here we report a rigorous evaluation of target-absent search behavior using the COCO-Search18 dataset to train state-of-the-art models. We focus on two specific aims. First, we characterize the presence of a target guidance signal in target-absent search behavior by comparing it to target-present guidance and free viewing. We do this by comparing how well a model trained on one type of fixation behavior (target-present, target-absent, free viewing) can predict behavior in either the same or different task. To compare target-absent search to free viewing behavior we created COCO-FreeView, a dataset of free-viewing fixations for the same images used in COCO-Search18. These comparisons revealed the existence of a target guidance signal in target-absent search, albeit one much less dominant compared to when a target actually appeared in an image, and that the target-absent guidance signal was similar to free viewing in that saliency and center bias were both weighted more than guidance from target features. Our second aim focused on the stopping criteria, a question intrinsic to target-absent search. Here we propose to train a foveated target detector whose target detection representation is sensitive to the relationship between distance from the fovea. Then combining the predicted target detection representation with other information such as fixation history and subject ID, our model outperforms the baselines in predicting when a person stops moving his attention during target-absent search.", "classified_sentences": [ { "sentence": "Human efficiency in finding a target in an image has attracted the attention of machine learning researchers, but what about when no target is there?", "category": "background" }, { "sentence": "Knowing how people search in the absence of a target, and when they stop, is important for Human-computer-interaction systems attempting to predict human gaze behavior in the wild.", "category": "background" }, { "sentence": "Here we report a rigorous evaluation of target-absent search behavior using the COCO-Search18 dataset to train state-of-the-art models.", "category": "method" }, { "sentence": "We focus on two specific aims.", "category": "method" }, { "sentence": "First, we characterize the presence of a target guidance signal in target-absent search behavior by comparing it to target-present guidance and free viewing.", "category": "method" }, { "sentence": "We do this by comparing how well a model trained on one type of fixation behavior (target-present, target-absent, free viewing) can predict behavior in either the same or different task.", "category": "method" }, { "sentence": "To compare target-absent search to free viewing behavior we created COCO-FreeView, a dataset of free-viewing fixations for the same images used in COCO-Search18.", "category": "method" }, { "sentence": "These comparisons revealed the existence of a target guidance signal in target-absent search, albeit one much less dominant compared to when a target actually appeared in an image, and that the target-absent guidance signal was similar to free viewing in that saliency and center bias were both weighted more than guidance from target features.", "category": "result" }, { "sentence": "Our second aim focused on the stopping criteria, a question intrinsic to target-absent search.", "category": "method" }, { "sentence": "Here we propose to train a foveated target detector whose target detection representation is sensitive to the relationship between distance from the fovea.", "category": "method" }, { "sentence": "Then combining the predicted target detection representation with other information such as fixation history and subject ID, our model outperforms the baselines in predicting when a person stops moving his attention during target-absent search.", "category": "result" } ] }, { "paper_id": "251943573", "title": "An investigation of ML techniques to detect Phishing Websites by complexity reduction", "abstract": "In today's digital age, one of the predominant causes of the security breaches is phishing web sites that disguise them-selves as legitimate web sites and trick unsuspecting users into revealing sensitive information. With the proliferation of high-speed internet and the popularization of IT education, there is an increase in unscrupulous actors on the web who are always ready to counterfeit a legitimate website and use it to deceive and ma-nipulate users. Software and non-software-based techniques have been used to try to unmask the phishers. Phishing web sites have many characteristics in them. Thus, classifying and detecting those is unavoidably time-consuming and complex. Our research analyzed several hybrid machine learning models, including a bespoke preprocessing step of reducing minimally correlated features and then training with four boosting algorithms and three SVM models for classification. These models have also been trained after hyperparameter tuning. Among the investigated models, XGBoost brought the highest accuracy of 97.0455% after the hyperparameter tuning.", "classified_sentences": [ { "sentence": "In today's digital age, one of the predominant causes of the security breaches is phishing web sites that disguise them-selves as legitimate web sites and trick unsuspecting users into revealing sensitive information.", "category": "background" }, { "sentence": "With the proliferation of high-speed internet and the popularization of IT education, there is an increase in unscrupulous actors on the web who are always ready to counterfeit a legitimate website and use it to deceive and ma-nipulate users.", "category": "background" }, { "sentence": "Software and non-software-based techniques have been used to try to unmask the phishers.", "category": "background" }, { "sentence": "Phishing web sites have many characteristics in them.", "category": "background" }, { "sentence": "Thus, classifying and detecting those is unavoidably time-consuming and complex.", "category": "background" }, { "sentence": "Our research analyzed several hybrid machine learning models, including a bespoke preprocessing step of reducing minimally correlated features and then training with four boosting algorithms and three SVM models for classification.", "category": "method" }, { "sentence": "These models have also been trained after hyperparameter tuning.", "category": "method" }, { "sentence": "Among the investigated models, XGBoost brought the highest accuracy of 97.0455% after the hyperparameter tuning.", "category": "result" } ] }, { "paper_id": "121449822", "title": "A global numerical weather prediction model with variable resolution: Application to the shallow‐water equations", "abstract": "We follow the approach suggested by F. Schmidt to implement a spectral global shallow-water model with variable resolution. A conformal mapping is built between the earth and a computational sphere and the equations are discretized on the latter using the standard spectral technique associated with a collocation (Gaussian) grid. We prove that the only non-trivial conformal mapping which exists between the two spheres is based on the transformation introduced by Schmidt, but the pole of the collocation grid has no longer to coincide with the pole of dilatation. We implement the technique in an explicit model, where only minor modifications to a uniform resolution model are needed. The semi-implicit scheme and the nonlinear normal mode initialization are proved to work satisfactorily. 24-hour forecasts show that the method is successful in dealing with the shallow-water equations and allow us to discuss the potential of the approach.", "classified_sentences": [ { "sentence": "We follow the approach suggested by F.", "category": "method" }, { "sentence": "Schmidt to implement a spectral global shallow-water model with variable resolution.", "category": "method" }, { "sentence": "A conformal mapping is built between the earth and a computational sphere and the equations are discretized on the latter using the standard spectral technique associated with a collocation (Gaussian) grid.", "category": "method" }, { "sentence": "We prove that the only non-trivial conformal mapping which exists between the two spheres is based on the transformation introduced by Schmidt, but the pole of the collocation grid has no longer to coincide with the pole of dilatation.", "category": "method" }, { "sentence": "We implement the technique in an explicit model, where only minor modifications to a uniform resolution model are needed.", "category": "method" }, { "sentence": "The semi-implicit scheme and the nonlinear normal mode initialization are proved to work satisfactorily.", "category": "result" }, { "sentence": "24-hour forecasts show that the method is successful in dealing with the shallow-water equations and allow us to discuss the potential of the approach.", "category": "result" } ] }, { "paper_id": "123224620", "title": "Ability of a Poor Man's Ensemble to Predict the Probability and Distribution of Precipitation", "abstract": "Abstract A poor man's ensemble is a set of independent numerical weather prediction (NWP) model forecasts from several operational centers. Because it samples uncertainties in both the initial conditions and model formulation through the variation of input data, analysis, and forecast methodologies of its component members, it is less prone to systematic biases and errors that cause underdispersive behavior in single-model ensemble prediction systems (EPSs). It is also essentially cost-free. Its main disadvantage is its relatively small size. This paper investigates the ability of a poor man's ensemble to provide forecasts of the probability and distribution of rainfall in the short range, 1–2 days. The poor man's ensemble described here consists of 24- and 48-h daily quantitative precipitation forecasts (QPFs) from seven operational NWP models. The ensemble forecasts were verified for a 28-month period over Australia using gridded daily rain gauge analyses. Forecasts of the probability of precipitation (.", "classified_sentences": [ { "sentence": "Abstract A poor man's ensemble is a set of independent numerical weather prediction (NWP) model forecasts from several operational centers.", "category": "background" }, { "sentence": "Because it samples uncertainties in both the initial conditions and model formulation through the variation of input data, analysis, and forecast methodologies of its component members, it is less prone to systematic biases and errors that cause underdispersive behavior in single-model ensemble prediction systems (EPSs).", "category": "background" }, { "sentence": "It is also essentially cost-free.", "category": "background" }, { "sentence": "Its main disadvantage is its relatively small size.", "category": "background" }, { "sentence": "This paper investigates the ability of a poor man's ensemble to provide forecasts of the probability and distribution of rainfall in the short range, 1–2 days.", "category": "method" }, { "sentence": "The poor man's ensemble described here consists of 24- and 48-h daily quantitative precipitation forecasts (QPFs) from seven operational NWP models.", "category": "method" }, { "sentence": "The ensemble forecasts were verified for a 28-month period over Australia using gridded daily rain gauge analyses.", "category": "method" }, { "sentence": "Forecasts of the probability of precipitation (.", "category": "result" } ] }, { "paper_id": "123272833", "title": "Modeling the Distribution of Precipitation Forecasts from the Canadian Ensemble Prediction System Using Kernel Density Estimation", "abstract": "Kernel density estimation is employed to fit smooth probabilistic models to precipitation forecasts of the Canadian ensemble prediction system. An intuitive nonparametric technique, kernel density estimation has become a powerful tool widely used in the approximation of probability density functions. The density estimators were constructed using the gamma kernels prescribed by S. -X. Chen, confined as they are to the nonnegative real axis, which constitutes the support of the random variable representing precipitation accumulation. Performance of kernel density estimators for several different smoothing bandwidths is compared with the discrete probabilistic model obtained as the fraction of member forecasts predicting the events, which for this study consisted of threshold exceedances. A propitious choice of the smoothing bandwidth yields smooth forecasts comparable, or sometimes superior, to the discrete probabilistic forecast, depending on the character of the raw ensemble forecasts. At the same time more realistic models of the probability density are achieved, particularly in the tail of the distribution, yielding forecasts that can be optimally calibrated for extreme events.", "classified_sentences": [ { "sentence": "Kernel density estimation is employed to fit smooth probabilistic models to precipitation forecasts of the Canadian ensemble prediction system.", "category": "method" }, { "sentence": "An intuitive nonparametric technique, kernel density estimation has become a powerful tool widely used in the approximation of probability density functions.", "category": "background" }, { "sentence": "The density estimators were constructed using the gamma kernels prescribed by S.-X. Chen, confined as they are to the nonnegative real axis, which constitutes the support of the random variable representing precipitation accumulation.", "category": "method" }, { "sentence": "Performance of kernel density estimators for several different smoothing bandwidths is compared with the discrete probabilistic model obtained as the fraction of member forecasts predicting the events, which for this study consisted of threshold exceedances.", "category": "method" }, { "sentence": "A propitious choice of the smoothing bandwidth yields smooth forecasts comparable, or sometimes superior, to the discrete probabilistic forecast, depending on the character of the raw ensemble forecasts.", "category": "result" }, { "sentence": "At the same time more realistic models of the probability density are achieved, particularly in the tail of the distribution, yielding forecasts that can be optimally calibrated for extreme events.", "category": "result" } ] }, { "paper_id": "123743583", "title": "Impact of Ensemble Size on Ensemble Prediction", "abstract": "Abstract The impact of ensemble size on the performance of the European Centre for Medium-Range Weather Forecasts ensemble prediction system (EPS) is analyzed. The skill of ensembles generated using 2, 4, 8, 16, and 32 perturbed ensemble members are compared for a period of 45 days—from 1 October to 15 November 1996. For each ensemble configuration, the skill is compared with the potential skill, measured by randomly choosing one of the 32 ensemble members as verification (idealized ensemble). Results are based on the analyses of the prediction of the 500-hPa geopotential height field. Various measures of performance are applied: skill of the ensemble mean, spread–skill relationship, skill of most accurate ensemble member, Brier score, ranked probability score, relative operating characteristic, and the outlier statistic. The relation between ensemble spread and control error is studied using L2, L8, and L∞ norms to measure distances between ensemble members and the control forecast or the verification. I.", "classified_sentences": [ { "sentence": "Abstract The impact of ensemble size on the performance of the European Centre for Medium-Range Weather Forecasts ensemble prediction system (EPS) is analyzed.", "category": "background" }, { "sentence": "The skill of ensembles generated using 2, 4, 8, 16, and 32 perturbed ensemble members are compared for a period of 45 days—from 1 October to 15 November 1996.", "category": "method" }, { "sentence": "For each ensemble configuration, the skill is compared with the potential skill, measured by randomly choosing one of the 32 ensemble members as verification (idealized ensemble).", "category": "method" }, { "sentence": "Results are based on the analyses of the prediction of the 500-hPa geopotential height field.", "category": "result" }, { "sentence": "Various measures of performance are applied: skill of the ensemble mean, spread–skill relationship, skill of most accurate ensemble member, Brier score, ranked probability score, relative operating characteristic, and the outlier statistic.", "category": "method" }, { "sentence": "The relation between ensemble spread and control error is studied using L2, L8, and L∞ norms to measure distances between ensemble members and the control forecast or the verification.", "category": "method" } ] }, { "paper_id": "123879230", "title": "Wind Forecast Error and Trajectory Prediction for En-Route Scheduling", "abstract": "This work examines wind forecast forecast error as it pertains to aircraft trajectory prediction for scheduling and spacing optimized profile descents. We first perform a statistcal analysis of the RUC-2 wind forecast by comparing the forecast with one year of archived ACARS wind speed observations. This analysis provided insights into the RUC-2 forecast’s tendency to over or under predict wind speed contributions to aircraft trajectory predictions depending on the direction of flight. These results were supported by an additional simuilation study using historical ACARS paths and airspeed profiles wand RUC-2 wind forecasts to predict flight times along multiple routings from west-coast origin airports to KATL. These predictions were then compared to the actual flights times of the ACARS path and used to form flight time error distributions as a function of aircraft routing and prediction horizon. We found that flights originating from more northernly located airports were less susceptiable to error in the RUC-2 forecasts, becuase a significant difference existed bewtween their ground track angle and the prevailing wind direction from the wind forecast. The ultimate contribution of this work is the recognition and evidence that the spatial distribution of wind forecast errors plays a significant role in trajectory prediction error. More importantly, these values must not be seen as static values, rather as continuous parameters that must be reevaluated as a predicted trajectory enters different regions of the national airspace.", "classified_sentences": [ { "sentence": "This work examines wind forecast forecast error as it pertains to aircraft trajectory prediction for scheduling and spacing optimized profile descents.", "category": "background" }, { "sentence": "We first perform a statistcal analysis of the RUC-2 wind forecast by comparing the forecast with one year of archived ACARS wind speed observations.", "category": "method" }, { "sentence": "This analysis provided insights into the RUC-2 forecast’s tendency to over or under predict wind speed contributions to aircraft trajectory predictions depending on the direction of flight.", "category": "result" }, { "sentence": "These results were supported by an additional simuilation study using historical ACARS paths and airspeed profiles wand RUC-2 wind forecasts to predict flight times along multiple routings from west-coast origin airports to KATL.", "category": "method" }, { "sentence": "These predictions were then compared to the actual flights times of the ACARS path and used to form flight time error distributions as a function of aircraft routing and prediction horizon.", "category": "result" }, { "sentence": "We found that flights originating from more northernly located airports were less susceptiable to error in the RUC-2 forecasts, becuase a significant difference existed bewtween their ground track angle and the prevailing wind direction from the wind forecast.", "category": "result" }, { "sentence": "The ultimate contribution of this work is the recognition and evidence that the spatial distribution of wind forecast errors plays a significant role in trajectory prediction error.", "category": "result" }, { "sentence": "More importantly, these values must not be seen as static values, rather as continuous parameters that must be reevaluated as a predicted trajectory enters different regions of the national airspace.", "category": "result" } ] }, { "paper_id": "124640597", "title": "An Improved Operational System for Forecasting Precipitation Type", "abstract": "Abstract A Model Output Statistics system for forecasting the conditional probability of precipitation type (PoPT) became operational within the National Weather Service in September 1978. Forecasts are provided for three precipitation type categories: snow or ice pellets, freezing rain, and rain. To develop the forecast equations, data are combined from different stations because of the limited amount of developmental data. To justify combining the data, the Limited-area Fine Mesh (LFM) model predictors are transformed from their original values through the use of the logit model. In one experiment, it is shown that probability of snow forecasts are made more accurate through an improved use of the logit model for predictor transformation. The new transformation procedure is then used in the development of a set of experimental PoPT forecast equations. The experimental equations differ from the operational equations in other ways also. The developmental sample for the experimental equations included appr.", "classified_sentences": [ { "sentence": "Abstract A Model Output Statistics system for forecasting the conditional probability of precipitation type (PoPT) became operational within the National Weather Service in September 1978.", "category": "background" }, { "sentence": "Forecasts are provided for three precipitation type categories: snow or ice pellets, freezing rain, and rain.", "category": "background" }, { "sentence": "To develop the forecast equations, data are combined from different stations because of the limited amount of developmental data.", "category": "method" }, { "sentence": "To justify combining the data, the Limited-area Fine Mesh (LFM) model predictors are transformed from their original values through the use of the logit model.", "category": "method" }, { "sentence": "In one experiment, it is shown that probability of snow forecasts are made more accurate through an improved use of the logit model for predictor transformation.", "category": "result" }, { "sentence": "The new transformation procedure is then used in the development of a set of experimental PoPT forecast equations.", "category": "method" }, { "sentence": "The experimental equations differ from the operational equations in other ways also.", "category": "result" }, { "sentence": "The developmental sample for the experimental equations included appr.", "category": "method" } ] }, { "paper_id": "258912148", "title": "A Machine Learning approach for Detecting Malicious URL using different algorithms and NLP techniques", "abstract": "A malicious URL is one that was made specifically to attack through spam or fraud. Due to the billions of dollars that are compromised, malicious URLs pose a severe threat to security software. Finding secure and phishing links is therefore crucial. Therefore, machine learning is quite helpful for resolving security-related challenges. In this study, we use about 5 lakh URLs that were retrieved from the Kaggle dataset. We are utilizing three NLP approaches, including the count vectorizer, hash vectorizer, and TF-IDF vectorizer. Six machine learning classifiers, including the decision tree, random forest, K-NN, NB, SVM, and logistic regression, were used in conjunction with all these techniques. The highest accuracy results of 98.2 percent are produced by random forest. To determine whether or not the URL supplied is malicious, we built a web app using Flask.", "classified_sentences": [ { "sentence": "A malicious URL is one that was made specifically to attack through spam or fraud.", "category": "background" }, { "sentence": "Due to the billions of dollars that are compromised, malicious URLs pose a severe threat to security software.", "category": "background" }, { "sentence": "Finding secure and phishing links is therefore crucial.", "category": "background" }, { "sentence": "Therefore, machine learning is quite helpful for resolving security-related challenges.", "category": "background" }, { "sentence": "In this study, we use about 5 lakh URLs that were retrieved from the Kaggle dataset.", "category": "method" }, { "sentence": "We are utilizing three NLP approaches, including the count vectorizer, hash vectorizer, and TF-IDF vectorizer.", "category": "method" }, { "sentence": "Six machine learning classifiers, including the decision tree, random forest, K-NN, NB, SVM, and logistic regression, were used in conjunction with all these techniques.", "category": "method" }, { "sentence": "The highest accuracy results of 98.2 percent are produced by random forest.", "category": "result" }, { "sentence": "To determine whether or not the URL supplied is malicious, we built a web app using Flask.", "category": "method" } ] }, { "paper_id": "259892387", "title": "CATNet: A Cascaded and Aggregated Transformer Network for RGB-D Salient Object Detection", "abstract": "Salient object detection (SOD) is an important preprocessing operation for various computer vision tasks. Most of existing RGB-D SOD models employ additive or connected strategies to directly aggregate and decode multi-scale features to predict salient maps. However, due to the large differences between the features of different scales, these aggregation strategies adopted may lead to information loss or redundancy, and few methods explicitly consider how to establish connections between features at different scales in the decoding process, which consequently deteriorates the detection performance of the models. To this end, we propose a cascaded and aggregated Transformer Network (CATNet) which consists of three key modules, i.e., attention feature enhancement module (AFEM), cross-modal fusion module (CMFM) and cascaded correction decoder (CCD). Specifically, the AFEM is designed on the basis of atrous spatial pyramid pooling to obtain multi-scale semantic information and global context information in high-level features through dilated convolution and multi-head self-attention mechanism, enhancing high-level features. The role of the CMFM is to enhance and thereafter fuse the RGB features and depth features, alleviating the problem of poor-quality depth maps. The CCD is composed of two subdecoders in a cascading fashion. It is designed to suppress noise in low-level features and mitigate the differences between features at different scales. Moreover, the CCD uses a feedback mechanism to correct and repair the output of the subdecoder by exploiting supervised features, so that the problem of information loss caused by the upsampling operation during the multi-scale features aggregation process can be mitigated. Extensive experimental results demonstrate that the proposed CATNet achieves superior performance over 14 state-of-the-art RGB-D methods on 7 challenging benchmarks.", "classified_sentences": [ { "sentence": "Salient object detection (SOD) is an important preprocessing operation for various computer vision tasks.", "category": "background" }, { "sentence": "Most of existing RGB-D SOD models employ additive or connected strategies to directly aggregate and decode multi-scale features to predict salient maps.", "category": "background" }, { "sentence": "However, due to the large differences between the features of different scales, these aggregation strategies adopted may lead to information loss or redundancy, and few methods explicitly consider how to establish connections between features at different scales in the decoding process, which consequently deteriorates the detection performance of the models.", "category": "background" }, { "sentence": "To this end, we propose a cascaded and aggregated Transformer Network (CATNet) which consists of three key modules, i.e., attention feature enhancement module (AFEM), cross-modal fusion module (CMFM) and cascaded correction decoder (CCD).", "category": "method" }, { "sentence": "Specifically, the AFEM is designed on the basis of atrous spatial pyramid pooling to obtain multi-scale semantic information and global context information in high-level features through dilated convolution and multi-head self-attention mechanism, enhancing high-level features.", "category": "method" }, { "sentence": "The role of the CMFM is to enhance and thereafter fuse the RGB features and depth features, alleviating the problem of poor-quality depth maps.", "category": "method" }, { "sentence": "The CCD is composed of two subdecoders in a cascading fashion.", "category": "method" }, { "sentence": "It is designed to suppress noise in low-level features and mitigate the differences between features at different scales.", "category": "method" }, { "sentence": "Moreover, the CCD uses a feedback mechanism to correct and repair the output of the subdecoder by exploiting supervised features, so that the problem of information loss caused by the upsampling operation during the multi-scale features aggregation process can be mitigated.", "category": "method" }, { "sentence": "Extensive experimental results demonstrate that the proposed CATNet achieves superior performance over 14 state-of-the-art RGB-D methods on 7 challenging benchmarks.", "category": "result" } ] }, { "paper_id": "128667935", "title": "Visual saliency detection based on modeling the spatial Gaussianity", "abstract": "In this paper, a novel salient object detection method based on modeling the spatial anomalies is presented. The proposed framework is inspired by the biological mechanism that human eyes are sensitive to the unusual and anomalous objects among complex background. It is supposed that a natural image can be seen as a combination of some similar or dissimilar basic patches, and there is a direct relationship between its saliency and anomaly. Some patches share high degree of similarity and have a vast number of quantity. They usually make up the background of an image. On the other hand, some patches present strong rarity and specificity. We name these patches “anomalies”. Generally, anomalous patch is a reflection of the edge or some special colors and textures in an image, and these pattern cannot be well “explained” by their surroundings. Human eyes show great interests in these anomalous patterns, and will automatically pick out the anomalous parts of an image as the salient regions. To better evaluate the anomaly degree of the basic patches and exploit their nonlinear statistical characteristics, a multivariate Gaussian distribution saliency evaluation model is proposed. In this way, objects with anomalous patterns usually appear as the outliers in the Gaussian distribution, and we identify these anomalous objects as salient ones. Experiments are conducted on the well-known MSRA saliency detection dataset. Compared with other recent developed visual saliency detection methods, our method suggests significant advantages.", "classified_sentences": [ { "sentence": "In this paper, a novel salient object detection method based on modeling the spatial anomalies is presented.", "category": "method" }, { "sentence": "The proposed framework is inspired by the biological mechanism that human eyes are sensitive to the unusual and anomalous objects among complex background.", "category": "method" }, { "sentence": "It is supposed that a natural image can be seen as a combination of some similar or dissimilar basic patches, and there is a direct relationship between its saliency and anomaly.", "category": "method" }, { "sentence": "Some patches share high degree of similarity and have a vast number of quantity.", "category": "background" }, { "sentence": "They usually make up the background of an image.", "category": "background" }, { "sentence": "On the other hand, some patches present strong rarity and specificity.", "category": "background" }, { "sentence": "We name these patches “anomalies”.", "category": "background" }, { "sentence": "Generally, anomalous patch is a reflection of the edge or some special colors and textures in an image, and these pattern cannot be well “explained” by their surroundings.", "category": "background" }, { "sentence": "Human eyes show great interests in these anomalous patterns, and will automatically pick out the anomalous parts of an image as the salient regions.", "category": "background" }, { "sentence": "To better evaluate the anomaly degree of the basic patches and exploit their nonlinear statistical characteristics, a multivariate Gaussian distribution saliency evaluation model is proposed.", "category": "method" }, { "sentence": "In this way, objects with anomalous patterns usually appear as the outliers in the Gaussian distribution, and we identify these anomalous objects as salient ones.", "category": "method" }, { "sentence": "Experiments are conducted on the well-known MSRA saliency detection dataset.", "category": "result" }, { "sentence": "Compared with other recent developed visual saliency detection methods, our method suggests significant advantages.", "category": "result" } ] }, { "paper_id": "263681952", "title": "Evaluación de diferentes estrategias para la generación de sistemas de predicción por conjuntos regionales de escala convectiva en un caso de precipitación intensa", "abstract": "El pronóstico por conjuntos constituye una metodología consolidada para incorporar la incertidumbre asociada a los pronósticos en diversas escalas espaciales y temporales. En particular, en la mesoescala, no es claro aún cuáles son las técnicas más efectivas para representar la incertidumbre asociada a las condiciones iniciales y a los errores de modelo. En este trabajo se evalúan tres alternativas diferentes para la generación de pronósticos por conjuntos en alta resolución, y se realiza una comparación con un sistema de predicción por conjuntos global de baja resolución. Cada conjunto se construyó con 20 miembros utilizando el modelo WRF-ARW y 4 km de resolución horizontal sobre un dominio que abarca el centro noreste de Argentina. Se explora el desempeño de los conjuntos para un caso de estudio de precipitación intensa entre el 22 y 24 de diciembre de 2015. Los resultados se centran en el análisis del desempeño del pronóstico de precipitación y muestran que los conjuntos en alta resolución tienen mejor desempeño que el sistema global de menor resolución tanto en términos de la precisión del pronóstico como en términos de la cuantificación de su incertidumbre. En este trabajo, los conjuntos donde solo se perturban las condiciones iniciales y de borde tienden a mostrar una menor dispersión que aquellos en donde se combinan diferentes parametrizaciones de los procesos de escala sub-reticular para la representación de los errores de modelo. Estos ´últimos presentan además un menor sesgo para umbrales mayores a 10 mm. Asimismo, aumentar la resolución de las condiciones iniciales y de borde de la media del ensamble aumenta levemente la dispersión y mejora la representación espacial de los patrones de precipitación para todos los umbrales considerados.", "classified_sentences": [ { "sentence": "El pronóstico por conjuntos constituye una metodología consolidada para incorporar la incertidumbre asociada a los pronósticos en diversas escalas espaciales y temporales.", "category": "background" }, { "sentence": "En particular, en la mesoescala, no es claro aún cuáles son las técnicas más efectivas para representar la incertidumbre asociada a las condiciones iniciales y a los errores de modelo.", "category": "background" }, { "sentence": "En este trabajo se evalúan tres alternativas diferentes para la generación de pronósticos por conjuntos en alta resolución, y se realiza una comparación con un sistema de predicción por conjuntos global de baja resolución.", "category": "method" }, { "sentence": "Cada conjunto se construyó con 20 miembros utilizando el modelo WRF-ARW y 4 km de resolución horizontal sobre un dominio que abarca el centro noreste de Argentina.", "category": "method" }, { "sentence": "Se explora el desempeño de los conjuntos para un caso de estudio de precipitación intensa entre el 22 y 24 de diciembre de 2015.", "category": "method" }, { "sentence": "Los resultados se centran en el análisis del desempeño del pronóstico de precipitación y muestran que los conjuntos en alta resolución tienen mejor desempeño que el sistema global de menor resolución tanto en términos de la precisión del pronóstico como en términos de la cuantificación de su incertidumbre.", "category": "result" }, { "sentence": "En este trabajo, los conjuntos donde solo se perturban las condiciones iniciales y de borde tienden a mostrar una menor dispersión que aquellos en donde se combinan diferentes parametrizaciones de los procesos de escala sub-reticular para la representación de los errores de modelo.", "category": "result" }, { "sentence": "Estos ´últimos presentan además un menor sesgo para umbrales mayores a 10 mm.", "category": "result" }, { "sentence": "Asimismo, aumentar la resolución de las condiciones iniciales y de borde de la media del ensamble aumenta levemente la dispersión y mejora la representación espacial de los patrones de precipitación para todos los umbrales considerados.", "category": "result" } ] }, { "paper_id": "267701303", "title": "PhishDetect: A BiLSTM based phishing URL detection framework using FastText embeddings", "abstract": "As per the Anti-Phishing Working Group's (APWG) report, there were approximately 4.7 million phishing attacks in the year 2022. A significant portion of these phishing attacks were carried out by alluring unsuspecting users into visiting counterfeit web-pages, which are hosted using malicious uniform resource locators (URLs). This paper proposes a lightweight Bidirectional Long Short Term Memory (BiLSTM) based phishing URL detection framework for mitigating the threats posed by phishing attacks. The proposed framework initially splits the URLs into four distinct components namely, Protocol type, domain, sub-domain and top level domain(TLD) using a set of special characters as delimiters. Different feature values corresponding to these four components are then used to build a vocabulary database of the URL corpus. Thereafter, a customized FastText word embedding technique is used to learn numeric feature vector representations of the tokens(URL features) present in the vocabulary database. These learned feature vectors, along with the pre-processed instances of the URL corpus are then provided as input to train a BiLSTM based classifier model for detection of malicious phishing URLs. Experimental results on a proprietary URL corpus comprising 200,000 normal and phishing URL instances show that the proposed framework achieves high accuracy in detecting malicious phishing URLs with minimal computational overhead.", "classified_sentences": [ { "sentence": "As per the Anti-Phishing Working Group's (APWG) report, there were approximately 4.7 million phishing attacks in the year 2022.", "category": "background" }, { "sentence": "A significant portion of these phishing attacks were carried out by alluring unsuspecting users into visiting counterfeit web-pages, which are hosted using malicious uniform resource locators (URLs).", "category": "background" }, { "sentence": "This paper proposes a lightweight Bidirectional Long Short Term Memory (BiLSTM) based phishing URL detection framework for mitigating the threats posed by phishing attacks.", "category": "method" }, { "sentence": "The proposed framework initially splits the URLs into four distinct components namely, Protocol type, domain, sub-domain and top level domain(TLD) using a set of special characters as delimiters.", "category": "method" }, { "sentence": "Different feature values corresponding to these four components are then used to build a vocabulary database of the URL corpus.", "category": "method" }, { "sentence": "Thereafter, a customized FastText word embedding technique is used to learn numeric feature vector representations of the tokens(URL features) present in the vocabulary database.", "category": "method" }, { "sentence": "These learned feature vectors, along with the pre-processed instances of the URL corpus are then provided as input to train a BiLSTM based classifier model for detection of malicious phishing URLs.", "category": "method" }, { "sentence": "Experimental results on a proprietary URL corpus comprising 200,000 normal and phishing URL instances show that the proposed framework achieves high accuracy in detecting malicious phishing URLs with minimal computational overhead.", "category": "result" } ] } ]