id
stringlengths
9
10
submitter
stringlengths
1
64
authors
stringlengths
4
20.7k
title
stringlengths
4
246
comments
stringlengths
1
523
journal-ref
stringlengths
4
404
doi
stringlengths
11
153
report-no
stringlengths
2
254
categories
stringlengths
5
98
license
stringclasses
9 values
orig_abstract
stringlengths
14
3.35k
versions
listlengths
1
60
update_date
stringlengths
10
10
authors_parsed
listlengths
1
1.35k
abstract
stringlengths
11
3.34k
1907.01727
Darion Cassel
Darion Cassel, Yan Huang, Limin Jia
Uncovering Information Flow Policy Violations in C Programs
null
null
null
null
cs.CR cs.PL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Programmers of cryptographic applications written in C need to avoid common mistakes such as sending private data over public channels, modifying trusted data with untrusted functions, or improperly ordering protocol steps. These secrecy, integrity, and sequencing policies can be cumbersome to check with existing general-purpose tools. We have developed a novel means of specifying and uncovering violations of these policies that allows for a much lighter-weight approach than previous tools. We embed the policy annotations in C's type system via a source-to-source translation and leverage existing C compilers to check for policy violations, achieving high performance and scalability. We show through case studies of recent cryptographic libraries and applications that our work is able to express detailed policies for large bodies of C code and can find subtle policy violations. To gain formal understanding of our policy annotations, we show formal connections between the policy annotations and an information flow type system and prove a noninterference guarantee.
[ { "created": "Wed, 3 Jul 2019 04:12:11 GMT", "version": "v1" } ]
2019-07-04
[ [ "Cassel", "Darion", "" ], [ "Huang", "Yan", "" ], [ "Jia", "Limin", "" ] ]
Programmers of cryptographic applications written in C need to avoid common mistakes such as sending private data over public channels, modifying trusted data with untrusted functions, or improperly ordering protocol steps. These secrecy, integrity, and sequencing policies can be cumbersome to check with existing general-purpose tools. We have developed a novel means of specifying and uncovering violations of these policies that allows for a much lighter-weight approach than previous tools. We embed the policy annotations in C's type system via a source-to-source translation and leverage existing C compilers to check for policy violations, achieving high performance and scalability. We show through case studies of recent cryptographic libraries and applications that our work is able to express detailed policies for large bodies of C code and can find subtle policy violations. To gain formal understanding of our policy annotations, we show formal connections between the policy annotations and an information flow type system and prove a noninterference guarantee.
1906.09463
Lucas Gren
Emanuel Mellblom, Isar Arason, Lucas Gren and Richard Torkar
The Connection Between Burnout and Personality Types in Software Developers
null
IEEE Software, 36(5), 2019
10.1109/MS.2019.2924769
null
cs.SE
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
This paper examines the connection between the Five Factor Model personality traits and burnout in software developers. This study aims to validate generalizations of findings in other fields. An online survey consisting of a miniaturized International Personality Item Pool questionnaire for measuring the Five Factor Model personality traits, and the Shirom-Melamed Burnout Measure for measuring burnout, were distributed to open source developer mailing lists, obtaining 47 valid responses. The results from a Bayesian Linear Regression analysis indicate a strong link between neuroticism and burnout confirming previous work, while the other Five Factor Model traits were not adding power to the model. It is important to note that we did not investigate the quality of work in connection to personality, nor did we take any other confounding factors into account like, for example, teamwork. Nonetheless, employers could be aware of, and support, software developers with high neuroticism.
[ { "created": "Sat, 22 Jun 2019 15:40:54 GMT", "version": "v1" } ]
2019-06-25
[ [ "Mellblom", "Emanuel", "" ], [ "Arason", "Isar", "" ], [ "Gren", "Lucas", "" ], [ "Torkar", "Richard", "" ] ]
This paper examines the connection between the Five Factor Model personality traits and burnout in software developers. This study aims to validate generalizations of findings in other fields. An online survey consisting of a miniaturized International Personality Item Pool questionnaire for measuring the Five Factor Model personality traits, and the Shirom-Melamed Burnout Measure for measuring burnout, were distributed to open source developer mailing lists, obtaining 47 valid responses. The results from a Bayesian Linear Regression analysis indicate a strong link between neuroticism and burnout confirming previous work, while the other Five Factor Model traits were not adding power to the model. It is important to note that we did not investigate the quality of work in connection to personality, nor did we take any other confounding factors into account like, for example, teamwork. Nonetheless, employers could be aware of, and support, software developers with high neuroticism.
1703.05486
Oscar De Somer
Oscar De Somer, Ana Soares, Tristan Kuijpers, Koen Vossen, Koen Vanthournout and Fred Spiessens
Using Reinforcement Learning for Demand Response of Domestic Hot Water Buffers: a Real-Life Demonstration
Submitted to IEEE ISGT Europe 2017
null
null
null
cs.SY cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
This paper demonstrates a data-driven control approach for demand response in real-life residential buildings. The objective is to optimally schedule the heating cycles of the Domestic Hot Water (DHW) buffer to maximize the self-consumption of the local photovoltaic (PV) production. A model-based reinforcement learning technique is used to tackle the underlying sequential decision-making problem. The proposed algorithm learns the stochastic occupant behavior, predicts the PV production and takes into account the dynamics of the system. A real-life experiment with six residential buildings is performed using this algorithm. The results show that the self-consumption of the PV production is significantly increased, compared to the default thermostat control.
[ { "created": "Thu, 16 Mar 2017 06:42:07 GMT", "version": "v1" } ]
2017-03-17
[ [ "De Somer", "Oscar", "" ], [ "Soares", "Ana", "" ], [ "Kuijpers", "Tristan", "" ], [ "Vossen", "Koen", "" ], [ "Vanthournout", "Koen", "" ], [ "Spiessens", "Fred", "" ] ]
This paper demonstrates a data-driven control approach for demand response in real-life residential buildings. The objective is to optimally schedule the heating cycles of the Domestic Hot Water (DHW) buffer to maximize the self-consumption of the local photovoltaic (PV) production. A model-based reinforcement learning technique is used to tackle the underlying sequential decision-making problem. The proposed algorithm learns the stochastic occupant behavior, predicts the PV production and takes into account the dynamics of the system. A real-life experiment with six residential buildings is performed using this algorithm. The results show that the self-consumption of the PV production is significantly increased, compared to the default thermostat control.
2406.16629
Melanie JI M\"uller
Melanie J.I. M\"uller
Meta-experiments: Improving experimentation through experimentation
6 pages, 2 figures, 1 table
null
null
null
cs.IR stat.AP
http://creativecommons.org/licenses/by-nc-sa/4.0/
A/B testing is widexly used in the industry to optimize customer facing websites. Many companies employ experimentation specialists to facilitate and improve the process of A/B testing. Here, we present the application of A/B testing to this improvement effort itself, by running experiments on the experimentation process, which we call 'meta-experiments'. We discuss the challenges of this approach using the example of one of our meta-experiments, which helped experimenters to run more sufficiently powered A/B tests. We also point out the benefits of 'dog fooding' for the experimentation specialists when running their own experiments.
[ { "created": "Mon, 24 Jun 2024 13:23:00 GMT", "version": "v1" } ]
2024-06-25
[ [ "Müller", "Melanie J. I.", "" ] ]
A/B testing is widexly used in the industry to optimize customer facing websites. Many companies employ experimentation specialists to facilitate and improve the process of A/B testing. Here, we present the application of A/B testing to this improvement effort itself, by running experiments on the experimentation process, which we call 'meta-experiments'. We discuss the challenges of this approach using the example of one of our meta-experiments, which helped experimenters to run more sufficiently powered A/B tests. We also point out the benefits of 'dog fooding' for the experimentation specialists when running their own experiments.
2203.17191
Daniel Gehrig
Stepan Tulyakov, Alfredo Bochicchio, Daniel Gehrig, Stamatios Georgoulis, Yuanyou Li, and Davide Scaramuzza
Time Lens++: Event-based Frame Interpolation with Parametric Non-linear Flow and Multi-scale Fusion
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, 2022
null
null
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Recently, video frame interpolation using a combination of frame- and event-based cameras has surpassed traditional image-based methods both in terms of performance and memory efficiency. However, current methods still suffer from (i) brittle image-level fusion of complementary interpolation results, that fails in the presence of artifacts in the fused image, (ii) potentially temporally inconsistent and inefficient motion estimation procedures, that run for every inserted frame and (iii) low contrast regions that do not trigger events, and thus cause events-only motion estimation to generate artifacts. Moreover, previous methods were only tested on datasets consisting of planar and faraway scenes, which do not capture the full complexity of the real world. In this work, we address the above problems by introducing multi-scale feature-level fusion and computing one-shot non-linear inter-frame motion from events and images, which can be efficiently sampled for image warping. We also collect the first large-scale events and frames dataset consisting of more than 100 challenging scenes with depth variations, captured with a new experimental setup based on a beamsplitter. We show that our method improves the reconstruction quality by up to 0.2 dB in terms of PSNR and up to 15% in LPIPS score.
[ { "created": "Thu, 31 Mar 2022 17:14:58 GMT", "version": "v1" }, { "created": "Mon, 25 Apr 2022 09:39:00 GMT", "version": "v2" } ]
2022-04-26
[ [ "Tulyakov", "Stepan", "" ], [ "Bochicchio", "Alfredo", "" ], [ "Gehrig", "Daniel", "" ], [ "Georgoulis", "Stamatios", "" ], [ "Li", "Yuanyou", "" ], [ "Scaramuzza", "Davide", "" ] ]
Recently, video frame interpolation using a combination of frame- and event-based cameras has surpassed traditional image-based methods both in terms of performance and memory efficiency. However, current methods still suffer from (i) brittle image-level fusion of complementary interpolation results, that fails in the presence of artifacts in the fused image, (ii) potentially temporally inconsistent and inefficient motion estimation procedures, that run for every inserted frame and (iii) low contrast regions that do not trigger events, and thus cause events-only motion estimation to generate artifacts. Moreover, previous methods were only tested on datasets consisting of planar and faraway scenes, which do not capture the full complexity of the real world. In this work, we address the above problems by introducing multi-scale feature-level fusion and computing one-shot non-linear inter-frame motion from events and images, which can be efficiently sampled for image warping. We also collect the first large-scale events and frames dataset consisting of more than 100 challenging scenes with depth variations, captured with a new experimental setup based on a beamsplitter. We show that our method improves the reconstruction quality by up to 0.2 dB in terms of PSNR and up to 15% in LPIPS score.
2004.04305
Shahin Shayandeh
Swadheen Shukla, Lars Liden, Shahin Shayandeh, Eslam Kamal, Jinchao Li, Matt Mazzola, Thomas Park, Baolin Peng, Jianfeng Gao
Conversation Learner -- A Machine Teaching Tool for Building Dialog Managers for Task-Oriented Dialog Systems
Accepted to ACL 2020 Demonstration Track
null
null
null
cs.CL cs.AI
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Traditionally, industry solutions for building a task-oriented dialog system have relied on helping dialog authors define rule-based dialog managers, represented as dialog flows. While dialog flows are intuitively interpretable and good for simple scenarios, they fall short of performance in terms of the flexibility needed to handle complex dialogs. On the other hand, purely machine-learned models can handle complex dialogs, but they are considered to be black boxes and require large amounts of training data. In this demonstration, we showcase Conversation Learner, a machine teaching tool for building dialog managers. It combines the best of both approaches by enabling dialog authors to create a dialog flow using familiar tools, converting the dialog flow into a parametric model (e.g., neural networks), and allowing dialog authors to improve the dialog manager (i.e., the parametric model) over time by leveraging user-system dialog logs as training data through a machine teaching interface.
[ { "created": "Thu, 9 Apr 2020 00:10:54 GMT", "version": "v1" }, { "created": "Fri, 1 May 2020 20:14:05 GMT", "version": "v2" } ]
2020-05-05
[ [ "Shukla", "Swadheen", "" ], [ "Liden", "Lars", "" ], [ "Shayandeh", "Shahin", "" ], [ "Kamal", "Eslam", "" ], [ "Li", "Jinchao", "" ], [ "Mazzola", "Matt", "" ], [ "Park", "Thomas", "" ], [ "Peng", "Baolin", "" ], [ "Gao", "Jianfeng", "" ] ]
Traditionally, industry solutions for building a task-oriented dialog system have relied on helping dialog authors define rule-based dialog managers, represented as dialog flows. While dialog flows are intuitively interpretable and good for simple scenarios, they fall short of performance in terms of the flexibility needed to handle complex dialogs. On the other hand, purely machine-learned models can handle complex dialogs, but they are considered to be black boxes and require large amounts of training data. In this demonstration, we showcase Conversation Learner, a machine teaching tool for building dialog managers. It combines the best of both approaches by enabling dialog authors to create a dialog flow using familiar tools, converting the dialog flow into a parametric model (e.g., neural networks), and allowing dialog authors to improve the dialog manager (i.e., the parametric model) over time by leveraging user-system dialog logs as training data through a machine teaching interface.
2105.11989
Rushabh Patel
Rushabh Patel, Yanhui Guo
Graph Based Link Prediction between Human Phenotypes and Genes
null
null
null
null
cs.AI cs.LG cs.NE
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Background: The learning of genotype-phenotype associations and history of human disease by doing detailed and precise analysis of phenotypic abnormalities can be defined as deep phenotyping. To understand and detect this interaction between phenotype and genotype is a fundamental step when translating precision medicine to clinical practice. The recent advances in the field of machine learning is efficient to predict these interactions between abnormal human phenotypes and genes. Methods: In this study, we developed a framework to predict links between human phenotype ontology (HPO) and genes. The annotation data from the heterogeneous knowledge resources i.e., orphanet, is used to parse human phenotype-gene associations. To generate the embeddings for the nodes (HPO & genes), an algorithm called node2vec was used. It performs node sampling on this graph based on random walks, then learns features over these sampled nodes to generate embeddings. These embeddings were used to perform the downstream task to predict the presence of the link between these nodes using 5 different supervised machine learning algorithms. Results: The downstream link prediction task shows that the Gradient Boosting Decision Tree based model (LightGBM) achieved an optimal AUROC 0.904 and AUCPR 0.784. In addition, LightGBM achieved an optimal weighted F1 score of 0.87. Compared to the other 4 methods LightGBM is able to find more accurate interaction/link between human phenotype & gene pairs.
[ { "created": "Tue, 25 May 2021 14:47:07 GMT", "version": "v1" }, { "created": "Tue, 1 Jun 2021 18:28:30 GMT", "version": "v2" } ]
2021-06-04
[ [ "Patel", "Rushabh", "" ], [ "Guo", "Yanhui", "" ] ]
Background: The learning of genotype-phenotype associations and history of human disease by doing detailed and precise analysis of phenotypic abnormalities can be defined as deep phenotyping. To understand and detect this interaction between phenotype and genotype is a fundamental step when translating precision medicine to clinical practice. The recent advances in the field of machine learning is efficient to predict these interactions between abnormal human phenotypes and genes. Methods: In this study, we developed a framework to predict links between human phenotype ontology (HPO) and genes. The annotation data from the heterogeneous knowledge resources i.e., orphanet, is used to parse human phenotype-gene associations. To generate the embeddings for the nodes (HPO & genes), an algorithm called node2vec was used. It performs node sampling on this graph based on random walks, then learns features over these sampled nodes to generate embeddings. These embeddings were used to perform the downstream task to predict the presence of the link between these nodes using 5 different supervised machine learning algorithms. Results: The downstream link prediction task shows that the Gradient Boosting Decision Tree based model (LightGBM) achieved an optimal AUROC 0.904 and AUCPR 0.784. In addition, LightGBM achieved an optimal weighted F1 score of 0.87. Compared to the other 4 methods LightGBM is able to find more accurate interaction/link between human phenotype & gene pairs.
2205.06266
Jonas Pfeiffer
Jonas Pfeiffer, Naman Goyal, Xi Victoria Lin, Xian Li, James Cross, Sebastian Riedel, Mikel Artetxe
Lifting the Curse of Multilinguality by Pre-training Modular Transformers
NAACL 2022
null
null
null
cs.CL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Multilingual pre-trained models are known to suffer from the curse of multilinguality, which causes per-language performance to drop as they cover more languages. We address this issue by introducing language-specific modules, which allows us to grow the total capacity of the model, while keeping the total number of trainable parameters per language constant. In contrast with prior work that learns language-specific components post-hoc, we pre-train the modules of our Cross-lingual Modular (X-Mod) models from the start. Our experiments on natural language inference, named entity recognition and question answering show that our approach not only mitigates the negative interference between languages, but also enables positive transfer, resulting in improved monolingual and cross-lingual performance. Furthermore, our approach enables adding languages post-hoc with no measurable drop in performance, no longer limiting the model usage to the set of pre-trained languages.
[ { "created": "Thu, 12 May 2022 17:59:56 GMT", "version": "v1" } ]
2022-05-13
[ [ "Pfeiffer", "Jonas", "" ], [ "Goyal", "Naman", "" ], [ "Lin", "Xi Victoria", "" ], [ "Li", "Xian", "" ], [ "Cross", "James", "" ], [ "Riedel", "Sebastian", "" ], [ "Artetxe", "Mikel", "" ] ]
Multilingual pre-trained models are known to suffer from the curse of multilinguality, which causes per-language performance to drop as they cover more languages. We address this issue by introducing language-specific modules, which allows us to grow the total capacity of the model, while keeping the total number of trainable parameters per language constant. In contrast with prior work that learns language-specific components post-hoc, we pre-train the modules of our Cross-lingual Modular (X-Mod) models from the start. Our experiments on natural language inference, named entity recognition and question answering show that our approach not only mitigates the negative interference between languages, but also enables positive transfer, resulting in improved monolingual and cross-lingual performance. Furthermore, our approach enables adding languages post-hoc with no measurable drop in performance, no longer limiting the model usage to the set of pre-trained languages.
2402.05687
Gabriel Martins de Jesus
Gabriel Martins de Jesus, Onel Luis Alcaraz Lopez, Richard Demo Souza, Nurul Huda Mahmood, Markku Juntti, Matti Latva-Aho
Assessment of the Sparsity-Diversity Trade-offs in Active Users Detection for mMTC
5 pages, 5 figures. Manuscript submitted to IEEE Wireless Communications Letters for review
null
null
null
cs.IT eess.SP math.IT
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Wireless communication systems must increasingly support a multitude of machine-type communications (MTC) devices, thus calling for advanced strategies for active user detection (AUD). Recent literature has delved into AUD techniques based on compressed sensing, highlighting the critical role of signal sparsity. This study investigates the relationship between frequency diversity and signal sparsity in the AUD problem. Single-antenna users transmit multiple copies of non-orthogonal pilots across multiple frequency channels and the base station independently performs AUD in each channel using the orthogonal matching pursuit algorithm. We note that, although frequency diversity may improve the likelihood of successful reception of the signals, it may also damage the channel sparsity level, leading to important trade-offs. We show that a sparser signal significantly benefits AUD, surpassing the advantages brought by frequency diversity in scenarios with limited temporal resources and/or high numbers of receive antennas. Conversely, with longer pilots and fewer receive antennas, investing in frequency diversity becomes more impactful, resulting in a tenfold AUD performance improvement.
[ { "created": "Thu, 8 Feb 2024 14:06:01 GMT", "version": "v1" } ]
2024-02-09
[ [ "de Jesus", "Gabriel Martins", "" ], [ "Lopez", "Onel Luis Alcaraz", "" ], [ "Souza", "Richard Demo", "" ], [ "Mahmood", "Nurul Huda", "" ], [ "Juntti", "Markku", "" ], [ "Latva-Aho", "Matti", "" ] ]
Wireless communication systems must increasingly support a multitude of machine-type communications (MTC) devices, thus calling for advanced strategies for active user detection (AUD). Recent literature has delved into AUD techniques based on compressed sensing, highlighting the critical role of signal sparsity. This study investigates the relationship between frequency diversity and signal sparsity in the AUD problem. Single-antenna users transmit multiple copies of non-orthogonal pilots across multiple frequency channels and the base station independently performs AUD in each channel using the orthogonal matching pursuit algorithm. We note that, although frequency diversity may improve the likelihood of successful reception of the signals, it may also damage the channel sparsity level, leading to important trade-offs. We show that a sparser signal significantly benefits AUD, surpassing the advantages brought by frequency diversity in scenarios with limited temporal resources and/or high numbers of receive antennas. Conversely, with longer pilots and fewer receive antennas, investing in frequency diversity becomes more impactful, resulting in a tenfold AUD performance improvement.
1809.10842
Yi Wu
Yi Wu, Yuxin Wu, Aviv Tamar, Stuart Russell, Georgia Gkioxari, Yuandong Tian
Learning and Planning with a Semantic Model
submitted to ICLR 2019
null
null
null
cs.LG cs.AI stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Building deep reinforcement learning agents that can generalize and adapt to unseen environments remains a fundamental challenge for AI. This paper describes progresses on this challenge in the context of man-made environments, which are visually diverse but contain intrinsic semantic regularities. We propose a hybrid model-based and model-free approach, LEArning and Planning with Semantics (LEAPS), consisting of a multi-target sub-policy that acts on visual inputs, and a Bayesian model over semantic structures. When placed in an unseen environment, the agent plans with the semantic model to make high-level decisions, proposes the next sub-target for the sub-policy to execute, and updates the semantic model based on new observations. We perform experiments in visual navigation tasks using House3D, a 3D environment that contains diverse human-designed indoor scenes with real-world objects. LEAPS outperforms strong baselines that do not explicitly plan using the semantic content.
[ { "created": "Fri, 28 Sep 2018 03:30:37 GMT", "version": "v1" } ]
2018-10-01
[ [ "Wu", "Yi", "" ], [ "Wu", "Yuxin", "" ], [ "Tamar", "Aviv", "" ], [ "Russell", "Stuart", "" ], [ "Gkioxari", "Georgia", "" ], [ "Tian", "Yuandong", "" ] ]
Building deep reinforcement learning agents that can generalize and adapt to unseen environments remains a fundamental challenge for AI. This paper describes progresses on this challenge in the context of man-made environments, which are visually diverse but contain intrinsic semantic regularities. We propose a hybrid model-based and model-free approach, LEArning and Planning with Semantics (LEAPS), consisting of a multi-target sub-policy that acts on visual inputs, and a Bayesian model over semantic structures. When placed in an unseen environment, the agent plans with the semantic model to make high-level decisions, proposes the next sub-target for the sub-policy to execute, and updates the semantic model based on new observations. We perform experiments in visual navigation tasks using House3D, a 3D environment that contains diverse human-designed indoor scenes with real-world objects. LEAPS outperforms strong baselines that do not explicitly plan using the semantic content.
1810.11530
Pascal Lamblin
Bart van Merri\"enboer, Olivier Breuleux, Arnaud Bergeron, Pascal Lamblin
Automatic differentiation in ML: Where we are and where we should be going
null
null
null
null
cs.LG cs.PL stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We review the current state of automatic differentiation (AD) for array programming in machine learning (ML), including the different approaches such as operator overloading (OO) and source transformation (ST) used for AD, graph-based intermediate representations for programs, and source languages. Based on these insights, we introduce a new graph-based intermediate representation (IR) which specifically aims to efficiently support fully-general AD for array programming. Unlike existing dataflow programming representations in ML frameworks, our IR naturally supports function calls, higher-order functions and recursion, making ML models easier to implement. The ability to represent closures allows us to perform AD using ST without a tape, making the resulting derivative (adjoint) program amenable to ahead-of-time optimization using tools from functional language compilers, and enabling higher-order derivatives. Lastly, we introduce a proof of concept compiler toolchain called Myia which uses a subset of Python as a front end.
[ { "created": "Fri, 26 Oct 2018 21:09:07 GMT", "version": "v1" }, { "created": "Wed, 2 Jan 2019 21:16:54 GMT", "version": "v2" } ]
2019-01-04
[ [ "van Merriënboer", "Bart", "" ], [ "Breuleux", "Olivier", "" ], [ "Bergeron", "Arnaud", "" ], [ "Lamblin", "Pascal", "" ] ]
We review the current state of automatic differentiation (AD) for array programming in machine learning (ML), including the different approaches such as operator overloading (OO) and source transformation (ST) used for AD, graph-based intermediate representations for programs, and source languages. Based on these insights, we introduce a new graph-based intermediate representation (IR) which specifically aims to efficiently support fully-general AD for array programming. Unlike existing dataflow programming representations in ML frameworks, our IR naturally supports function calls, higher-order functions and recursion, making ML models easier to implement. The ability to represent closures allows us to perform AD using ST without a tape, making the resulting derivative (adjoint) program amenable to ahead-of-time optimization using tools from functional language compilers, and enabling higher-order derivatives. Lastly, we introduce a proof of concept compiler toolchain called Myia which uses a subset of Python as a front end.
2108.04197
Haokai Hong
Haokai Hong, Kai Ye, Min Jiang, Donglin Cao, Kay Chen Tan
Solving Large-Scale Multi-Objective Optimization via Probabilistic Prediction Model
17 pages, 2 figures
null
null
null
cs.NE cs.AI
http://creativecommons.org/licenses/by/4.0/
The main feature of large-scale multi-objective optimization problems (LSMOP) is to optimize multiple conflicting objectives while considering thousands of decision variables at the same time. An efficient LSMOP algorithm should have the ability to escape the local optimal solution from the huge search space and find the global optimal. Most of the current researches focus on how to deal with decision variables. However, due to the large number of decision variables, it is easy to lead to high computational cost. Maintaining the diversity of the population is one of the effective ways to improve search efficiency. In this paper, we propose a probabilistic prediction model based on trend prediction model and generating-filtering strategy, called LT-PPM, to tackle the LSMOP. The proposed method enhances the diversity of the population through importance sampling. At the same time, due to the adoption of an individual-based evolution mechanism, the computational cost of the proposed method is independent of the number of decision variables, thus avoiding the problem of exponential growth of the search space. We compared the proposed algorithm with several state-of-the-art algorithms for different benchmark functions. The experimental results and complexity analysis have demonstrated that the proposed algorithm has significant improvement in terms of its performance and computational efficiency in large-scale multi-objective optimization.
[ { "created": "Fri, 16 Jul 2021 09:43:35 GMT", "version": "v1" } ]
2021-08-10
[ [ "Hong", "Haokai", "" ], [ "Ye", "Kai", "" ], [ "Jiang", "Min", "" ], [ "Cao", "Donglin", "" ], [ "Tan", "Kay Chen", "" ] ]
The main feature of large-scale multi-objective optimization problems (LSMOP) is to optimize multiple conflicting objectives while considering thousands of decision variables at the same time. An efficient LSMOP algorithm should have the ability to escape the local optimal solution from the huge search space and find the global optimal. Most of the current researches focus on how to deal with decision variables. However, due to the large number of decision variables, it is easy to lead to high computational cost. Maintaining the diversity of the population is one of the effective ways to improve search efficiency. In this paper, we propose a probabilistic prediction model based on trend prediction model and generating-filtering strategy, called LT-PPM, to tackle the LSMOP. The proposed method enhances the diversity of the population through importance sampling. At the same time, due to the adoption of an individual-based evolution mechanism, the computational cost of the proposed method is independent of the number of decision variables, thus avoiding the problem of exponential growth of the search space. We compared the proposed algorithm with several state-of-the-art algorithms for different benchmark functions. The experimental results and complexity analysis have demonstrated that the proposed algorithm has significant improvement in terms of its performance and computational efficiency in large-scale multi-objective optimization.
1911.08706
Xing Niu
Xing Niu, Marine Carpuat
Controlling Neural Machine Translation Formality with Synthetic Supervision
Accepted at AAAI 2020
null
null
null
cs.CL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
This work aims to produce translations that convey source language content at a formality level that is appropriate for a particular audience. Framing this problem as a neural sequence-to-sequence task ideally requires training triplets consisting of a bilingual sentence pair labeled with target language formality. However, in practice, available training examples are limited to English sentence pairs of different styles, and bilingual parallel sentences of unknown formality. We introduce a novel training scheme for multi-task models that automatically generates synthetic training triplets by inferring the missing element on the fly, thus enabling end-to-end training. Comprehensive automatic and human assessments show that our best model outperforms existing models by producing translations that better match desired formality levels while preserving the source meaning.
[ { "created": "Wed, 20 Nov 2019 04:54:21 GMT", "version": "v1" }, { "created": "Thu, 28 Nov 2019 00:55:36 GMT", "version": "v2" } ]
2019-12-02
[ [ "Niu", "Xing", "" ], [ "Carpuat", "Marine", "" ] ]
This work aims to produce translations that convey source language content at a formality level that is appropriate for a particular audience. Framing this problem as a neural sequence-to-sequence task ideally requires training triplets consisting of a bilingual sentence pair labeled with target language formality. However, in practice, available training examples are limited to English sentence pairs of different styles, and bilingual parallel sentences of unknown formality. We introduce a novel training scheme for multi-task models that automatically generates synthetic training triplets by inferring the missing element on the fly, thus enabling end-to-end training. Comprehensive automatic and human assessments show that our best model outperforms existing models by producing translations that better match desired formality levels while preserving the source meaning.
1805.12393
Yuyu Zhang
Yuyu Zhang, Hanjun Dai, Kamil Toraman, Le Song
KG^2: Learning to Reason Science Exam Questions with Contextual Knowledge Graph Embeddings
null
null
null
null
cs.LG cs.AI cs.CL stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
The AI2 Reasoning Challenge (ARC), a new benchmark dataset for question answering (QA) has been recently released. ARC only contains natural science questions authored for human exams, which are hard to answer and require advanced logic reasoning. On the ARC Challenge Set, existing state-of-the-art QA systems fail to significantly outperform random baseline, reflecting the difficult nature of this task. In this paper, we propose a novel framework for answering science exam questions, which mimics human solving process in an open-book exam. To address the reasoning challenge, we construct contextual knowledge graphs respectively for the question itself and supporting sentences. Our model learns to reason with neural embeddings of both knowledge graphs. Experiments on the ARC Challenge Set show that our model outperforms the previous state-of-the-art QA systems.
[ { "created": "Thu, 31 May 2018 09:39:14 GMT", "version": "v1" } ]
2018-06-01
[ [ "Zhang", "Yuyu", "" ], [ "Dai", "Hanjun", "" ], [ "Toraman", "Kamil", "" ], [ "Song", "Le", "" ] ]
The AI2 Reasoning Challenge (ARC), a new benchmark dataset for question answering (QA) has been recently released. ARC only contains natural science questions authored for human exams, which are hard to answer and require advanced logic reasoning. On the ARC Challenge Set, existing state-of-the-art QA systems fail to significantly outperform random baseline, reflecting the difficult nature of this task. In this paper, we propose a novel framework for answering science exam questions, which mimics human solving process in an open-book exam. To address the reasoning challenge, we construct contextual knowledge graphs respectively for the question itself and supporting sentences. Our model learns to reason with neural embeddings of both knowledge graphs. Experiments on the ARC Challenge Set show that our model outperforms the previous state-of-the-art QA systems.
1307.3054
Sucheta Shrivastava
Sayali Nimkar, Sanal Varghese, Sucheta Shrivastava
Contrast Enhancement And Brightness Preservation Using Multi- Decomposition Histogram Equalization
9 pages,13 figures
SIPIJ, Vol.4, Issue.3, pp. 85-93
10.5121/sipij.2013.4308
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Histogram Equalization (HE) has been an essential addition to the Image Enhancement world. Enhancement techniques like Classical Histogram Equalization (CHE), Adaptive Histogram Equalization (ADHE), Bi-Histogram Equalization (BHE) and Recursive Mean Separate Histogram Equalization (RMSHE) methods enhance contrast, however, brightness is not well preserved with these methods, which gives an unpleasant look to the final image obtained. Thus, we introduce a novel technique Multi-Decomposition Histogram Equalization (MDHE) to eliminate the drawbacks of the earlier methods. In MDHE, we have decomposed the input sixty-four parts, applied CHE in each of the sub-images and then finally interpolated them in correct order. The final image after MDHE results in contrast enhanced and brightness preserved image compared to all other techniques mentioned above. We have calculated the various parameters like PSNR, SNR, RMSE, MSE, etc. for every technique. Our results are well supported by bar graphs, histograms and the parameter calculations at the end.
[ { "created": "Thu, 11 Jul 2013 11:02:57 GMT", "version": "v1" } ]
2013-07-12
[ [ "Nimkar", "Sayali", "" ], [ "Varghese", "Sanal", "" ], [ "Shrivastava", "Sucheta", "" ] ]
Histogram Equalization (HE) has been an essential addition to the Image Enhancement world. Enhancement techniques like Classical Histogram Equalization (CHE), Adaptive Histogram Equalization (ADHE), Bi-Histogram Equalization (BHE) and Recursive Mean Separate Histogram Equalization (RMSHE) methods enhance contrast, however, brightness is not well preserved with these methods, which gives an unpleasant look to the final image obtained. Thus, we introduce a novel technique Multi-Decomposition Histogram Equalization (MDHE) to eliminate the drawbacks of the earlier methods. In MDHE, we have decomposed the input sixty-four parts, applied CHE in each of the sub-images and then finally interpolated them in correct order. The final image after MDHE results in contrast enhanced and brightness preserved image compared to all other techniques mentioned above. We have calculated the various parameters like PSNR, SNR, RMSE, MSE, etc. for every technique. Our results are well supported by bar graphs, histograms and the parameter calculations at the end.
1011.3170
James Aspnes
James Aspnes
Slightly smaller splitter networks
null
null
null
null
cs.DC
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
The classic renaming protocol of Moir and Anderson (1995) uses a network of Theta(n^2) splitters to assign unique names to n processes with unbounded initial names. We show how to reduce this bound to Theta(n^{3/2}) splitters.
[ { "created": "Sun, 14 Nov 2010 00:52:14 GMT", "version": "v1" } ]
2010-11-16
[ [ "Aspnes", "James", "" ] ]
The classic renaming protocol of Moir and Anderson (1995) uses a network of Theta(n^2) splitters to assign unique names to n processes with unbounded initial names. We show how to reduce this bound to Theta(n^{3/2}) splitters.
1907.11034
Petra Van Den Bos
Petra van den Bos and Frits Vaandrager
State Identification for Labeled Transition Systems with Inputs and Outputs
null
null
null
null
cs.FL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
For Finite State Machines (FSMs) a rich testing theory has been developed to discover aspects of their behavior and ensure their correct functioning. Although this theory is widely used, e.g., to check conformance of protocol implementations, its applicability is limited by restrictions of the FSM framework: the fact that inputs and outputs alternate in an FSM, and outputs are fully determined by the previous input and state. Labeled Transition Systems with inputs and outputs (LTSs), as studied in ioco testing theory, provide a richer framework for testing component oriented systems, but lack the algorithms for test generation from FSM theory. In this article, we propose an algorithm for the fundamental problem of state identification during testing of LTSs. Our algorithm is a direct generalization of the well-known algorithm for computing adaptive distinguishing sequences for FSMs proposed by Lee & Yannakakis. Our algorithm has to deal with so-called compatible states, states that cannot be distinguished in case of an adversarial system-under-test. Analogous to the result of Lee & Yannakakis, we prove that if an (adaptive) test exists that distinguishes all pairs of incompatible states of an LTS, our algorithm will find one. In practice, such adaptive tests typically do not exist. However, in experiments with an implementation of our algorithm on an industrial benchmark, we find that tests produced by our algorithm still distinguish more than 99% of the incompatible state pairs.
[ { "created": "Thu, 25 Jul 2019 13:28:31 GMT", "version": "v1" }, { "created": "Tue, 22 Oct 2019 14:43:22 GMT", "version": "v2" } ]
2019-10-23
[ [ "Bos", "Petra van den", "" ], [ "Vaandrager", "Frits", "" ] ]
For Finite State Machines (FSMs) a rich testing theory has been developed to discover aspects of their behavior and ensure their correct functioning. Although this theory is widely used, e.g., to check conformance of protocol implementations, its applicability is limited by restrictions of the FSM framework: the fact that inputs and outputs alternate in an FSM, and outputs are fully determined by the previous input and state. Labeled Transition Systems with inputs and outputs (LTSs), as studied in ioco testing theory, provide a richer framework for testing component oriented systems, but lack the algorithms for test generation from FSM theory. In this article, we propose an algorithm for the fundamental problem of state identification during testing of LTSs. Our algorithm is a direct generalization of the well-known algorithm for computing adaptive distinguishing sequences for FSMs proposed by Lee & Yannakakis. Our algorithm has to deal with so-called compatible states, states that cannot be distinguished in case of an adversarial system-under-test. Analogous to the result of Lee & Yannakakis, we prove that if an (adaptive) test exists that distinguishes all pairs of incompatible states of an LTS, our algorithm will find one. In practice, such adaptive tests typically do not exist. However, in experiments with an implementation of our algorithm on an industrial benchmark, we find that tests produced by our algorithm still distinguish more than 99% of the incompatible state pairs.
1509.02796
Johannes K\"oster
Johannes K\"oster
Rust-Bio - a fast and safe bioinformatics library
null
null
10.1093/bioinformatics/btv573
null
cs.MS
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We present Rust-Bio, the first general purpose bioinformatics library for the innovative Rust programming language. Rust-Bio leverages the unique combination of speed, memory safety and high-level syntax offered by Rust to provide a fast and safe set of bioinformatics algorithms and data structures with a focus on sequence analysis.
[ { "created": "Wed, 9 Sep 2015 14:53:02 GMT", "version": "v1" } ]
2015-10-08
[ [ "Köster", "Johannes", "" ] ]
We present Rust-Bio, the first general purpose bioinformatics library for the innovative Rust programming language. Rust-Bio leverages the unique combination of speed, memory safety and high-level syntax offered by Rust to provide a fast and safe set of bioinformatics algorithms and data structures with a focus on sequence analysis.
2406.13188
Naiming (Lucy) Liu
Naiming Liu, Zichao Wang, Richard Baraniuk
Synthetic Context Generation for Question Generation
null
null
null
null
cs.CL cs.LG
http://creativecommons.org/licenses/by/4.0/
Despite rapid advancements in large language models (LLMs), QG remains a challenging problem due to its complicated process, open-ended nature, and the diverse settings in which question generation occurs. A common approach to address these challenges involves fine-tuning smaller, custom models using datasets containing background context, question, and answer. However, obtaining suitable domain-specific datasets with appropriate context is often more difficult than acquiring question-answer pairs. In this paper, we investigate training QG models using synthetic contexts generated by LLMs from readily available question-answer pairs. We conduct a comprehensive study to answer critical research questions related to the performance of models trained on synthetic contexts and their potential impact on QG research and applications. Our empirical results reveal: 1) contexts are essential for QG tasks, even if they are synthetic; 2) fine-tuning smaller language models has the capability of achieving better performances as compared to prompting larger language models; and 3) synthetic context and real context could achieve comparable performances. These findings highlight the effectiveness of synthetic contexts in QG and paves the way for future advancements in the field.
[ { "created": "Wed, 19 Jun 2024 03:37:52 GMT", "version": "v1" } ]
2024-06-21
[ [ "Liu", "Naiming", "" ], [ "Wang", "Zichao", "" ], [ "Baraniuk", "Richard", "" ] ]
Despite rapid advancements in large language models (LLMs), QG remains a challenging problem due to its complicated process, open-ended nature, and the diverse settings in which question generation occurs. A common approach to address these challenges involves fine-tuning smaller, custom models using datasets containing background context, question, and answer. However, obtaining suitable domain-specific datasets with appropriate context is often more difficult than acquiring question-answer pairs. In this paper, we investigate training QG models using synthetic contexts generated by LLMs from readily available question-answer pairs. We conduct a comprehensive study to answer critical research questions related to the performance of models trained on synthetic contexts and their potential impact on QG research and applications. Our empirical results reveal: 1) contexts are essential for QG tasks, even if they are synthetic; 2) fine-tuning smaller language models has the capability of achieving better performances as compared to prompting larger language models; and 3) synthetic context and real context could achieve comparable performances. These findings highlight the effectiveness of synthetic contexts in QG and paves the way for future advancements in the field.
2103.16185
Jakub Ruszil
Jakub Ruszil
Approximation algorithm for finding short synchronizing words in weighted automata
9 pages, 2 figures
null
null
null
cs.FL cs.DM cs.DS math.CO
http://creativecommons.org/licenses/by/4.0/
In this paper we are dealing with the issue of finding possibly short synchronizing words in automata with weight assigned to each letter in the alphabet $\Sigma$. First we discuss some complexity problems, and then we present new approximation algorithm in four variations.
[ { "created": "Tue, 30 Mar 2021 09:07:57 GMT", "version": "v1" } ]
2021-03-31
[ [ "Ruszil", "Jakub", "" ] ]
In this paper we are dealing with the issue of finding possibly short synchronizing words in automata with weight assigned to each letter in the alphabet $\Sigma$. First we discuss some complexity problems, and then we present new approximation algorithm in four variations.
1206.3634
Ankur Sahai
Ankur Sahai
Balls into Bins: strict Capacities and Edge Weights
null
null
null
null
cs.DS cs.DC
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We explore a novel theoretical model for studying the performance of distributed storage management systems where the data-centers have limited capacities (as compared to storage space requested by the users). Prior schemes such as Balls-into-bins (used for load balancing) neither consider bin (consumer) capacities (multiple balls into a bin) nor the future performance of the system after, balls (producer requests) are allocated to bins and restrict number of balls as a function of the number of bins. Our problem consists of finding an optimal assignment of the online producer requests to consumers (via weighted edges) in a complete bipartite graph while ensuring that the total size of request assigned on a consumer is limited by its capacity. The metric used to measure the performance in this model is the (minimization of) weighted sum of the requests assigned on the edges (loads) and their corresponding weights. We first explore the optimal offline algorithms followed by competitive analysis of different online techniques. Using oblivious adversary. LP and Primal-Dual algorithms are used for calculating the optimal offline solution in O(r*n) time (where r and n are the number of requests and consumers respectively) while randomized algorithms are used for the online case. For the simplified model with equal consumer capacities an average-case competitive ratio of AVG(d) / MIN(d) (where d is the edge weight / distance) is achieved using an algorithm that has equal probability for selecting any of the available edges with a running time of $O(r)$. In the extending the model to arbitrary consumer capacities we show an average case competitive ratio of AVG(d*c) / (AVG(c) *MIN(d)).
[ { "created": "Sat, 16 Jun 2012 07:33:30 GMT", "version": "v1" } ]
2012-06-19
[ [ "Sahai", "Ankur", "" ] ]
We explore a novel theoretical model for studying the performance of distributed storage management systems where the data-centers have limited capacities (as compared to storage space requested by the users). Prior schemes such as Balls-into-bins (used for load balancing) neither consider bin (consumer) capacities (multiple balls into a bin) nor the future performance of the system after, balls (producer requests) are allocated to bins and restrict number of balls as a function of the number of bins. Our problem consists of finding an optimal assignment of the online producer requests to consumers (via weighted edges) in a complete bipartite graph while ensuring that the total size of request assigned on a consumer is limited by its capacity. The metric used to measure the performance in this model is the (minimization of) weighted sum of the requests assigned on the edges (loads) and their corresponding weights. We first explore the optimal offline algorithms followed by competitive analysis of different online techniques. Using oblivious adversary. LP and Primal-Dual algorithms are used for calculating the optimal offline solution in O(r*n) time (where r and n are the number of requests and consumers respectively) while randomized algorithms are used for the online case. For the simplified model with equal consumer capacities an average-case competitive ratio of AVG(d) / MIN(d) (where d is the edge weight / distance) is achieved using an algorithm that has equal probability for selecting any of the available edges with a running time of $O(r)$. In the extending the model to arbitrary consumer capacities we show an average case competitive ratio of AVG(d*c) / (AVG(c) *MIN(d)).
2403.12031
Xiuyu Li
Qitian Jason Hu, Jacob Bieker, Xiuyu Li, Nan Jiang, Benjamin Keigwin, Gaurav Ranganath, Kurt Keutzer, Shriyash Kaustubh Upadhyay
RouterBench: A Benchmark for Multi-LLM Routing System
null
null
null
null
cs.LG cs.AI
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
As the range of applications for Large Language Models (LLMs) continues to grow, the demand for effective serving solutions becomes increasingly critical. Despite the versatility of LLMs, no single model can optimally address all tasks and applications, particularly when balancing performance with cost. This limitation has led to the development of LLM routing systems, which combine the strengths of various models to overcome the constraints of individual LLMs. Yet, the absence of a standardized benchmark for evaluating the performance of LLM routers hinders progress in this area. To bridge this gap, we present RouterBench, a novel evaluation framework designed to systematically assess the efficacy of LLM routing systems, along with a comprehensive dataset comprising over 405k inference outcomes from representative LLMs to support the development of routing strategies. We further propose a theoretical framework for LLM routing, and deliver a comparative analysis of various routing approaches through RouterBench, highlighting their potentials and limitations within our evaluation framework. This work not only formalizes and advances the development of LLM routing systems but also sets a standard for their assessment, paving the way for more accessible and economically viable LLM deployments. The code and data are available at https://github.com/withmartian/routerbench.
[ { "created": "Mon, 18 Mar 2024 17:59:04 GMT", "version": "v1" }, { "created": "Thu, 28 Mar 2024 17:56:28 GMT", "version": "v2" } ]
2024-03-29
[ [ "Hu", "Qitian Jason", "" ], [ "Bieker", "Jacob", "" ], [ "Li", "Xiuyu", "" ], [ "Jiang", "Nan", "" ], [ "Keigwin", "Benjamin", "" ], [ "Ranganath", "Gaurav", "" ], [ "Keutzer", "Kurt", "" ], [ "Upadhyay", "Shriyash Kaustubh", "" ] ]
As the range of applications for Large Language Models (LLMs) continues to grow, the demand for effective serving solutions becomes increasingly critical. Despite the versatility of LLMs, no single model can optimally address all tasks and applications, particularly when balancing performance with cost. This limitation has led to the development of LLM routing systems, which combine the strengths of various models to overcome the constraints of individual LLMs. Yet, the absence of a standardized benchmark for evaluating the performance of LLM routers hinders progress in this area. To bridge this gap, we present RouterBench, a novel evaluation framework designed to systematically assess the efficacy of LLM routing systems, along with a comprehensive dataset comprising over 405k inference outcomes from representative LLMs to support the development of routing strategies. We further propose a theoretical framework for LLM routing, and deliver a comparative analysis of various routing approaches through RouterBench, highlighting their potentials and limitations within our evaluation framework. This work not only formalizes and advances the development of LLM routing systems but also sets a standard for their assessment, paving the way for more accessible and economically viable LLM deployments. The code and data are available at https://github.com/withmartian/routerbench.
2403.13258
Zengqiang Yan
Xian Lin, Yangyang Xiang, Zhehao Wang, Kwang-Ting Cheng, Zengqiang Yan, Li Yu
SAMCT: Segment Any CT Allowing Labor-Free Task-Indicator Prompts
null
null
null
null
cs.CV
http://creativecommons.org/licenses/by/4.0/
Segment anything model (SAM), a foundation model with superior versatility and generalization across diverse segmentation tasks, has attracted widespread attention in medical imaging. However, it has been proved that SAM would encounter severe performance degradation due to the lack of medical knowledge in training and local feature encoding. Though several SAM-based models have been proposed for tuning SAM in medical imaging, they still suffer from insufficient feature extraction and highly rely on high-quality prompts. In this paper, we construct a large CT dataset consisting of 1.1M CT images and 5M masks from public datasets and propose a powerful foundation model SAMCT allowing labor-free prompts. Specifically, based on SAM, SAMCT is further equipped with a U-shaped CNN image encoder, a cross-branch interaction module, and a task-indicator prompt encoder. The U-shaped CNN image encoder works in parallel with the ViT image encoder in SAM to supplement local features. Cross-branch interaction enhances the feature expression capability of the CNN image encoder and the ViT image encoder by exchanging global perception and local features from one to the other. The task-indicator prompt encoder is a plug-and-play component to effortlessly encode task-related indicators into prompt embeddings. In this way, SAMCT can work in an automatic manner in addition to the semi-automatic interactive strategy in SAM. Extensive experiments demonstrate the superiority of SAMCT against the state-of-the-art task-specific and SAM-based medical foundation models on various tasks. The code, data, and models are released at https://github.com/xianlin7/SAMCT.
[ { "created": "Wed, 20 Mar 2024 02:39:15 GMT", "version": "v1" } ]
2024-03-21
[ [ "Lin", "Xian", "" ], [ "Xiang", "Yangyang", "" ], [ "Wang", "Zhehao", "" ], [ "Cheng", "Kwang-Ting", "" ], [ "Yan", "Zengqiang", "" ], [ "Yu", "Li", "" ] ]
Segment anything model (SAM), a foundation model with superior versatility and generalization across diverse segmentation tasks, has attracted widespread attention in medical imaging. However, it has been proved that SAM would encounter severe performance degradation due to the lack of medical knowledge in training and local feature encoding. Though several SAM-based models have been proposed for tuning SAM in medical imaging, they still suffer from insufficient feature extraction and highly rely on high-quality prompts. In this paper, we construct a large CT dataset consisting of 1.1M CT images and 5M masks from public datasets and propose a powerful foundation model SAMCT allowing labor-free prompts. Specifically, based on SAM, SAMCT is further equipped with a U-shaped CNN image encoder, a cross-branch interaction module, and a task-indicator prompt encoder. The U-shaped CNN image encoder works in parallel with the ViT image encoder in SAM to supplement local features. Cross-branch interaction enhances the feature expression capability of the CNN image encoder and the ViT image encoder by exchanging global perception and local features from one to the other. The task-indicator prompt encoder is a plug-and-play component to effortlessly encode task-related indicators into prompt embeddings. In this way, SAMCT can work in an automatic manner in addition to the semi-automatic interactive strategy in SAM. Extensive experiments demonstrate the superiority of SAMCT against the state-of-the-art task-specific and SAM-based medical foundation models on various tasks. The code, data, and models are released at https://github.com/xianlin7/SAMCT.
1609.05876
Roberto Alonso
Roberto Alonso, Ra\'ul Monroy, Eduardo Aguirre
On the Phase Transition of Finding a Biclique in a larger Bipartite Graph
null
null
null
null
cs.AI cs.CC
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We report on the phase transition of finding a complete subgraph, of specified dimensions, in a bipartite graph. Finding a complete subgraph in a bipartite graph is a problem that has growing attention in several domains, including bioinformatics, social network analysis and domain clustering. A key step for a successful phase transition study is identifying a suitable order parameter, when none is known. To this purpose, we have applied a decision tree classifier to real-world instances of this problem, in order to understand what problem features separate an instance that is hard to solve from those that is not. We have successfully identified one such order parameter and with it the phase transition of finding a complete bipartite subgraph of specified dimensions. Our phase transition study shows an easy-to-hard-to-easy-to-hard-to-easy pattern. Further, our results indicate that the hardest instances are in a region where it is more likely that the corresponding bipartite graph will have a complete subgraph of specified dimensions, a positive answer. By contrast, instances with a negative answer are more likely to appear in a region where the computational cost is negligible. This behaviour is remarkably similar for problems of a number of different sizes.
[ { "created": "Mon, 19 Sep 2016 19:22:08 GMT", "version": "v1" } ]
2016-09-20
[ [ "Alonso", "Roberto", "" ], [ "Monroy", "Raúl", "" ], [ "Aguirre", "Eduardo", "" ] ]
We report on the phase transition of finding a complete subgraph, of specified dimensions, in a bipartite graph. Finding a complete subgraph in a bipartite graph is a problem that has growing attention in several domains, including bioinformatics, social network analysis and domain clustering. A key step for a successful phase transition study is identifying a suitable order parameter, when none is known. To this purpose, we have applied a decision tree classifier to real-world instances of this problem, in order to understand what problem features separate an instance that is hard to solve from those that is not. We have successfully identified one such order parameter and with it the phase transition of finding a complete bipartite subgraph of specified dimensions. Our phase transition study shows an easy-to-hard-to-easy-to-hard-to-easy pattern. Further, our results indicate that the hardest instances are in a region where it is more likely that the corresponding bipartite graph will have a complete subgraph of specified dimensions, a positive answer. By contrast, instances with a negative answer are more likely to appear in a region where the computational cost is negligible. This behaviour is remarkably similar for problems of a number of different sizes.
2110.11240
Nathaniel Hoy
Nathaniel Hoy, Theodora Koulouri
A Systematic Review on the Detection of Fake News Articles
22 Pages, 16 Figures, Currently submitted to ACM TIST - Awaiting Peer-Review
null
null
null
cs.CL cs.AI cs.LG
http://creativecommons.org/licenses/by-nc-nd/4.0/
It has been argued that fake news and the spread of false information pose a threat to societies throughout the world, from influencing the results of elections to hindering the efforts to manage the COVID-19 pandemic. To combat this threat, a number of Natural Language Processing (NLP) approaches have been developed. These leverage a number of datasets, feature extraction/selection techniques and machine learning (ML) algorithms to detect fake news before it spreads. While these methods are well-documented, there is less evidence regarding their efficacy in this domain. By systematically reviewing the literature, this paper aims to delineate the approaches for fake news detection that are most performant, identify limitations with existing approaches, and suggest ways these can be mitigated. The analysis of the results indicates that Ensemble Methods using a combination of news content and socially-based features are currently the most effective. Finally, it is proposed that future research should focus on developing approaches that address generalisability issues (which, in part, arise from limitations with current datasets), explainability and bias.
[ { "created": "Mon, 18 Oct 2021 21:29:11 GMT", "version": "v1" } ]
2021-10-22
[ [ "Hoy", "Nathaniel", "" ], [ "Koulouri", "Theodora", "" ] ]
It has been argued that fake news and the spread of false information pose a threat to societies throughout the world, from influencing the results of elections to hindering the efforts to manage the COVID-19 pandemic. To combat this threat, a number of Natural Language Processing (NLP) approaches have been developed. These leverage a number of datasets, feature extraction/selection techniques and machine learning (ML) algorithms to detect fake news before it spreads. While these methods are well-documented, there is less evidence regarding their efficacy in this domain. By systematically reviewing the literature, this paper aims to delineate the approaches for fake news detection that are most performant, identify limitations with existing approaches, and suggest ways these can be mitigated. The analysis of the results indicates that Ensemble Methods using a combination of news content and socially-based features are currently the most effective. Finally, it is proposed that future research should focus on developing approaches that address generalisability issues (which, in part, arise from limitations with current datasets), explainability and bias.
1511.08952
Ndapa Nakashole
Ndapandula Nakashole
Bootstrapping Ternary Relation Extractors
6 pages
null
null
null
cs.CL cs.AI
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Binary relation extraction methods have been widely studied in recent years. However, few methods have been developed for higher n-ary relation extraction. One limiting factor is the effort required to generate training data. For binary relations, one only has to provide a few dozen pairs of entities per relation, as training data. For ternary relations (n=3), each training instance is a triplet of entities, placing a greater cognitive load on people. For example, many people know that Google acquired Youtube but not the dollar amount or the date of the acquisition and many people know that Hillary Clinton is married to Bill Clinton by not the location or date of their wedding. This makes higher n-nary training data generation a time consuming exercise in searching the Web. We present a resource for training ternary relation extractors. This was generated using a minimally supervised yet effective approach. We present statistics on the size and the quality of the dataset.
[ { "created": "Sun, 29 Nov 2015 00:49:13 GMT", "version": "v1" }, { "created": "Wed, 17 Jul 2019 02:52:34 GMT", "version": "v2" } ]
2019-07-18
[ [ "Nakashole", "Ndapandula", "" ] ]
Binary relation extraction methods have been widely studied in recent years. However, few methods have been developed for higher n-ary relation extraction. One limiting factor is the effort required to generate training data. For binary relations, one only has to provide a few dozen pairs of entities per relation, as training data. For ternary relations (n=3), each training instance is a triplet of entities, placing a greater cognitive load on people. For example, many people know that Google acquired Youtube but not the dollar amount or the date of the acquisition and many people know that Hillary Clinton is married to Bill Clinton by not the location or date of their wedding. This makes higher n-nary training data generation a time consuming exercise in searching the Web. We present a resource for training ternary relation extractors. This was generated using a minimally supervised yet effective approach. We present statistics on the size and the quality of the dataset.
2301.03728
Armen Aghajanyan
Armen Aghajanyan, Lili Yu, Alexis Conneau, Wei-Ning Hsu, Karen Hambardzumyan, Susan Zhang, Stephen Roller, Naman Goyal, Omer Levy, Luke Zettlemoyer
Scaling Laws for Generative Mixed-Modal Language Models
null
null
null
null
cs.CL cs.AI cs.LG
http://creativecommons.org/licenses/by/4.0/
Generative language models define distributions over sequences of tokens that can represent essentially any combination of data modalities (e.g., any permutation of image tokens from VQ-VAEs, speech tokens from HuBERT, BPE tokens for language or code, and so on). To better understand the scaling properties of such mixed-modal models, we conducted over 250 experiments using seven different modalities and model sizes ranging from 8 million to 30 billion, trained on 5-100 billion tokens. We report new mixed-modal scaling laws that unify the contributions of individual modalities and the interactions between them. Specifically, we explicitly model the optimal synergy and competition due to data and model size as an additive term to previous uni-modal scaling laws. We also find four empirical phenomena observed during the training, such as emergent coordinate-ascent style training that naturally alternates between modalities, guidelines for selecting critical hyper-parameters, and connections between mixed-modal competition and training stability. Finally, we test our scaling law by training a 30B speech-text model, which significantly outperforms the corresponding unimodal models. Overall, our research provides valuable insights into the design and training of mixed-modal generative models, an important new class of unified models that have unique distributional properties.
[ { "created": "Tue, 10 Jan 2023 00:20:06 GMT", "version": "v1" } ]
2023-01-11
[ [ "Aghajanyan", "Armen", "" ], [ "Yu", "Lili", "" ], [ "Conneau", "Alexis", "" ], [ "Hsu", "Wei-Ning", "" ], [ "Hambardzumyan", "Karen", "" ], [ "Zhang", "Susan", "" ], [ "Roller", "Stephen", "" ], [ "Goyal", "Naman", "" ], [ "Levy", "Omer", "" ], [ "Zettlemoyer", "Luke", "" ] ]
Generative language models define distributions over sequences of tokens that can represent essentially any combination of data modalities (e.g., any permutation of image tokens from VQ-VAEs, speech tokens from HuBERT, BPE tokens for language or code, and so on). To better understand the scaling properties of such mixed-modal models, we conducted over 250 experiments using seven different modalities and model sizes ranging from 8 million to 30 billion, trained on 5-100 billion tokens. We report new mixed-modal scaling laws that unify the contributions of individual modalities and the interactions between them. Specifically, we explicitly model the optimal synergy and competition due to data and model size as an additive term to previous uni-modal scaling laws. We also find four empirical phenomena observed during the training, such as emergent coordinate-ascent style training that naturally alternates between modalities, guidelines for selecting critical hyper-parameters, and connections between mixed-modal competition and training stability. Finally, we test our scaling law by training a 30B speech-text model, which significantly outperforms the corresponding unimodal models. Overall, our research provides valuable insights into the design and training of mixed-modal generative models, an important new class of unified models that have unique distributional properties.
1602.00430
Biao Sun
Biao Sun, Wenfeng Zhao, Xinshan Zhu
Compressed Sensing for Implantable Neural Recordings Using Co-sparse Analysis Model and Weighted $\ell_1$-Optimization
22 pages, 11 figures
null
null
null
cs.IT math.IT
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Reliable and energy-efficient wireless data transmission remains a major challenge in resource-constrained wireless neural recording tasks, where data compression is generally adopted to relax the burdens on the wireless data link. Recently, Compressed Sensing (CS) theory has successfully demonstrated its potential in neural recording application. The main limitation of CS, however, is that the neural signals have no good sparse representation with commonly used dictionaries and learning a reliable dictionary is often data dependent and computationally demanding. In this paper, a novel CS approach for implantable neural recording is proposed. The main contributions are: 1) The co-sparse analysis model is adopted to enforce co-sparsity of the neural signals, therefore overcoming the drawbacks of conventional synthesis model and enhancing the reconstruction performance. 2) A multi-fractional-order difference matrix is constructed as the analysis dictionary, thus avoiding the dictionary learning procedure and reducing the need for previously acquired data and computational resources. 3) By exploiting the statistical priors of the analysis coefficients, a weighted analysis $\ell_1$-minimization (WALM) algorithm is proposed to reconstruct the neural signals. Experimental results on Leicester neural signal database reveal that the proposed approach outperforms the state-of-the-art CS-based methods. On the challenging high compression ratio task, the proposed approach still achieves high reconstruction performance and spike classification accuracy.
[ { "created": "Mon, 1 Feb 2016 08:57:14 GMT", "version": "v1" } ]
2016-02-02
[ [ "Sun", "Biao", "" ], [ "Zhao", "Wenfeng", "" ], [ "Zhu", "Xinshan", "" ] ]
Reliable and energy-efficient wireless data transmission remains a major challenge in resource-constrained wireless neural recording tasks, where data compression is generally adopted to relax the burdens on the wireless data link. Recently, Compressed Sensing (CS) theory has successfully demonstrated its potential in neural recording application. The main limitation of CS, however, is that the neural signals have no good sparse representation with commonly used dictionaries and learning a reliable dictionary is often data dependent and computationally demanding. In this paper, a novel CS approach for implantable neural recording is proposed. The main contributions are: 1) The co-sparse analysis model is adopted to enforce co-sparsity of the neural signals, therefore overcoming the drawbacks of conventional synthesis model and enhancing the reconstruction performance. 2) A multi-fractional-order difference matrix is constructed as the analysis dictionary, thus avoiding the dictionary learning procedure and reducing the need for previously acquired data and computational resources. 3) By exploiting the statistical priors of the analysis coefficients, a weighted analysis $\ell_1$-minimization (WALM) algorithm is proposed to reconstruct the neural signals. Experimental results on Leicester neural signal database reveal that the proposed approach outperforms the state-of-the-art CS-based methods. On the challenging high compression ratio task, the proposed approach still achieves high reconstruction performance and spike classification accuracy.
2208.11188
Vincent Cicirello
Vincent A. Cicirello
On Fitness Landscape Analysis of Permutation Problems: From Distance Metrics to Mutation Operator Selection
Accepted by: Mobile Networks and Applications, Special Issue on Foundational Studies in Bio-Inspired Technologies
Mobile Networks and Applications, 28(2): 507-517, April 2023
10.1007/s11036-022-02060-z
null
cs.NE cs.AI cs.DM
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In this paper, we explore the theory and expand upon the practice of fitness landscape analysis for optimization problems over the space of permutations. Many of the computational and analytical tools for fitness landscape analysis, such as fitness distance correlation, require identifying a distance metric for measuring the similarity of different solutions to the problem. We begin with a survey of the available distance metrics for permutations, and then use principal component analysis to classify these metrics. The result of this analysis aligns with existing classifications of permutation problem types produced through less formal means, including the A-permutation, R-permutation, and P-permutation types, which classifies problems by whether absolute position of permutation elements, relative positions of elements, or general precedence of pairs of elements, is the dominant influence over solution fitness. Additionally, the formal analysis identifies subtypes within these problem categories. We see that the classification can assist in identifying appropriate metrics based on optimization problem feature for use in fitness landscape analysis. Using optimization problems of each class, we also demonstrate how the classification scheme can subsequently inform the choice of mutation operator within an evolutionary algorithm. From this, we present a classification of a variety of mutation operators as a counterpart to that of the metrics. Our implementations of the permutation metrics, permutation mutation operators, and associated evolutionary algorithm, are available in a pair of open source Java libraries. All of the code necessary to recreate our analysis and experimental results are also available as open source.
[ { "created": "Tue, 23 Aug 2022 20:46:49 GMT", "version": "v1" } ]
2023-11-10
[ [ "Cicirello", "Vincent A.", "" ] ]
In this paper, we explore the theory and expand upon the practice of fitness landscape analysis for optimization problems over the space of permutations. Many of the computational and analytical tools for fitness landscape analysis, such as fitness distance correlation, require identifying a distance metric for measuring the similarity of different solutions to the problem. We begin with a survey of the available distance metrics for permutations, and then use principal component analysis to classify these metrics. The result of this analysis aligns with existing classifications of permutation problem types produced through less formal means, including the A-permutation, R-permutation, and P-permutation types, which classifies problems by whether absolute position of permutation elements, relative positions of elements, or general precedence of pairs of elements, is the dominant influence over solution fitness. Additionally, the formal analysis identifies subtypes within these problem categories. We see that the classification can assist in identifying appropriate metrics based on optimization problem feature for use in fitness landscape analysis. Using optimization problems of each class, we also demonstrate how the classification scheme can subsequently inform the choice of mutation operator within an evolutionary algorithm. From this, we present a classification of a variety of mutation operators as a counterpart to that of the metrics. Our implementations of the permutation metrics, permutation mutation operators, and associated evolutionary algorithm, are available in a pair of open source Java libraries. All of the code necessary to recreate our analysis and experimental results are also available as open source.
1804.05258
Ross M. McConnell
Pavol Hell and Jing Huang and Ross M. McConnell and Arash Rafiey
Interval-Like Graphs and Digraphs
null
null
null
null
cs.DM
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We unify several seemingly different graph and digraph classes under one umbrella. These classes are all broadly speaking different generalizations of interval graphs, and include, in addition to interval graphs, also adjusted interval digraphs, threshold graphs, complements of threshold tolerance graphs (known as `co-TT' graphs), bipartite interval containment graphs, bipartite co-circular arc graphs, and two-directional orthogonal ray graphs. (The last three classes coincide, but have been investigated in different contexts.) This common view is made possible by introducing loops. We also show that all the above classes are united by a common ordering characterization, the existence of a min ordering. We propose a common generalization of all these graph and digraph classes, namely signed-interval digraphs, and show that they are precisely the digraphs that are characterized by the existence of a min ordering. We also offer an alternative geometric characterization of these digraphs. For most of the above example graph and digraph classes, we show that they are exactly those signed-interval digraphs that satisfy a suitable natural restriction on the digraph, like having all loops, or having a symmetric edge-set, or being bipartite. (For instance co-TT graphs are precisely those signed-interval digraphs that have each edge symmetric.) We also offer some discussion of recognition algorithms and characterizations, saving the details for future papers.
[ { "created": "Sat, 14 Apr 2018 18:19:33 GMT", "version": "v1" }, { "created": "Tue, 26 Jun 2018 21:41:10 GMT", "version": "v2" } ]
2018-06-28
[ [ "Hell", "Pavol", "" ], [ "Huang", "Jing", "" ], [ "McConnell", "Ross M.", "" ], [ "Rafiey", "Arash", "" ] ]
We unify several seemingly different graph and digraph classes under one umbrella. These classes are all broadly speaking different generalizations of interval graphs, and include, in addition to interval graphs, also adjusted interval digraphs, threshold graphs, complements of threshold tolerance graphs (known as `co-TT' graphs), bipartite interval containment graphs, bipartite co-circular arc graphs, and two-directional orthogonal ray graphs. (The last three classes coincide, but have been investigated in different contexts.) This common view is made possible by introducing loops. We also show that all the above classes are united by a common ordering characterization, the existence of a min ordering. We propose a common generalization of all these graph and digraph classes, namely signed-interval digraphs, and show that they are precisely the digraphs that are characterized by the existence of a min ordering. We also offer an alternative geometric characterization of these digraphs. For most of the above example graph and digraph classes, we show that they are exactly those signed-interval digraphs that satisfy a suitable natural restriction on the digraph, like having all loops, or having a symmetric edge-set, or being bipartite. (For instance co-TT graphs are precisely those signed-interval digraphs that have each edge symmetric.) We also offer some discussion of recognition algorithms and characterizations, saving the details for future papers.
2210.12181
Weizi Li
Weizi Li
Urban Socio-Technical Systems: An Autonomy and Mobility Perspective
null
null
null
null
cs.CY
http://creativecommons.org/licenses/by/4.0/
The future of the human race is urban. The world's population is projected to grow an additional 2.5 billion by 2050, with all expected to live in urban areas. This will increase the percentage of urban population from 55% today to 70% within three decades and further strengthen the role of cities as the hub for information, transportation, and overall socio-economic development. Unlike any other time in human history, the increasing levels of autonomy and machine intelligence are transforming cities to be no longer just human agglomerations but a fusion of humans, machines, and algorithms making collective decisions, thus complex socio-technical systems. This manuscript summarizes and discusses my efforts from the urban autonomy and mobility perspective to develop the urban socio-technical system.
[ { "created": "Fri, 21 Oct 2022 18:15:41 GMT", "version": "v1" } ]
2022-10-25
[ [ "Li", "Weizi", "" ] ]
The future of the human race is urban. The world's population is projected to grow an additional 2.5 billion by 2050, with all expected to live in urban areas. This will increase the percentage of urban population from 55% today to 70% within three decades and further strengthen the role of cities as the hub for information, transportation, and overall socio-economic development. Unlike any other time in human history, the increasing levels of autonomy and machine intelligence are transforming cities to be no longer just human agglomerations but a fusion of humans, machines, and algorithms making collective decisions, thus complex socio-technical systems. This manuscript summarizes and discusses my efforts from the urban autonomy and mobility perspective to develop the urban socio-technical system.
2006.09873
Debanjan Ghosh
Debanjan Ghosh, Beata Beigman Klebanov, Yi Song
An Exploratory Study of Argumentative Writing by Young Students: A Transformer-based Approach
15th Workshop on Innovative Use of NLP for Building Educational Applications, ACL 2020
null
null
null
cs.CL cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We present a computational exploration of argument critique writing by young students. Middle school students were asked to criticize an argument presented in the prompt, focusing on identifying and explaining the reasoning flaws. This task resembles an established college-level argument critique task. Lexical and discourse features that utilize detailed domain knowledge to identify critiques exist for the college task but do not perform well on the young students data. Instead, transformer-based architecture (e.g., BERT) fine-tuned on a large corpus of critique essays from the college task performs much better (over 20% improvement in F1 score). Analysis of the performance of various configurations of the system suggests that while children's writing does not exhibit the standard discourse structure of an argumentative essay, it does share basic local sequential structures with the more mature writers.
[ { "created": "Wed, 17 Jun 2020 13:55:31 GMT", "version": "v1" } ]
2020-06-18
[ [ "Ghosh", "Debanjan", "" ], [ "Klebanov", "Beata Beigman", "" ], [ "Song", "Yi", "" ] ]
We present a computational exploration of argument critique writing by young students. Middle school students were asked to criticize an argument presented in the prompt, focusing on identifying and explaining the reasoning flaws. This task resembles an established college-level argument critique task. Lexical and discourse features that utilize detailed domain knowledge to identify critiques exist for the college task but do not perform well on the young students data. Instead, transformer-based architecture (e.g., BERT) fine-tuned on a large corpus of critique essays from the college task performs much better (over 20% improvement in F1 score). Analysis of the performance of various configurations of the system suggests that while children's writing does not exhibit the standard discourse structure of an argumentative essay, it does share basic local sequential structures with the more mature writers.
2204.01690
Rahim Taheri
Meysam Ghahramani, Rahim Taheri, Mohammad Shojafar, Reza Javidan, Shaohua Wan
Deep Image: A precious image based deep learning method for online malware detection in IoT Environment
10 pages, 17 figures, SUBMITTED TO IEEE INTERNET OF THINGS JOURNAL, MARCH 2022
null
null
null
cs.CR cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
The volume of malware and the number of attacks in IoT devices are rising everyday, which encourages security professionals to continually enhance their malware analysis tools. Researchers in the field of cyber security have extensively explored the usage of sophisticated analytics and the efficiency of malware detection. With the introduction of new malware kinds and attack routes, security experts confront considerable challenges in developing efficient malware detection and analysis solutions. In this paper, a different view of malware analysis is considered and the risk level of each sample feature is computed, and based on that the risk level of that sample is calculated. In this way, a criterion is introduced that is used together with accuracy and FPR criteria for malware analysis in IoT environment. In this paper, three malware detection methods based on visualization techniques called the clustering approach, the probabilistic approach, and the deep learning approach are proposed. Then, in addition to the usual machine learning criteria namely accuracy and FPR, a proposed criterion based on the risk of samples has also been used for comparison, with the results showing that the deep learning approach performed better in detecting malware
[ { "created": "Mon, 4 Apr 2022 17:56:55 GMT", "version": "v1" } ]
2022-04-05
[ [ "Ghahramani", "Meysam", "" ], [ "Taheri", "Rahim", "" ], [ "Shojafar", "Mohammad", "" ], [ "Javidan", "Reza", "" ], [ "Wan", "Shaohua", "" ] ]
The volume of malware and the number of attacks in IoT devices are rising everyday, which encourages security professionals to continually enhance their malware analysis tools. Researchers in the field of cyber security have extensively explored the usage of sophisticated analytics and the efficiency of malware detection. With the introduction of new malware kinds and attack routes, security experts confront considerable challenges in developing efficient malware detection and analysis solutions. In this paper, a different view of malware analysis is considered and the risk level of each sample feature is computed, and based on that the risk level of that sample is calculated. In this way, a criterion is introduced that is used together with accuracy and FPR criteria for malware analysis in IoT environment. In this paper, three malware detection methods based on visualization techniques called the clustering approach, the probabilistic approach, and the deep learning approach are proposed. Then, in addition to the usual machine learning criteria namely accuracy and FPR, a proposed criterion based on the risk of samples has also been used for comparison, with the results showing that the deep learning approach performed better in detecting malware
1701.05973
Amirhossein Reisizadeh
Amirhossein Reisizadeh, Saurav Prakash, Ramtin Pedarsani, Amir Salman Avestimehr
Coded Computation over Heterogeneous Clusters
This work is published in IEEE Transaction on Information Theory (2019). A preliminary version of this work was published in IEEE International Symposium on Information Theory (ISIT) 2017
null
null
null
cs.DC cs.IT math.IT
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In large-scale distributed computing clusters, such as Amazon EC2, there are several types of "system noise" that can result in major degradation of performance: bottlenecks due to limited communication bandwidth, latency due to straggler nodes, etc. On the other hand, these systems enjoy abundance of redundancy - a vast number of computing nodes and large storage capacity. There have been recent results that demonstrate the impact of coding for efficient utilization of computation and storage redundancy to alleviate the effect of stragglers and communication bottlenecks in homogeneous clusters. In this paper, we focus on general heterogeneous distributed computing clusters consisting of a variety of computing machines with different capabilities. We propose a coding framework for speeding up distributed computing in heterogeneous clusters by trading redundancy for reducing the latency of computation. In particular, we propose Heterogeneous Coded Matrix Multiplication (HCMM) algorithm for performing distributed matrix multiplication over heterogeneous clusters that is provably asymptotically optimal for a broad class of processing time distributions. Moreover, we show that HCMM is unboundedly faster than any uncoded scheme. To demonstrate practicality of HCMM, we carry out experiments over Amazon EC2 clusters where HCMM is found to be up to $61\%$, $46\%$ and $36\%$ respectively faster than three benchmark load allocation schemes - Uniform Uncoded, Load-balanced Uncoded, and Uniform Coded. Additionally, we provide a generalization to the problem of optimal load allocation in heterogeneous settings, where we take into account the monetary costs associated with the clusters. We argue that HCMM is asymptotically optimal for budget-constrained scenarios as well, and we develop a heuristic algorithm for (HCMM) load allocation for budget-limited computation tasks.
[ { "created": "Sat, 21 Jan 2017 03:11:47 GMT", "version": "v1" }, { "created": "Fri, 16 Jun 2017 23:06:31 GMT", "version": "v2" }, { "created": "Mon, 2 Apr 2018 21:43:11 GMT", "version": "v3" }, { "created": "Tue, 9 Oct 2018 06:03:03 GMT", "version": "v4" }, { "created": "Wed, 19 Jun 2019 23:58:44 GMT", "version": "v5" } ]
2019-06-21
[ [ "Reisizadeh", "Amirhossein", "" ], [ "Prakash", "Saurav", "" ], [ "Pedarsani", "Ramtin", "" ], [ "Avestimehr", "Amir Salman", "" ] ]
In large-scale distributed computing clusters, such as Amazon EC2, there are several types of "system noise" that can result in major degradation of performance: bottlenecks due to limited communication bandwidth, latency due to straggler nodes, etc. On the other hand, these systems enjoy abundance of redundancy - a vast number of computing nodes and large storage capacity. There have been recent results that demonstrate the impact of coding for efficient utilization of computation and storage redundancy to alleviate the effect of stragglers and communication bottlenecks in homogeneous clusters. In this paper, we focus on general heterogeneous distributed computing clusters consisting of a variety of computing machines with different capabilities. We propose a coding framework for speeding up distributed computing in heterogeneous clusters by trading redundancy for reducing the latency of computation. In particular, we propose Heterogeneous Coded Matrix Multiplication (HCMM) algorithm for performing distributed matrix multiplication over heterogeneous clusters that is provably asymptotically optimal for a broad class of processing time distributions. Moreover, we show that HCMM is unboundedly faster than any uncoded scheme. To demonstrate practicality of HCMM, we carry out experiments over Amazon EC2 clusters where HCMM is found to be up to $61\%$, $46\%$ and $36\%$ respectively faster than three benchmark load allocation schemes - Uniform Uncoded, Load-balanced Uncoded, and Uniform Coded. Additionally, we provide a generalization to the problem of optimal load allocation in heterogeneous settings, where we take into account the monetary costs associated with the clusters. We argue that HCMM is asymptotically optimal for budget-constrained scenarios as well, and we develop a heuristic algorithm for (HCMM) load allocation for budget-limited computation tasks.
1904.08562
Jonathan Sterling
Jonathan Sterling, Carlo Angiuli, Daniel Gratzer
Cubical Syntax for Reflection-Free Extensional Equality
Extended version; International Conference on Formal Structures for Computation and Deduction (FSCD), 2019
null
10.4230/LIPIcs.FSCD.2019.31
null
cs.LO math.LO
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We contribute XTT, a cubical reconstruction of Observational Type Theory which extends Martin-L\"of's intensional type theory with a dependent equality type that enjoys function extensionality and a judgmental version of the unicity of identity types principle (UIP): any two elements of the same equality type are judgmentally equal. Moreover, we conjecture that the typing relation can be decided in a practical way. In this paper, we establish an algebraic canonicity theorem using a novel cubical extension (independently proposed by Awodey) of the logical families or categorical gluing argument inspired by Coquand and Shulman: every closed element of boolean type is derivably equal to either 'true' or 'false'.
[ { "created": "Thu, 18 Apr 2019 01:52:13 GMT", "version": "v1" }, { "created": "Thu, 6 Jun 2019 19:43:41 GMT", "version": "v2" } ]
2021-04-20
[ [ "Sterling", "Jonathan", "" ], [ "Angiuli", "Carlo", "" ], [ "Gratzer", "Daniel", "" ] ]
We contribute XTT, a cubical reconstruction of Observational Type Theory which extends Martin-L\"of's intensional type theory with a dependent equality type that enjoys function extensionality and a judgmental version of the unicity of identity types principle (UIP): any two elements of the same equality type are judgmentally equal. Moreover, we conjecture that the typing relation can be decided in a practical way. In this paper, we establish an algebraic canonicity theorem using a novel cubical extension (independently proposed by Awodey) of the logical families or categorical gluing argument inspired by Coquand and Shulman: every closed element of boolean type is derivably equal to either 'true' or 'false'.
2403.15786
Kaiwen Wang
Kaiwen Wang, Yinzhe Shen, Martin Lauer
Adversarial Defense Teacher for Cross-Domain Object Detection under Poor Visibility Conditions
null
null
null
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Existing object detectors encounter challenges in handling domain shifts between training and real-world data, particularly under poor visibility conditions like fog and night. Cutting-edge cross-domain object detection methods use teacher-student frameworks and compel teacher and student models to produce consistent predictions under weak and strong augmentations, respectively. In this paper, we reveal that manually crafted augmentations are insufficient for optimal teaching and present a simple yet effective framework named Adversarial Defense Teacher (ADT), leveraging adversarial defense to enhance teaching quality. Specifically, we employ adversarial attacks, encouraging the model to generalize on subtly perturbed inputs that effectively deceive the model. To address small objects under poor visibility conditions, we propose a Zoom-in Zoom-out strategy, which zooms-in images for better pseudo-labels and zooms-out images and pseudo-labels to learn refined features. Our results demonstrate that ADT achieves superior performance, reaching 54.5% mAP on Foggy Cityscapes, surpassing the previous state-of-the-art by 2.6% mAP.
[ { "created": "Sat, 23 Mar 2024 10:16:05 GMT", "version": "v1" } ]
2024-03-26
[ [ "Wang", "Kaiwen", "" ], [ "Shen", "Yinzhe", "" ], [ "Lauer", "Martin", "" ] ]
Existing object detectors encounter challenges in handling domain shifts between training and real-world data, particularly under poor visibility conditions like fog and night. Cutting-edge cross-domain object detection methods use teacher-student frameworks and compel teacher and student models to produce consistent predictions under weak and strong augmentations, respectively. In this paper, we reveal that manually crafted augmentations are insufficient for optimal teaching and present a simple yet effective framework named Adversarial Defense Teacher (ADT), leveraging adversarial defense to enhance teaching quality. Specifically, we employ adversarial attacks, encouraging the model to generalize on subtly perturbed inputs that effectively deceive the model. To address small objects under poor visibility conditions, we propose a Zoom-in Zoom-out strategy, which zooms-in images for better pseudo-labels and zooms-out images and pseudo-labels to learn refined features. Our results demonstrate that ADT achieves superior performance, reaching 54.5% mAP on Foggy Cityscapes, surpassing the previous state-of-the-art by 2.6% mAP.
1903.07290
Wonseok Ha
Wonseok Ha and Juhoon Back
An Output Feedback Stabilizer for MIMO Nonlinear Systems with Uncertain Input Gain: Nonlinear Nominal Model
8 pages, 6 figures
null
null
null
cs.SY
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
This paper deals with the output feedback stabilization problem of nonlinear multi-input multi-output systems having an uncertain input gain matrix. It is assumed that the system has a well-defined vector relative degree and that the zero dynamics is input-to-state stable. Based on the assumption that there exists a state feedback controller which globally asymptotically stabilizes the origin of the nominal closed-loop system, we present an output feedback stabilizer which recovers the stability of the nominal closed-loop system in the semi-global practical sense. Compared to previous results, we allow that the nominal system can have a nonlinear input gain matrix that is a function of state and this is done by modifying the structure of the disturbance observer-based robust output feedback controller. It is expected that the proposed controller can be well applied to the case when the system's nonlinearity is to be exploited rather than canceled.
[ { "created": "Mon, 18 Mar 2019 07:56:01 GMT", "version": "v1" } ]
2019-03-19
[ [ "Ha", "Wonseok", "" ], [ "Back", "Juhoon", "" ] ]
This paper deals with the output feedback stabilization problem of nonlinear multi-input multi-output systems having an uncertain input gain matrix. It is assumed that the system has a well-defined vector relative degree and that the zero dynamics is input-to-state stable. Based on the assumption that there exists a state feedback controller which globally asymptotically stabilizes the origin of the nominal closed-loop system, we present an output feedback stabilizer which recovers the stability of the nominal closed-loop system in the semi-global practical sense. Compared to previous results, we allow that the nominal system can have a nonlinear input gain matrix that is a function of state and this is done by modifying the structure of the disturbance observer-based robust output feedback controller. It is expected that the proposed controller can be well applied to the case when the system's nonlinearity is to be exploited rather than canceled.
2404.07963
Songlin Xu
Songlin Xu, Xinyu Zhang, Lianhui Qin
EduAgent: Generative Student Agents in Learning
null
null
null
null
cs.CY cs.AI cs.CL cs.HC cs.LG
http://creativecommons.org/licenses/by-nc-nd/4.0/
Student simulation in online education is important to address dynamic learning behaviors of students with diverse backgrounds. Existing simulation models based on deep learning usually need massive training data, lacking prior knowledge in educational contexts. Large language models (LLMs) may contain such prior knowledge since they are pre-trained from a large corpus. However, because student behaviors are dynamic and multifaceted with individual differences, directly prompting LLMs is not robust nor accurate enough to capture fine-grained interactions among diverse student personas, learning behaviors, and learning outcomes. This work tackles this problem by presenting a newly annotated fine-grained large-scale dataset and proposing EduAgent, a novel generative agent framework incorporating cognitive prior knowledge (i.e., theoretical findings revealed in cognitive science) to guide LLMs to first reason correlations among various behaviors and then make simulations. Our two experiments show that EduAgent could not only mimic and predict learning behaviors of real students but also generate realistic learning behaviors of virtual students without real data.
[ { "created": "Sat, 23 Mar 2024 18:19:17 GMT", "version": "v1" } ]
2024-04-12
[ [ "Xu", "Songlin", "" ], [ "Zhang", "Xinyu", "" ], [ "Qin", "Lianhui", "" ] ]
Student simulation in online education is important to address dynamic learning behaviors of students with diverse backgrounds. Existing simulation models based on deep learning usually need massive training data, lacking prior knowledge in educational contexts. Large language models (LLMs) may contain such prior knowledge since they are pre-trained from a large corpus. However, because student behaviors are dynamic and multifaceted with individual differences, directly prompting LLMs is not robust nor accurate enough to capture fine-grained interactions among diverse student personas, learning behaviors, and learning outcomes. This work tackles this problem by presenting a newly annotated fine-grained large-scale dataset and proposing EduAgent, a novel generative agent framework incorporating cognitive prior knowledge (i.e., theoretical findings revealed in cognitive science) to guide LLMs to first reason correlations among various behaviors and then make simulations. Our two experiments show that EduAgent could not only mimic and predict learning behaviors of real students but also generate realistic learning behaviors of virtual students without real data.
2209.07386
Mete \c{S}eref Ahunbay
Mete \c{S}eref Ahunbay and Martin Bichler and Johannes Kn\"orr
Pricing Optimal Outcomes in Coupled and Non-Convex Markets: Theory and Applications to Electricity Markets
44 pages, 2 figures
null
null
null
cs.GT
http://creativecommons.org/licenses/by/4.0/
According to the fundamental theorems of welfare economics, any competitive equilibrium is Pareto efficient. Unfortunately, competitive equilibrium prices only exist under strong assumptions such as perfectly divisible goods and convex preferences. In many real-world markets, participants have non-convex preferences and the allocation problem needs to consider complex constraints. Electricity markets are a prime example, but similar problems appear in many real-world markets, which has led to a growing literature in market design. Power markets use heuristic pricing rules based on the dual of a relaxed allocation problem today. With increasing levels of renewables, these rules have come under scrutiny as they lead to high out-of-market side-payments to some participants and to inadequate congestion signals. We show that existing pricing heuristics optimize specific design goals that can be conflicting. The trade-offs can be substantial, and we establish that the design of pricing rules is fundamentally a multi-objective optimization problem addressing different incentives. In addition to traditional multi-objective optimization techniques using weighing of individual objectives, we introduce a novel parameter-free pricing rule that minimizes incentives for market participants to deviate locally. Our theoretical and experimental findings show how the new pricing rule capitalizes on the upsides of existing pricing rules under scrutiny today. It leads to prices that incur low make-whole payments while providing adequate congestion signals and low lost opportunity costs. Our suggested pricing rule does not require weighing of objectives, it is computationally scalable, and balances trade-offs in a principled manner, addressing an important policy issue in electricity markets.
[ { "created": "Thu, 15 Sep 2022 15:51:46 GMT", "version": "v1" }, { "created": "Tue, 23 May 2023 13:43:12 GMT", "version": "v2" } ]
2023-05-24
[ [ "Ahunbay", "Mete Şeref", "" ], [ "Bichler", "Martin", "" ], [ "Knörr", "Johannes", "" ] ]
According to the fundamental theorems of welfare economics, any competitive equilibrium is Pareto efficient. Unfortunately, competitive equilibrium prices only exist under strong assumptions such as perfectly divisible goods and convex preferences. In many real-world markets, participants have non-convex preferences and the allocation problem needs to consider complex constraints. Electricity markets are a prime example, but similar problems appear in many real-world markets, which has led to a growing literature in market design. Power markets use heuristic pricing rules based on the dual of a relaxed allocation problem today. With increasing levels of renewables, these rules have come under scrutiny as they lead to high out-of-market side-payments to some participants and to inadequate congestion signals. We show that existing pricing heuristics optimize specific design goals that can be conflicting. The trade-offs can be substantial, and we establish that the design of pricing rules is fundamentally a multi-objective optimization problem addressing different incentives. In addition to traditional multi-objective optimization techniques using weighing of individual objectives, we introduce a novel parameter-free pricing rule that minimizes incentives for market participants to deviate locally. Our theoretical and experimental findings show how the new pricing rule capitalizes on the upsides of existing pricing rules under scrutiny today. It leads to prices that incur low make-whole payments while providing adequate congestion signals and low lost opportunity costs. Our suggested pricing rule does not require weighing of objectives, it is computationally scalable, and balances trade-offs in a principled manner, addressing an important policy issue in electricity markets.
2306.14142
Eugene T.Y. Ang
Eugene Ang, Prasanta Bhattacharya, Andrew Lim
Estimating Policy Effects in a Social Network with Independent Set Sampling
null
null
null
null
cs.SI stat.ME
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Evaluating the impact of policy interventions on respondents who are embedded in a social network is often challenging due to the presence of network interference within the treatment groups, as well as between treatment and non-treatment groups throughout the network. In this paper, we propose a modeling strategy that combines existing work on stochastic actor-oriented models (SAOM) with a novel network sampling method based on the identification of independent sets. By assigning respondents from an independent set to the treatment, we are able to block any spillover of the treatment and network influence, thereby allowing us to isolate the direct effect of the treatment from the indirect network-induced effects, in the immediate term. As a result, our method allows for the estimation of both the \textit{direct} as well as the \textit{net effect} of a chosen policy intervention, in the presence of network effects in the population. We perform a comparative simulation analysis to show that our proposed sampling technique leads to distinct direct and net effects of the policy, as well as significant network effects driven by policy-linked homophily. This study highlights the importance of network sampling techniques in improving policy evaluation studies and has the potential to help researchers and policymakers with better planning, designing, and anticipating policy responses in a networked society.
[ { "created": "Sun, 25 Jun 2023 06:30:58 GMT", "version": "v1" }, { "created": "Fri, 1 Sep 2023 12:30:40 GMT", "version": "v2" }, { "created": "Mon, 26 Feb 2024 03:58:58 GMT", "version": "v3" } ]
2024-02-27
[ [ "Ang", "Eugene", "" ], [ "Bhattacharya", "Prasanta", "" ], [ "Lim", "Andrew", "" ] ]
Evaluating the impact of policy interventions on respondents who are embedded in a social network is often challenging due to the presence of network interference within the treatment groups, as well as between treatment and non-treatment groups throughout the network. In this paper, we propose a modeling strategy that combines existing work on stochastic actor-oriented models (SAOM) with a novel network sampling method based on the identification of independent sets. By assigning respondents from an independent set to the treatment, we are able to block any spillover of the treatment and network influence, thereby allowing us to isolate the direct effect of the treatment from the indirect network-induced effects, in the immediate term. As a result, our method allows for the estimation of both the \textit{direct} as well as the \textit{net effect} of a chosen policy intervention, in the presence of network effects in the population. We perform a comparative simulation analysis to show that our proposed sampling technique leads to distinct direct and net effects of the policy, as well as significant network effects driven by policy-linked homophily. This study highlights the importance of network sampling techniques in improving policy evaluation studies and has the potential to help researchers and policymakers with better planning, designing, and anticipating policy responses in a networked society.
2312.15964
Hyun Kang
Hyun Kang, Dohae Lee, Myungjin Shin, In-Kwon Lee
Semantic Guidance Tuning for Text-To-Image Diffusion Models
Rework is being done
null
null
null
cs.CV cs.AI
http://creativecommons.org/licenses/by/4.0/
Recent advancements in Text-to-Image (T2I) diffusion models have demonstrated impressive success in generating high-quality images with zero-shot generalization capabilities. Yet, current models struggle to closely adhere to prompt semantics, often misrepresenting or overlooking specific attributes. To address this, we propose a simple, training-free approach that modulates the guidance direction of diffusion models during inference. We first decompose the prompt semantics into a set of concepts, and monitor the guidance trajectory in relation to each concept. Our key observation is that deviations in model's adherence to prompt semantics are highly correlated with divergence of the guidance from one or more of these concepts. Based on this observation, we devise a technique to steer the guidance direction towards any concept from which the model diverges. Extensive experimentation validates that our method improves the semantic alignment of images generated by diffusion models in response to prompts. Project page is available at: https://korguy.github.io/
[ { "created": "Tue, 26 Dec 2023 09:02:17 GMT", "version": "v1" }, { "created": "Tue, 30 Jan 2024 04:55:29 GMT", "version": "v2" } ]
2024-01-31
[ [ "Kang", "Hyun", "" ], [ "Lee", "Dohae", "" ], [ "Shin", "Myungjin", "" ], [ "Lee", "In-Kwon", "" ] ]
Recent advancements in Text-to-Image (T2I) diffusion models have demonstrated impressive success in generating high-quality images with zero-shot generalization capabilities. Yet, current models struggle to closely adhere to prompt semantics, often misrepresenting or overlooking specific attributes. To address this, we propose a simple, training-free approach that modulates the guidance direction of diffusion models during inference. We first decompose the prompt semantics into a set of concepts, and monitor the guidance trajectory in relation to each concept. Our key observation is that deviations in model's adherence to prompt semantics are highly correlated with divergence of the guidance from one or more of these concepts. Based on this observation, we devise a technique to steer the guidance direction towards any concept from which the model diverges. Extensive experimentation validates that our method improves the semantic alignment of images generated by diffusion models in response to prompts. Project page is available at: https://korguy.github.io/
0710.4905
Oliver Kosut
Oliver Kosut, Lang Tong
Distributed Source Coding in the Presence of Byzantine Sensors
14 pages. Submitted to IEEE Trans. on IT, Information-theoretic Security special issue, February 2007. Revised October 2007
null
10.1109/TIT.2008.921867
null
cs.IT math.IT
null
The distributed source coding problem is considered when the sensors, or encoders, are under Byzantine attack; that is, an unknown group of sensors have been reprogrammed by a malicious intruder to undermine the reconstruction at the fusion center. Three different forms of the problem are considered. The first is a variable-rate setup, in which the decoder adaptively chooses the rates at which the sensors transmit. An explicit characterization of the variable-rate achievable sum rates is given for any number of sensors and any groups of traitors. The converse is proved constructively by letting the traitors simulate a fake distribution and report the generated values as the true ones. This fake distribution is chosen so that the decoder cannot determine which sensors are traitors while maximizing the required rate to decode every value. Achievability is proved using a scheme in which the decoder receives small packets of information from a sensor until its message can be decoded, before moving on to the next sensor. The sensors use randomization to choose from a set of coding functions, which makes it probabilistically impossible for the traitors to cause the decoder to make an error. Two forms of the fixed-rate problem are considered, one with deterministic coding and one with randomized coding. The achievable rate regions are given for both these problems, and it is shown that lower rates can be achieved with randomized coding.
[ { "created": "Thu, 25 Oct 2007 16:22:59 GMT", "version": "v1" } ]
2016-11-15
[ [ "Kosut", "Oliver", "" ], [ "Tong", "Lang", "" ] ]
The distributed source coding problem is considered when the sensors, or encoders, are under Byzantine attack; that is, an unknown group of sensors have been reprogrammed by a malicious intruder to undermine the reconstruction at the fusion center. Three different forms of the problem are considered. The first is a variable-rate setup, in which the decoder adaptively chooses the rates at which the sensors transmit. An explicit characterization of the variable-rate achievable sum rates is given for any number of sensors and any groups of traitors. The converse is proved constructively by letting the traitors simulate a fake distribution and report the generated values as the true ones. This fake distribution is chosen so that the decoder cannot determine which sensors are traitors while maximizing the required rate to decode every value. Achievability is proved using a scheme in which the decoder receives small packets of information from a sensor until its message can be decoded, before moving on to the next sensor. The sensors use randomization to choose from a set of coding functions, which makes it probabilistically impossible for the traitors to cause the decoder to make an error. Two forms of the fixed-rate problem are considered, one with deterministic coding and one with randomized coding. The achievable rate regions are given for both these problems, and it is shown that lower rates can be achieved with randomized coding.
0906.1182
Paolo Mancarella
P. Mancarella, G. Terreni, F. Sadri, F. Toni, U. Endriss
The CIFF Proof Procedure for Abductive Logic Programming with Constraints: Theory, Implementation and Experiments
null
null
null
null
cs.AI cs.LO
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We present the CIFF proof procedure for abductive logic programming with constraints, and we prove its correctness. CIFF is an extension of the IFF proof procedure for abductive logic programming, relaxing the original restrictions over variable quantification (allowedness conditions) and incorporating a constraint solver to deal with numerical constraints as in constraint logic programming. Finally, we describe the CIFF system, comparing it with state of the art abductive systems and answer set solvers and showing how to use it to program some applications. (To appear in Theory and Practice of Logic Programming - TPLP).
[ { "created": "Fri, 5 Jun 2009 16:13:23 GMT", "version": "v1" } ]
2009-06-08
[ [ "Mancarella", "P.", "" ], [ "Terreni", "G.", "" ], [ "Sadri", "F.", "" ], [ "Toni", "F.", "" ], [ "Endriss", "U.", "" ] ]
We present the CIFF proof procedure for abductive logic programming with constraints, and we prove its correctness. CIFF is an extension of the IFF proof procedure for abductive logic programming, relaxing the original restrictions over variable quantification (allowedness conditions) and incorporating a constraint solver to deal with numerical constraints as in constraint logic programming. Finally, we describe the CIFF system, comparing it with state of the art abductive systems and answer set solvers and showing how to use it to program some applications. (To appear in Theory and Practice of Logic Programming - TPLP).
2408.04092
Steven Xia
Siyuan Xia, Chris Zhu, Tapan Srivastava, Bridget Fahey, Raul Castro Fernandez
Programmable Dataflows: Abstraction and Programming Model for Data Sharing
null
null
null
null
cs.DB
http://creativecommons.org/licenses/by-nc-nd/4.0/
Data sharing is central to a wide variety of applications such as fraud detection, ad matching, and research. The lack of data sharing abstractions makes the solution to each data sharing problem bespoke and cost-intensive, hampering value generation. In this paper, we first introduce a data sharing model to represent every data sharing problem with a sequence of dataflows. From the model, we distill an abstraction, the contract, which agents use to communicate the intent of a dataflow and evaluate its consequences, before the dataflow takes place. This helps agents move towards a common sharing goal without violating any regulatory and privacy constraints. Then, we design and implement the contract programming model (CPM), which allows agents to program data sharing applications catered to each problem's needs. Contracts permit data sharing, but their interactive nature may introduce inefficiencies. To mitigate those inefficiencies, we extend the CPM so that it can save intermediate outputs of dataflows, and skip computation if a dataflow tries to access data that it does not have access to. In our evaluation, we show that 1) the contract abstraction is general enough to represent a wide range of sharing problems, 2) we can write programs for complex data sharing problems and exhibit qualitative improvements over other alternate technologies, and 3) quantitatively, our optimizations make sharing programs written with the CPM efficient.
[ { "created": "Wed, 7 Aug 2024 21:15:57 GMT", "version": "v1" } ]
2024-08-09
[ [ "Xia", "Siyuan", "" ], [ "Zhu", "Chris", "" ], [ "Srivastava", "Tapan", "" ], [ "Fahey", "Bridget", "" ], [ "Fernandez", "Raul Castro", "" ] ]
Data sharing is central to a wide variety of applications such as fraud detection, ad matching, and research. The lack of data sharing abstractions makes the solution to each data sharing problem bespoke and cost-intensive, hampering value generation. In this paper, we first introduce a data sharing model to represent every data sharing problem with a sequence of dataflows. From the model, we distill an abstraction, the contract, which agents use to communicate the intent of a dataflow and evaluate its consequences, before the dataflow takes place. This helps agents move towards a common sharing goal without violating any regulatory and privacy constraints. Then, we design and implement the contract programming model (CPM), which allows agents to program data sharing applications catered to each problem's needs. Contracts permit data sharing, but their interactive nature may introduce inefficiencies. To mitigate those inefficiencies, we extend the CPM so that it can save intermediate outputs of dataflows, and skip computation if a dataflow tries to access data that it does not have access to. In our evaluation, we show that 1) the contract abstraction is general enough to represent a wide range of sharing problems, 2) we can write programs for complex data sharing problems and exhibit qualitative improvements over other alternate technologies, and 3) quantitatively, our optimizations make sharing programs written with the CPM efficient.
1901.09002
Tai Sing Lee
Jielin Qiu, Ge Huang, Tai Sing Lee
A Neurally-Inspired Hierarchical Prediction Network for Spatiotemporal Sequence Learning and Prediction
Some of the results are not replicable
null
null
null
cs.NE cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In this paper we developed a hierarchical network model, called Hierarchical Prediction Network (HPNet), to understand how spatiotemporal memories might be learned and encoded in the recurrent circuits in the visual cortical hierarchy for predicting future video frames. This neurally inspired model operates in the analysis-by-synthesis framework. It contains a feed-forward path that computes and encodes spatiotemporal features of successive complexity and a feedback path for the successive levels to project their interpretations to the level below. Within each level, the feed-forward path and the feedback path intersect in a recurrent gated circuit, instantiated in a LSTM module, to generate a prediction or explanation of the incoming signals. The network learns its internal model of the world by minimizing the errors of its prediction of the incoming signals at each level of the hierarchy. We found that hierarchical interaction in the network increases semantic clustering of global movement patterns in the population codes of the units along the hierarchy, even in the earliest module. This facilitates the learning of relationships among movement patterns, yielding state-of-the-art performance in long range video sequence predictions in the benchmark datasets. The network model automatically reproduces a variety of prediction suppression and familiarity suppression neurophysiological phenomena observed in the visual cortex, suggesting that hierarchical prediction might indeed be an important principle for representational learning in the visual cortex.
[ { "created": "Fri, 25 Jan 2019 18:03:17 GMT", "version": "v1" }, { "created": "Fri, 1 Oct 2021 12:59:31 GMT", "version": "v2" } ]
2021-10-04
[ [ "Qiu", "Jielin", "" ], [ "Huang", "Ge", "" ], [ "Lee", "Tai Sing", "" ] ]
In this paper we developed a hierarchical network model, called Hierarchical Prediction Network (HPNet), to understand how spatiotemporal memories might be learned and encoded in the recurrent circuits in the visual cortical hierarchy for predicting future video frames. This neurally inspired model operates in the analysis-by-synthesis framework. It contains a feed-forward path that computes and encodes spatiotemporal features of successive complexity and a feedback path for the successive levels to project their interpretations to the level below. Within each level, the feed-forward path and the feedback path intersect in a recurrent gated circuit, instantiated in a LSTM module, to generate a prediction or explanation of the incoming signals. The network learns its internal model of the world by minimizing the errors of its prediction of the incoming signals at each level of the hierarchy. We found that hierarchical interaction in the network increases semantic clustering of global movement patterns in the population codes of the units along the hierarchy, even in the earliest module. This facilitates the learning of relationships among movement patterns, yielding state-of-the-art performance in long range video sequence predictions in the benchmark datasets. The network model automatically reproduces a variety of prediction suppression and familiarity suppression neurophysiological phenomena observed in the visual cortex, suggesting that hierarchical prediction might indeed be an important principle for representational learning in the visual cortex.
1207.1257
Anton Belov
Anton Belov and Joao Marques-Silva
Generalizing Redundancy in Propositional Logic: Foundations and Hitting Sets Duality
13 pages; first part of series on labelled CNF formulas; fixed some references
null
null
null
cs.LO cs.AI
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Detection and elimination of redundant clauses from propositional formulas in Conjunctive Normal Form (CNF) is a fundamental problem with numerous application domains, including AI, and has been the subject of extensive research. Moreover, a number of recent applications motivated various extensions of this problem. For example, unsatisfiable formulas partitioned into disjoint subsets of clauses (so-called groups) often need to be simplified by removing redundant groups, or may contain redundant variables, rather than clauses. In this report we present a generalized theoretical framework of labelled CNF formulas that unifies various extensions of the redundancy detection and removal problem and allows to derive a number of results that subsume and extend previous work. The follow-up reports contain a number of additional theoretical results and algorithms for various computational problems in the context of the proposed framework.
[ { "created": "Thu, 5 Jul 2012 13:57:40 GMT", "version": "v1" }, { "created": "Tue, 10 Jul 2012 01:15:08 GMT", "version": "v2" } ]
2012-07-11
[ [ "Belov", "Anton", "" ], [ "Marques-Silva", "Joao", "" ] ]
Detection and elimination of redundant clauses from propositional formulas in Conjunctive Normal Form (CNF) is a fundamental problem with numerous application domains, including AI, and has been the subject of extensive research. Moreover, a number of recent applications motivated various extensions of this problem. For example, unsatisfiable formulas partitioned into disjoint subsets of clauses (so-called groups) often need to be simplified by removing redundant groups, or may contain redundant variables, rather than clauses. In this report we present a generalized theoretical framework of labelled CNF formulas that unifies various extensions of the redundancy detection and removal problem and allows to derive a number of results that subsume and extend previous work. The follow-up reports contain a number of additional theoretical results and algorithms for various computational problems in the context of the proposed framework.
2011.10534
Olivier Carton
Olivier Carton
Ambiguity through the lens of measure theory
null
null
null
null
cs.FL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In this paper, we establish a strong link between the ambiguity for finite words of a B\"uchi automaton and the ambiguity for infinite words of the same automaton. This link is based on measure theory. More precisely, we show that such an automaton is unambiguous, in the sense that no finite word labels two runs with the same starting state and the same ending state if and only if for each state, the set of infinite sequences labelling two runs starting from that state has measure zero. The measure used to define these negligible sets, that is sets of measure zero, can be any measure computed by a weighted automaton which is compatible with the B\"uchi automaton. This latter condition is very natural: the measure must put weight on cylinders [w] where w is the label of some run in the B\"uchi automaton.
[ { "created": "Fri, 20 Nov 2020 18:06:10 GMT", "version": "v1" }, { "created": "Thu, 27 Jan 2022 17:48:57 GMT", "version": "v2" }, { "created": "Tue, 8 Feb 2022 14:26:46 GMT", "version": "v3" }, { "created": "Fri, 22 Apr 2022 16:13:46 GMT", "version": "v4" } ]
2022-04-25
[ [ "Carton", "Olivier", "" ] ]
In this paper, we establish a strong link between the ambiguity for finite words of a B\"uchi automaton and the ambiguity for infinite words of the same automaton. This link is based on measure theory. More precisely, we show that such an automaton is unambiguous, in the sense that no finite word labels two runs with the same starting state and the same ending state if and only if for each state, the set of infinite sequences labelling two runs starting from that state has measure zero. The measure used to define these negligible sets, that is sets of measure zero, can be any measure computed by a weighted automaton which is compatible with the B\"uchi automaton. This latter condition is very natural: the measure must put weight on cylinders [w] where w is the label of some run in the B\"uchi automaton.
2406.13355
Umberto Mart\'inez-Pe\~nas
Umberto Mart\'inez-Pe\~nas and Rub\'en Rodr\'iguez-Ballesteros
Linear codes in the folded Hamming distance and the quasi MDS property
null
null
null
null
cs.IT math.IT
http://creativecommons.org/publicdomain/zero/1.0/
In this work, we study linear codes with the folded Hamming distance, or equivalently, codes with the classical Hamming distance that are linear over a subfield. This includes additive codes. We study MDS codes in this setting and define quasi MDS (QMDS) codes and dually QMDS codes, which attain a more relaxed variant of the classical Singleton bound. We provide several general results concerning these codes, including restriction, shortening, weight distributions, existence, density, geometric description and bounds on their lengths relative to their field sizes. We provide explicit examples and a binary construction with optimal lengths relative to their field sizes, which beats any MDS code.
[ { "created": "Wed, 19 Jun 2024 09:01:11 GMT", "version": "v1" } ]
2024-06-21
[ [ "Martínez-Peñas", "Umberto", "" ], [ "Rodríguez-Ballesteros", "Rubén", "" ] ]
In this work, we study linear codes with the folded Hamming distance, or equivalently, codes with the classical Hamming distance that are linear over a subfield. This includes additive codes. We study MDS codes in this setting and define quasi MDS (QMDS) codes and dually QMDS codes, which attain a more relaxed variant of the classical Singleton bound. We provide several general results concerning these codes, including restriction, shortening, weight distributions, existence, density, geometric description and bounds on their lengths relative to their field sizes. We provide explicit examples and a binary construction with optimal lengths relative to their field sizes, which beats any MDS code.
1904.12058
Muhan Zhang
Muhan Zhang, Yixin Chen
Inductive Matrix Completion Based on Graph Neural Networks
Accepted as a spotlight presentation at ICLR-2020
null
null
null
cs.IR cs.LG stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We propose an inductive matrix completion model without using side information. By factorizing the (rating) matrix into the product of low-dimensional latent embeddings of rows (users) and columns (items), a majority of existing matrix completion methods are transductive, since the learned embeddings cannot generalize to unseen rows/columns or to new matrices. To make matrix completion inductive, most previous works use content (side information), such as user's age or movie's genre, to make predictions. However, high-quality content is not always available, and can be hard to extract. Under the extreme setting where not any side information is available other than the matrix to complete, can we still learn an inductive matrix completion model? In this paper, we propose an Inductive Graph-based Matrix Completion (IGMC) model to address this problem. IGMC trains a graph neural network (GNN) based purely on 1-hop subgraphs around (user, item) pairs generated from the rating matrix and maps these subgraphs to their corresponding ratings. It achieves highly competitive performance with state-of-the-art transductive baselines. In addition, IGMC is inductive -- it can generalize to users/items unseen during the training (given that their interactions exist), and can even transfer to new tasks. Our transfer learning experiments show that a model trained out of the MovieLens dataset can be directly used to predict Douban movie ratings with surprisingly good performance. Our work demonstrates that: 1) it is possible to train inductive matrix completion models without using side information while achieving similar or better performances than state-of-the-art transductive methods; 2) local graph patterns around a (user, item) pair are effective predictors of the rating this user gives to the item; and 3) Long-range dependencies might not be necessary for modeling recommender systems.
[ { "created": "Fri, 26 Apr 2019 21:58:46 GMT", "version": "v1" }, { "created": "Sun, 6 Oct 2019 03:08:54 GMT", "version": "v2" }, { "created": "Sun, 16 Feb 2020 04:27:14 GMT", "version": "v3" } ]
2020-02-18
[ [ "Zhang", "Muhan", "" ], [ "Chen", "Yixin", "" ] ]
We propose an inductive matrix completion model without using side information. By factorizing the (rating) matrix into the product of low-dimensional latent embeddings of rows (users) and columns (items), a majority of existing matrix completion methods are transductive, since the learned embeddings cannot generalize to unseen rows/columns or to new matrices. To make matrix completion inductive, most previous works use content (side information), such as user's age or movie's genre, to make predictions. However, high-quality content is not always available, and can be hard to extract. Under the extreme setting where not any side information is available other than the matrix to complete, can we still learn an inductive matrix completion model? In this paper, we propose an Inductive Graph-based Matrix Completion (IGMC) model to address this problem. IGMC trains a graph neural network (GNN) based purely on 1-hop subgraphs around (user, item) pairs generated from the rating matrix and maps these subgraphs to their corresponding ratings. It achieves highly competitive performance with state-of-the-art transductive baselines. In addition, IGMC is inductive -- it can generalize to users/items unseen during the training (given that their interactions exist), and can even transfer to new tasks. Our transfer learning experiments show that a model trained out of the MovieLens dataset can be directly used to predict Douban movie ratings with surprisingly good performance. Our work demonstrates that: 1) it is possible to train inductive matrix completion models without using side information while achieving similar or better performances than state-of-the-art transductive methods; 2) local graph patterns around a (user, item) pair are effective predictors of the rating this user gives to the item; and 3) Long-range dependencies might not be necessary for modeling recommender systems.
2203.10744
Chen Wu
Changran Hu, Akshara Reddi Methukupalli, Yutong Zhou, Chen Wu, Yubo Chen
Programming Language Agnostic Mining of Code and Language Pairs with Sequence Labeling Based Question Answering
null
null
null
null
cs.CL cs.PL
http://creativecommons.org/licenses/by/4.0/
Mining aligned natural language (NL) and programming language (PL) pairs is a critical task to NL-PL understanding. Existing methods applied specialized hand-crafted features or separately-trained models for each PL. However, they usually suffered from low transferability across multiple PLs, especially for niche PLs with less annotated data. Fortunately, a Stack Overflow answer post is essentially a sequence of text and code blocks and its global textual context can provide PL-agnostic supplementary information. In this paper, we propose a Sequence Labeling based Question Answering (SLQA) method to mine NL-PL pairs in a PL-agnostic manner. In particular, we propose to apply the BIO tagging scheme instead of the conventional binary scheme to mine the code solutions which are often composed of multiple blocks of a post. Experiments on current single-PL single-block benchmarks and a manually-labeled cross-PL multi-block benchmark prove the effectiveness and transferability of SLQA. We further present a parallel NL-PL corpus named Lang2Code automatically mined with SLQA, which contains about 1.4M pairs on 6 PLs. Under statistical analysis and downstream evaluation, we demonstrate that Lang2Code is a large-scale high-quality data resource for further NL-PL research.
[ { "created": "Mon, 21 Mar 2022 05:33:59 GMT", "version": "v1" } ]
2022-03-22
[ [ "Hu", "Changran", "" ], [ "Methukupalli", "Akshara Reddi", "" ], [ "Zhou", "Yutong", "" ], [ "Wu", "Chen", "" ], [ "Chen", "Yubo", "" ] ]
Mining aligned natural language (NL) and programming language (PL) pairs is a critical task to NL-PL understanding. Existing methods applied specialized hand-crafted features or separately-trained models for each PL. However, they usually suffered from low transferability across multiple PLs, especially for niche PLs with less annotated data. Fortunately, a Stack Overflow answer post is essentially a sequence of text and code blocks and its global textual context can provide PL-agnostic supplementary information. In this paper, we propose a Sequence Labeling based Question Answering (SLQA) method to mine NL-PL pairs in a PL-agnostic manner. In particular, we propose to apply the BIO tagging scheme instead of the conventional binary scheme to mine the code solutions which are often composed of multiple blocks of a post. Experiments on current single-PL single-block benchmarks and a manually-labeled cross-PL multi-block benchmark prove the effectiveness and transferability of SLQA. We further present a parallel NL-PL corpus named Lang2Code automatically mined with SLQA, which contains about 1.4M pairs on 6 PLs. Under statistical analysis and downstream evaluation, we demonstrate that Lang2Code is a large-scale high-quality data resource for further NL-PL research.
2302.04529
Martijn Goorden
Martijn A. Goorden, Kim G. Larsen, Axel Legay, Florian Lorber, Ulrik Nyman, Andrzej Wasowski
Timed I/O Automata: It is never too late to complete your timed specification theory
Version submitted for review
null
null
null
cs.FL cs.SE
http://creativecommons.org/licenses/by/4.0/
A specification theory combines notions of specifications and implementations with a satisfaction relation, a refinement relation and a set of operators supporting stepwise design. We develop a complete specification framework for real-time systems using Timed I/O Automata as the specification formalism, with the semantics expressed in terms of Timed I/O Transition Systems. We provide constructs for refinement, consistency checking, logical and structural composition, and quotient of specifications -- all indispensable ingredients of a compositional design methodology. The theory is backed by rigorous proofs and is being implemented in the open-source tool ECDAR.
[ { "created": "Thu, 9 Feb 2023 09:41:48 GMT", "version": "v1" }, { "created": "Thu, 13 Jul 2023 07:50:12 GMT", "version": "v2" } ]
2023-07-14
[ [ "Goorden", "Martijn A.", "" ], [ "Larsen", "Kim G.", "" ], [ "Legay", "Axel", "" ], [ "Lorber", "Florian", "" ], [ "Nyman", "Ulrik", "" ], [ "Wasowski", "Andrzej", "" ] ]
A specification theory combines notions of specifications and implementations with a satisfaction relation, a refinement relation and a set of operators supporting stepwise design. We develop a complete specification framework for real-time systems using Timed I/O Automata as the specification formalism, with the semantics expressed in terms of Timed I/O Transition Systems. We provide constructs for refinement, consistency checking, logical and structural composition, and quotient of specifications -- all indispensable ingredients of a compositional design methodology. The theory is backed by rigorous proofs and is being implemented in the open-source tool ECDAR.
2104.04474
Chavit Denninnart
Chavit Denninnart, Mohsen Amini Salehi
Harnessing the Potential of Function-Reuse in Multimedia Cloud Systems
null
null
null
null
cs.DC cs.MM
http://creativecommons.org/licenses/by/4.0/
Cloud-based computing systems can get oversubscribed due to the budget constraints of their users or limitations in certain resource types. The oversubscription can, in turn, degrade the users perceived Quality of Service (QoS). The approach we investigate to mitigate both the oversubscription and the incurred cost is based on smart reusing of the computation needed to process the service requests (i.e., tasks). We propose a reusing paradigm for the tasks that are waiting for execution. This paradigm can be particularly impactful in serverless platforms where multiple users can request similar services simultaneously. Our motivation is a multimedia streaming engine that processes the media segments in an on-demand manner. We propose a mechanism to identify various types of "mergeable" tasks and aggregate them to improve the QoS and mitigate the incurred cost. We develop novel approaches to determine when and how to perform task aggregation such that the QoS of other tasks is not affected. Evaluation results show that the proposed mechanism can improve the QoS by significantly reducing the percentage of tasks missing their deadlines %. In addition, it can and reduce the overall time (and subsequently the incurred cost) of utilizing cloud services by more than 9%.
[ { "created": "Fri, 9 Apr 2021 16:45:53 GMT", "version": "v1" } ]
2021-04-12
[ [ "Denninnart", "Chavit", "" ], [ "Salehi", "Mohsen Amini", "" ] ]
Cloud-based computing systems can get oversubscribed due to the budget constraints of their users or limitations in certain resource types. The oversubscription can, in turn, degrade the users perceived Quality of Service (QoS). The approach we investigate to mitigate both the oversubscription and the incurred cost is based on smart reusing of the computation needed to process the service requests (i.e., tasks). We propose a reusing paradigm for the tasks that are waiting for execution. This paradigm can be particularly impactful in serverless platforms where multiple users can request similar services simultaneously. Our motivation is a multimedia streaming engine that processes the media segments in an on-demand manner. We propose a mechanism to identify various types of "mergeable" tasks and aggregate them to improve the QoS and mitigate the incurred cost. We develop novel approaches to determine when and how to perform task aggregation such that the QoS of other tasks is not affected. Evaluation results show that the proposed mechanism can improve the QoS by significantly reducing the percentage of tasks missing their deadlines %. In addition, it can and reduce the overall time (and subsequently the incurred cost) of utilizing cloud services by more than 9%.
2108.05876
Emiliano De Cristofaro
Pujan Paudel, Jeremy Blackburn, Emiliano De Cristofaro, Savvas Zannettou, and Gianluca Stringhini
An Early Look at the Gettr Social Network
null
null
null
null
cs.CY cs.SI
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
This paper presents the first data-driven analysis of Gettr, a new social network platform launched by former US President Donald Trump's team. Among other things, we find that users on the platform heavily discuss politics, with a focus on the Trump campaign in the US and Bolsonaro's in Brazil. Activity on the platform has steadily been decreasing since its launch, although a core of verified users and early adopters kept posting and become central to it. Finally, although toxicity has been increasing over time, the average level of toxicity is still lower than the one recently observed on other fringe social networks like Gab and 4chan. Overall, we provide a first quantitative look at this new community, observing a lack of organic engagement and activity.
[ { "created": "Thu, 12 Aug 2021 17:49:30 GMT", "version": "v1" } ]
2021-08-13
[ [ "Paudel", "Pujan", "" ], [ "Blackburn", "Jeremy", "" ], [ "De Cristofaro", "Emiliano", "" ], [ "Zannettou", "Savvas", "" ], [ "Stringhini", "Gianluca", "" ] ]
This paper presents the first data-driven analysis of Gettr, a new social network platform launched by former US President Donald Trump's team. Among other things, we find that users on the platform heavily discuss politics, with a focus on the Trump campaign in the US and Bolsonaro's in Brazil. Activity on the platform has steadily been decreasing since its launch, although a core of verified users and early adopters kept posting and become central to it. Finally, although toxicity has been increasing over time, the average level of toxicity is still lower than the one recently observed on other fringe social networks like Gab and 4chan. Overall, we provide a first quantitative look at this new community, observing a lack of organic engagement and activity.
1712.07642
Fereshteh Sadeghi
Fereshteh Sadeghi, Alexander Toshev, Eric Jang, Sergey Levine
Sim2Real View Invariant Visual Servoing by Recurrent Control
Supplementary video: https://fsadeghi.github.io/Sim2RealViewInvariantServo
null
null
null
cs.CV cs.LG cs.RO
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Humans are remarkably proficient at controlling their limbs and tools from a wide range of viewpoints and angles, even in the presence of optical distortions. In robotics, this ability is referred to as visual servoing: moving a tool or end-point to a desired location using primarily visual feedback. In this paper, we study how viewpoint-invariant visual servoing skills can be learned automatically in a robotic manipulation scenario. To this end, we train a deep recurrent controller that can automatically determine which actions move the end-point of a robotic arm to a desired object. The problem that must be solved by this controller is fundamentally ambiguous: under severe variation in viewpoint, it may be impossible to determine the actions in a single feedforward operation. Instead, our visual servoing system must use its memory of past movements to understand how the actions affect the robot motion from the current viewpoint, correcting mistakes and gradually moving closer to the target. This ability is in stark contrast to most visual servoing methods, which either assume known dynamics or require a calibration phase. We show how we can learn this recurrent controller using simulated data and a reinforcement learning objective. We then describe how the resulting model can be transferred to a real-world robot by disentangling perception from control and only adapting the visual layers. The adapted model can servo to previously unseen objects from novel viewpoints on a real-world Kuka IIWA robotic arm. For supplementary videos, see: https://fsadeghi.github.io/Sim2RealViewInvariantServo
[ { "created": "Wed, 20 Dec 2017 18:54:29 GMT", "version": "v1" } ]
2017-12-21
[ [ "Sadeghi", "Fereshteh", "" ], [ "Toshev", "Alexander", "" ], [ "Jang", "Eric", "" ], [ "Levine", "Sergey", "" ] ]
Humans are remarkably proficient at controlling their limbs and tools from a wide range of viewpoints and angles, even in the presence of optical distortions. In robotics, this ability is referred to as visual servoing: moving a tool or end-point to a desired location using primarily visual feedback. In this paper, we study how viewpoint-invariant visual servoing skills can be learned automatically in a robotic manipulation scenario. To this end, we train a deep recurrent controller that can automatically determine which actions move the end-point of a robotic arm to a desired object. The problem that must be solved by this controller is fundamentally ambiguous: under severe variation in viewpoint, it may be impossible to determine the actions in a single feedforward operation. Instead, our visual servoing system must use its memory of past movements to understand how the actions affect the robot motion from the current viewpoint, correcting mistakes and gradually moving closer to the target. This ability is in stark contrast to most visual servoing methods, which either assume known dynamics or require a calibration phase. We show how we can learn this recurrent controller using simulated data and a reinforcement learning objective. We then describe how the resulting model can be transferred to a real-world robot by disentangling perception from control and only adapting the visual layers. The adapted model can servo to previously unseen objects from novel viewpoints on a real-world Kuka IIWA robotic arm. For supplementary videos, see: https://fsadeghi.github.io/Sim2RealViewInvariantServo
2403.13433
Zhouhong Gu
Zhouhong Gu, Xiaoxuan Zhu, Haoran Guo, Lin Zhang, Yin Cai, Hao Shen, Jiangjie Chen, Zheyu Ye, Yifei Dai, Yan Gao, Yao Hu, Hongwei Feng, Yanghua Xiao
AgentGroupChat: An Interactive Group Chat Simulacra For Better Eliciting Emergent Behavior
null
null
null
null
cs.AI cs.CL cs.CY
http://creativecommons.org/licenses/by/4.0/
Language significantly influences the formation and evolution of Human emergent behavior, which is crucial in understanding collective intelligence within human societies. Considering that the study of how language affects human behavior needs to put it into the dynamic scenarios in which it is used, we introduce AgentGroupChat in this paper, a simulation that delves into the complex role of language in shaping collective behavior through interactive debate scenarios. Central to this simulation are characters engaging in dynamic conversation interactions. To enable simulation, we introduce the Verbal Strategist Agent, utilizing large language models to enhance interaction strategies by incorporating elements of persona and action. We set four narrative scenarios based on AgentGroupChat to demonstrate the simulation's capacity to mimic complex language use in group dynamics. Evaluations focus on aligning agent behaviors with human expectations and the emergence of collective behaviors within the simulation. Results reveal that emergent behaviors materialize from a confluence of factors: a conducive environment for extensive information exchange, characters with diverse traits, high linguistic comprehension, and strategic adaptability. During discussions on ``the impact of AI on humanity'' in AgentGroupChat simulation, philosophers commonly agreed that ``AI could enhance societal welfare with judicious limitations'' and even come to a conclusion that ``the essence of true intelligence encompasses understanding the necessity to constrain self abilities''. Additionally, in the competitive domain of casting for primary roles in films in AgentGroupChat, certain actors were ready to reduce their remuneration or accept lesser roles, motivated by their deep-seated desire to contribute to the project.
[ { "created": "Wed, 20 Mar 2024 09:21:32 GMT", "version": "v1" }, { "created": "Thu, 4 Apr 2024 07:40:31 GMT", "version": "v2" } ]
2024-04-05
[ [ "Gu", "Zhouhong", "" ], [ "Zhu", "Xiaoxuan", "" ], [ "Guo", "Haoran", "" ], [ "Zhang", "Lin", "" ], [ "Cai", "Yin", "" ], [ "Shen", "Hao", "" ], [ "Chen", "Jiangjie", "" ], [ "Ye", "Zheyu", "" ], [ "Dai", "Yifei", "" ], [ "Gao", "Yan", "" ], [ "Hu", "Yao", "" ], [ "Feng", "Hongwei", "" ], [ "Xiao", "Yanghua", "" ] ]
Language significantly influences the formation and evolution of Human emergent behavior, which is crucial in understanding collective intelligence within human societies. Considering that the study of how language affects human behavior needs to put it into the dynamic scenarios in which it is used, we introduce AgentGroupChat in this paper, a simulation that delves into the complex role of language in shaping collective behavior through interactive debate scenarios. Central to this simulation are characters engaging in dynamic conversation interactions. To enable simulation, we introduce the Verbal Strategist Agent, utilizing large language models to enhance interaction strategies by incorporating elements of persona and action. We set four narrative scenarios based on AgentGroupChat to demonstrate the simulation's capacity to mimic complex language use in group dynamics. Evaluations focus on aligning agent behaviors with human expectations and the emergence of collective behaviors within the simulation. Results reveal that emergent behaviors materialize from a confluence of factors: a conducive environment for extensive information exchange, characters with diverse traits, high linguistic comprehension, and strategic adaptability. During discussions on ``the impact of AI on humanity'' in AgentGroupChat simulation, philosophers commonly agreed that ``AI could enhance societal welfare with judicious limitations'' and even come to a conclusion that ``the essence of true intelligence encompasses understanding the necessity to constrain self abilities''. Additionally, in the competitive domain of casting for primary roles in films in AgentGroupChat, certain actors were ready to reduce their remuneration or accept lesser roles, motivated by their deep-seated desire to contribute to the project.
2006.07123
Roman Pogodin
Roman Pogodin and Peter E. Latham
Kernelized information bottleneck leads to biologically plausible 3-factor Hebbian learning in deep networks
Accepted to NeurIPS 2020
null
null
null
cs.LG q-bio.NC stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
The state-of-the art machine learning approach to training deep neural networks, backpropagation, is implausible for real neural networks: neurons need to know their outgoing weights; training alternates between a bottom-up forward pass (computation) and a top-down backward pass (learning); and the algorithm often needs precise labels of many data points. Biologically plausible approximations to backpropagation, such as feedback alignment, solve the weight transport problem, but not the other two. Thus, fully biologically plausible learning rules have so far remained elusive. Here we present a family of learning rules that does not suffer from any of these problems. It is motivated by the information bottleneck principle (extended with kernel methods), in which networks learn to compress the input as much as possible without sacrificing prediction of the output. The resulting rules have a 3-factor Hebbian structure: they require pre- and post-synaptic firing rates and an error signal - the third factor - consisting of a global teaching signal and a layer-specific term, both available without a top-down pass. They do not require precise labels; instead, they rely on the similarity between pairs of desired outputs. Moreover, to obtain good performance on hard problems and retain biological plausibility, our rules need divisive normalization - a known feature of biological networks. Finally, simulations show that our rules perform nearly as well as backpropagation on image classification tasks.
[ { "created": "Fri, 12 Jun 2020 12:30:53 GMT", "version": "v1" }, { "created": "Fri, 23 Oct 2020 17:00:59 GMT", "version": "v2" } ]
2020-10-26
[ [ "Pogodin", "Roman", "" ], [ "Latham", "Peter E.", "" ] ]
The state-of-the art machine learning approach to training deep neural networks, backpropagation, is implausible for real neural networks: neurons need to know their outgoing weights; training alternates between a bottom-up forward pass (computation) and a top-down backward pass (learning); and the algorithm often needs precise labels of many data points. Biologically plausible approximations to backpropagation, such as feedback alignment, solve the weight transport problem, but not the other two. Thus, fully biologically plausible learning rules have so far remained elusive. Here we present a family of learning rules that does not suffer from any of these problems. It is motivated by the information bottleneck principle (extended with kernel methods), in which networks learn to compress the input as much as possible without sacrificing prediction of the output. The resulting rules have a 3-factor Hebbian structure: they require pre- and post-synaptic firing rates and an error signal - the third factor - consisting of a global teaching signal and a layer-specific term, both available without a top-down pass. They do not require precise labels; instead, they rely on the similarity between pairs of desired outputs. Moreover, to obtain good performance on hard problems and retain biological plausibility, our rules need divisive normalization - a known feature of biological networks. Finally, simulations show that our rules perform nearly as well as backpropagation on image classification tasks.
1706.00399
Veit Elser
Veit Elser, Ti-Yen Lan and Tamir Bendory
Benchmark problems for phase retrieval
27 pages, 10 figures, new references, modified appendix, improved presentation
null
null
null
cs.IT math.IT physics.data-an
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In recent years, the mathematical and algorithmic aspects of the phase retrieval problem have received considerable attention. Many papers in this area mention crystallography as a principal application. In crystallography, the signal to be recovered is periodic and comprised of atomic distributions arranged homogeneously in the unit cell of the crystal. The crystallographic problem is both the leading application and one of the hardest forms of phase retrieval. We have constructed a graded set of benchmark problems for evaluating algorithms that perform this type of phase retrieval. The data, publicly available online, is provided in an easily interpretable format. We also propose a simple and unambiguous success/failure criterion based on the actual needs in crystallography. Baseline runtimes were obtained with an iterative algorithm that is similar but more transparent than those used in crystallography. Empirically, the runtimes grow exponentially with respect to a new hardness parameter: the sparsity of the signal autocorrelation. We also review the algorithms used by the leading software packages. This set of benchmark problems, we hope, will encourage the development of new algorithms for the phase retrieval problem in general, and crystallography in particular.
[ { "created": "Thu, 1 Jun 2017 17:22:43 GMT", "version": "v1" }, { "created": "Tue, 13 Feb 2018 20:31:32 GMT", "version": "v2" }, { "created": "Thu, 14 Jun 2018 15:28:00 GMT", "version": "v3" } ]
2018-06-15
[ [ "Elser", "Veit", "" ], [ "Lan", "Ti-Yen", "" ], [ "Bendory", "Tamir", "" ] ]
In recent years, the mathematical and algorithmic aspects of the phase retrieval problem have received considerable attention. Many papers in this area mention crystallography as a principal application. In crystallography, the signal to be recovered is periodic and comprised of atomic distributions arranged homogeneously in the unit cell of the crystal. The crystallographic problem is both the leading application and one of the hardest forms of phase retrieval. We have constructed a graded set of benchmark problems for evaluating algorithms that perform this type of phase retrieval. The data, publicly available online, is provided in an easily interpretable format. We also propose a simple and unambiguous success/failure criterion based on the actual needs in crystallography. Baseline runtimes were obtained with an iterative algorithm that is similar but more transparent than those used in crystallography. Empirically, the runtimes grow exponentially with respect to a new hardness parameter: the sparsity of the signal autocorrelation. We also review the algorithms used by the leading software packages. This set of benchmark problems, we hope, will encourage the development of new algorithms for the phase retrieval problem in general, and crystallography in particular.
2312.09087
J\"ames M\'en\'etrey
J\"ames M\'en\'etrey and Marcelo Pasin and Pascal Felber and Valerio Schiavoni and Giovanni Mazzeo and Arne Hollum and Darshan Vaydia
A Comprehensive Trusted Runtime for WebAssembly with Intel SGX
This publication incorporates results from the VEDLIoT project, which received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 957197. arXiv admin note: text overlap with arXiv:2103.15860
TDSC: IEEE Transactions on Dependable and Secure Computing, November, 2023
10.1109/TDSC.2023.3334516
null
cs.CR cs.PF cs.PL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In real-world scenarios, trusted execution environments (TEEs) frequently host applications that lack the trust of the infrastructure provider, as well as data owners who have specifically outsourced their data for remote processing. We present Twine, a trusted runtime for running WebAssembly-compiled applications within TEEs, establishing a two-way sandbox. Twine leverages memory safety guarantees of WebAssembly (Wasm) and abstracts the complexity of TEEs, empowering the execution of legacy and language-agnostic applications. It extends the standard WebAssembly system interface (WASI), providing controlled OS services, focusing on I/O. Additionally, through built-in TEE mechanisms, Twine delivers attestation capabilities to ensure the integrity of the runtime and the OS services supplied to the application. We evaluate its performance using general-purpose benchmarks and real-world applications, showing it compares on par with state-of-the-art solutions. A case study involving fintech company Credora reveals that Twine can be deployed in production with reasonable performance trade-offs, ranging from a 0.7x slowdown to a 1.17x speedup compared to native run time. Finally, we identify performance improvement through library optimisation, showcasing one such adjustment that leads up to 4.1x speedup. Twine is open-source and has been upstreamed into the original Wasm runtime, WAMR.
[ { "created": "Thu, 14 Dec 2023 16:19:00 GMT", "version": "v1" } ]
2023-12-15
[ [ "Ménétrey", "Jämes", "" ], [ "Pasin", "Marcelo", "" ], [ "Felber", "Pascal", "" ], [ "Schiavoni", "Valerio", "" ], [ "Mazzeo", "Giovanni", "" ], [ "Hollum", "Arne", "" ], [ "Vaydia", "Darshan", "" ] ]
In real-world scenarios, trusted execution environments (TEEs) frequently host applications that lack the trust of the infrastructure provider, as well as data owners who have specifically outsourced their data for remote processing. We present Twine, a trusted runtime for running WebAssembly-compiled applications within TEEs, establishing a two-way sandbox. Twine leverages memory safety guarantees of WebAssembly (Wasm) and abstracts the complexity of TEEs, empowering the execution of legacy and language-agnostic applications. It extends the standard WebAssembly system interface (WASI), providing controlled OS services, focusing on I/O. Additionally, through built-in TEE mechanisms, Twine delivers attestation capabilities to ensure the integrity of the runtime and the OS services supplied to the application. We evaluate its performance using general-purpose benchmarks and real-world applications, showing it compares on par with state-of-the-art solutions. A case study involving fintech company Credora reveals that Twine can be deployed in production with reasonable performance trade-offs, ranging from a 0.7x slowdown to a 1.17x speedup compared to native run time. Finally, we identify performance improvement through library optimisation, showcasing one such adjustment that leads up to 4.1x speedup. Twine is open-source and has been upstreamed into the original Wasm runtime, WAMR.
0712.4075
Vitaly Skachek
Vitaly Skachek, Mark F. Flanagan, Eimear Byrne, Marcus Greferath
Polytope Representations for Linear-Programming Decoding of Non-Binary Linear Codes
5 pages, to appear in 2008 IEEE International Symposium on Information Theory
null
10.1109/ISIT.2008.4595239
null
cs.IT math.IT
null
In previous work, we demonstrated how decoding of a non-binary linear code could be formulated as a linear-programming problem. In this paper, we study different polytopes for use with linear-programming decoding, and show that for many classes of codes these polytopes yield a complexity advantage for decoding. These representations lead to polynomial-time decoders for a wide variety of classical non-binary linear codes.
[ { "created": "Tue, 25 Dec 2007 15:44:01 GMT", "version": "v1" }, { "created": "Thu, 17 Apr 2008 17:35:15 GMT", "version": "v2" } ]
2016-11-18
[ [ "Skachek", "Vitaly", "" ], [ "Flanagan", "Mark F.", "" ], [ "Byrne", "Eimear", "" ], [ "Greferath", "Marcus", "" ] ]
In previous work, we demonstrated how decoding of a non-binary linear code could be formulated as a linear-programming problem. In this paper, we study different polytopes for use with linear-programming decoding, and show that for many classes of codes these polytopes yield a complexity advantage for decoding. These representations lead to polynomial-time decoders for a wide variety of classical non-binary linear codes.
2405.03089
Ismail Alkhouri
Xitong Zhang, Ismail R. Alkhouri, Rongrong Wang
Structure-Preserving Network Compression Via Low-Rank Induced Training Through Linear Layers Composition
null
null
null
null
cs.LG
http://creativecommons.org/licenses/by/4.0/
Deep Neural Networks (DNNs) have achieved remarkable success in addressing many previously unsolvable tasks. However, the storage and computational requirements associated with DNNs pose a challenge for deploying these trained models on resource-limited devices. Therefore, a plethora of compression and pruning techniques have been proposed in recent years. Low-rank decomposition techniques are among the approaches most utilized to address this problem. Compared to post-training compression, compression-promoted training is still under-explored. In this paper, we present a theoretically-justified novel approach, termed Low-Rank Induced Training (LoRITa), that promotes low-rankness through the composition of linear layers and compresses by using singular value truncation. This is achieved without the need to change the structure at inference time or require constrained and/or additional optimization, other than the standard weight decay regularization. Moreover, LoRITa eliminates the need to (i) initialize with pre-trained models and (ii) specify rank selection prior to training. Our experimental results (i) demonstrate the effectiveness of our approach using MNIST on Fully Connected Networks, CIFAR10 on Vision Transformers, and CIFAR10/100 on Convolutional Neural Networks, and (ii) illustrate that we achieve either competitive or SOTA results when compared to leading structured pruning methods in terms of FLOPs and parameters drop.
[ { "created": "Mon, 6 May 2024 00:58:23 GMT", "version": "v1" } ]
2024-05-07
[ [ "Zhang", "Xitong", "" ], [ "Alkhouri", "Ismail R.", "" ], [ "Wang", "Rongrong", "" ] ]
Deep Neural Networks (DNNs) have achieved remarkable success in addressing many previously unsolvable tasks. However, the storage and computational requirements associated with DNNs pose a challenge for deploying these trained models on resource-limited devices. Therefore, a plethora of compression and pruning techniques have been proposed in recent years. Low-rank decomposition techniques are among the approaches most utilized to address this problem. Compared to post-training compression, compression-promoted training is still under-explored. In this paper, we present a theoretically-justified novel approach, termed Low-Rank Induced Training (LoRITa), that promotes low-rankness through the composition of linear layers and compresses by using singular value truncation. This is achieved without the need to change the structure at inference time or require constrained and/or additional optimization, other than the standard weight decay regularization. Moreover, LoRITa eliminates the need to (i) initialize with pre-trained models and (ii) specify rank selection prior to training. Our experimental results (i) demonstrate the effectiveness of our approach using MNIST on Fully Connected Networks, CIFAR10 on Vision Transformers, and CIFAR10/100 on Convolutional Neural Networks, and (ii) illustrate that we achieve either competitive or SOTA results when compared to leading structured pruning methods in terms of FLOPs and parameters drop.
2306.08925
Xiaoyi Bao
Xiaoyi Bao, Xiaotong Jiang, Zhongqing Wang, Yue Zhang, and Guodong Zhou
Opinion Tree Parsing for Aspect-based Sentiment Analysis
null
null
null
null
cs.CL cs.AI
http://creativecommons.org/licenses/by/4.0/
Extracting sentiment elements using pre-trained generative models has recently led to large improvements in aspect-based sentiment analysis benchmarks. However, these models always need large-scale computing resources, and they also ignore explicit modeling of structure between sentiment elements. To address these challenges, we propose an opinion tree parsing model, aiming to parse all the sentiment elements from an opinion tree, which is much faster, and can explicitly reveal a more comprehensive and complete aspect-level sentiment structure. In particular, we first introduce a novel context-free opinion grammar to normalize the opinion tree structure. We then employ a neural chart-based opinion tree parser to fully explore the correlations among sentiment elements and parse them into an opinion tree structure. Extensive experiments show the superiority of our proposed model and the capacity of the opinion tree parser with the proposed context-free opinion grammar. More importantly, the results also prove that our model is much faster than previous models.
[ { "created": "Thu, 15 Jun 2023 07:53:14 GMT", "version": "v1" } ]
2023-06-16
[ [ "Bao", "Xiaoyi", "" ], [ "Jiang", "Xiaotong", "" ], [ "Wang", "Zhongqing", "" ], [ "Zhang", "Yue", "" ], [ "Zhou", "Guodong", "" ] ]
Extracting sentiment elements using pre-trained generative models has recently led to large improvements in aspect-based sentiment analysis benchmarks. However, these models always need large-scale computing resources, and they also ignore explicit modeling of structure between sentiment elements. To address these challenges, we propose an opinion tree parsing model, aiming to parse all the sentiment elements from an opinion tree, which is much faster, and can explicitly reveal a more comprehensive and complete aspect-level sentiment structure. In particular, we first introduce a novel context-free opinion grammar to normalize the opinion tree structure. We then employ a neural chart-based opinion tree parser to fully explore the correlations among sentiment elements and parse them into an opinion tree structure. Extensive experiments show the superiority of our proposed model and the capacity of the opinion tree parser with the proposed context-free opinion grammar. More importantly, the results also prove that our model is much faster than previous models.
1911.09325
Yafeng Liu
Liu Yafeng, Chen Tian, Liu Zhongyu, Zhang Lei, Hu Yanjun and Ding Enjie
Simultaneous Implementation Features Extraction and Recognition Using C3D Network for WiFi-based Human Activity Recognition
11 pages, 8 figures, 5 tables
null
null
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Human actions recognition has attracted more and more people's attention. Many technology have been developed to express human action's features, such as image, skeleton-based, and channel state information(CSI). Among them, on account of CSI's easy to be equipped and undemanding for light, and it has gained more and more attention in some special scene. However, the relationship between CSI signal and human actions is very complex, and some preliminary work must be done to make CSI features easy to understand for computer. Nowadays, many work departed CSI-based features' action dealing into two parts. One part is for features extraction and dimension reduce, and the other part is for time series problems. Some of them even omitted one of the two part work. Therefore, the accuracies of current recognition systems are far from satisfactory. In this paper, we propose a new deep learning based approach, i.e. C3D network and C3D network with attention mechanism, for human actions recognition using CSI signals. This kind of network can make feature extraction from spatial convolution and temporal convolution simultaneously, and through this network the two part of CSI-based human actions recognition mentioned above can be realized at the same time. The entire algorithm structure is simplified. The experimental results show that our proposed C3D network is able to achieve the best recognition performance for all activities when compared with some benchmark approaches.
[ { "created": "Thu, 21 Nov 2019 07:45:46 GMT", "version": "v1" } ]
2019-11-22
[ [ "Yafeng", "Liu", "" ], [ "Tian", "Chen", "" ], [ "Zhongyu", "Liu", "" ], [ "Lei", "Zhang", "" ], [ "Yanjun", "Hu", "" ], [ "Enjie", "Ding", "" ] ]
Human actions recognition has attracted more and more people's attention. Many technology have been developed to express human action's features, such as image, skeleton-based, and channel state information(CSI). Among them, on account of CSI's easy to be equipped and undemanding for light, and it has gained more and more attention in some special scene. However, the relationship between CSI signal and human actions is very complex, and some preliminary work must be done to make CSI features easy to understand for computer. Nowadays, many work departed CSI-based features' action dealing into two parts. One part is for features extraction and dimension reduce, and the other part is for time series problems. Some of them even omitted one of the two part work. Therefore, the accuracies of current recognition systems are far from satisfactory. In this paper, we propose a new deep learning based approach, i.e. C3D network and C3D network with attention mechanism, for human actions recognition using CSI signals. This kind of network can make feature extraction from spatial convolution and temporal convolution simultaneously, and through this network the two part of CSI-based human actions recognition mentioned above can be realized at the same time. The entire algorithm structure is simplified. The experimental results show that our proposed C3D network is able to achieve the best recognition performance for all activities when compared with some benchmark approaches.
1607.00595
Datong Zhou
Datong Zhou, Maximilian Balandat, Claire Tomlin
Residential Demand Response Targeting Using Machine Learning with Observational Data
8 pages
null
null
null
cs.SY
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
The large scale deployment of Advanced Metering Infrastructure among residential energy customers has served as a boon for energy systems research relying on granular consumption data. Residential Demand Response aims to utilize the flexibility of consumers to reduce their energy usage during times when the grid is strained. Suitable incentive mechanisms to encourage customers to deviate from their usual behavior have to be implemented to correctly control the bids into the wholesale electricity market as a Demand Response provider. In this paper, we present a framework for short term load forecasting on an individual user level, and relate nonexperimental estimates of Demand Response efficacy, i.e. the estimated reduction of consumption during Demand Response events, to the variability of user consumption. We apply our framework on a data set from a residential Demand Response program in the Western United States. Our results suggest that users with more variable consumption patterns are more likely to reduce their consumption compared to users with a more regular consumption behavior.
[ { "created": "Sun, 3 Jul 2016 05:32:50 GMT", "version": "v1" } ]
2016-07-05
[ [ "Zhou", "Datong", "" ], [ "Balandat", "Maximilian", "" ], [ "Tomlin", "Claire", "" ] ]
The large scale deployment of Advanced Metering Infrastructure among residential energy customers has served as a boon for energy systems research relying on granular consumption data. Residential Demand Response aims to utilize the flexibility of consumers to reduce their energy usage during times when the grid is strained. Suitable incentive mechanisms to encourage customers to deviate from their usual behavior have to be implemented to correctly control the bids into the wholesale electricity market as a Demand Response provider. In this paper, we present a framework for short term load forecasting on an individual user level, and relate nonexperimental estimates of Demand Response efficacy, i.e. the estimated reduction of consumption during Demand Response events, to the variability of user consumption. We apply our framework on a data set from a residential Demand Response program in the Western United States. Our results suggest that users with more variable consumption patterns are more likely to reduce their consumption compared to users with a more regular consumption behavior.
2404.00357
Tao Li
Tao Li, Qinghua Tao, Weihao Yan, Zehao Lei, Yingwen Wu, Kun Fang, Mingzhen He, Xiaolin Huang
Revisiting Random Weight Perturbation for Efficiently Improving Generalization
Accepted to TMLR 2024
null
null
null
cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Improving the generalization ability of modern deep neural networks (DNNs) is a fundamental challenge in machine learning. Two branches of methods have been proposed to seek flat minima and improve generalization: one led by sharpness-aware minimization (SAM) minimizes the worst-case neighborhood loss through adversarial weight perturbation (AWP), and the other minimizes the expected Bayes objective with random weight perturbation (RWP). While RWP offers advantages in computation and is closely linked to AWP on a mathematical basis, its empirical performance has consistently lagged behind that of AWP. In this paper, we revisit the use of RWP for improving generalization and propose improvements from two perspectives: i) the trade-off between generalization and convergence and ii) the random perturbation generation. Through extensive experimental evaluations, we demonstrate that our enhanced RWP methods achieve greater efficiency in enhancing generalization, particularly in large-scale problems, while also offering comparable or even superior performance to SAM. The code is released at https://github.com/nblt/mARWP.
[ { "created": "Sat, 30 Mar 2024 13:18:27 GMT", "version": "v1" } ]
2024-04-02
[ [ "Li", "Tao", "" ], [ "Tao", "Qinghua", "" ], [ "Yan", "Weihao", "" ], [ "Lei", "Zehao", "" ], [ "Wu", "Yingwen", "" ], [ "Fang", "Kun", "" ], [ "He", "Mingzhen", "" ], [ "Huang", "Xiaolin", "" ] ]
Improving the generalization ability of modern deep neural networks (DNNs) is a fundamental challenge in machine learning. Two branches of methods have been proposed to seek flat minima and improve generalization: one led by sharpness-aware minimization (SAM) minimizes the worst-case neighborhood loss through adversarial weight perturbation (AWP), and the other minimizes the expected Bayes objective with random weight perturbation (RWP). While RWP offers advantages in computation and is closely linked to AWP on a mathematical basis, its empirical performance has consistently lagged behind that of AWP. In this paper, we revisit the use of RWP for improving generalization and propose improvements from two perspectives: i) the trade-off between generalization and convergence and ii) the random perturbation generation. Through extensive experimental evaluations, we demonstrate that our enhanced RWP methods achieve greater efficiency in enhancing generalization, particularly in large-scale problems, while also offering comparable or even superior performance to SAM. The code is released at https://github.com/nblt/mARWP.
2312.10714
Siqi Liu
Siqi Liu, Yong-Lu Li, Zhou Fang, Xinpeng Liu, Yang You, Cewu Lu
Primitive-based 3D Human-Object Interaction Modelling and Programming
AAAI2024
null
null
null
cs.CV
http://creativecommons.org/licenses/by-nc-nd/4.0/
Embedding Human and Articulated Object Interaction (HAOI) in 3D is an important direction for a deeper human activity understanding. Different from previous works that use parametric and CAD models to represent humans and objects, in this work, we propose a novel 3D geometric primitive-based language to encode both humans and objects. Given our new paradigm, humans and objects are all compositions of primitives instead of heterogeneous entities. Thus, mutual information learning may be achieved between the limited 3D data of humans and different object categories. Moreover, considering the simplicity of the expression and the richness of the information it contains, we choose the superquadric as the primitive representation. To explore an effective embedding of HAOI for the machine, we build a new benchmark on 3D HAOI consisting of primitives together with their images and propose a task requiring machines to recover 3D HAOI using primitives from images. Moreover, we propose a baseline of single-view 3D reconstruction on HAOI. We believe this primitive-based 3D HAOI representation would pave the way for 3D HAOI studies. Our code and data are available at https://mvig-rhos.com/p3haoi.
[ { "created": "Sun, 17 Dec 2023 13:16:49 GMT", "version": "v1" } ]
2023-12-19
[ [ "Liu", "Siqi", "" ], [ "Li", "Yong-Lu", "" ], [ "Fang", "Zhou", "" ], [ "Liu", "Xinpeng", "" ], [ "You", "Yang", "" ], [ "Lu", "Cewu", "" ] ]
Embedding Human and Articulated Object Interaction (HAOI) in 3D is an important direction for a deeper human activity understanding. Different from previous works that use parametric and CAD models to represent humans and objects, in this work, we propose a novel 3D geometric primitive-based language to encode both humans and objects. Given our new paradigm, humans and objects are all compositions of primitives instead of heterogeneous entities. Thus, mutual information learning may be achieved between the limited 3D data of humans and different object categories. Moreover, considering the simplicity of the expression and the richness of the information it contains, we choose the superquadric as the primitive representation. To explore an effective embedding of HAOI for the machine, we build a new benchmark on 3D HAOI consisting of primitives together with their images and propose a task requiring machines to recover 3D HAOI using primitives from images. Moreover, we propose a baseline of single-view 3D reconstruction on HAOI. We believe this primitive-based 3D HAOI representation would pave the way for 3D HAOI studies. Our code and data are available at https://mvig-rhos.com/p3haoi.
2206.04935
Max M\"uller-Eberstein
Max M\"uller-Eberstein, Rob van der Goot and Barbara Plank
Sort by Structure: Language Model Ranking as Dependency Probing
Accepted at NAACL 2022 (Main Conference)
null
null
null
cs.CL
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Making an informed choice of pre-trained language model (LM) is critical for performance, yet environmentally costly, and as such widely underexplored. The field of Computer Vision has begun to tackle encoder ranking, with promising forays into Natural Language Processing, however they lack coverage of linguistic tasks such as structured prediction. We propose probing to rank LMs, specifically for parsing dependencies in a given language, by measuring the degree to which labeled trees are recoverable from an LM's contextualized embeddings. Across 46 typologically and architecturally diverse LM-language pairs, our probing approach predicts the best LM choice 79% of the time using orders of magnitude less compute than training a full parser. Within this study, we identify and analyze one recently proposed decoupled LM - RemBERT - and find it strikingly contains less inherent dependency information, but often yields the best parser after full fine-tuning. Without this outlier our approach identifies the best LM in 89% of cases.
[ { "created": "Fri, 10 Jun 2022 08:10:29 GMT", "version": "v1" } ]
2022-06-13
[ [ "Müller-Eberstein", "Max", "" ], [ "van der Goot", "Rob", "" ], [ "Plank", "Barbara", "" ] ]
Making an informed choice of pre-trained language model (LM) is critical for performance, yet environmentally costly, and as such widely underexplored. The field of Computer Vision has begun to tackle encoder ranking, with promising forays into Natural Language Processing, however they lack coverage of linguistic tasks such as structured prediction. We propose probing to rank LMs, specifically for parsing dependencies in a given language, by measuring the degree to which labeled trees are recoverable from an LM's contextualized embeddings. Across 46 typologically and architecturally diverse LM-language pairs, our probing approach predicts the best LM choice 79% of the time using orders of magnitude less compute than training a full parser. Within this study, we identify and analyze one recently proposed decoupled LM - RemBERT - and find it strikingly contains less inherent dependency information, but often yields the best parser after full fine-tuning. Without this outlier our approach identifies the best LM in 89% of cases.
2403.18555
Philip Kenneweg
Philip Kenneweg, Sarah Schr\"oder, Alexander Schulz, Barbara Hammer
Debiasing Sentence Embedders through Contrastive Word Pairs
null
null
10.5220/0011615300003411
null
cs.CL
http://creativecommons.org/publicdomain/zero/1.0/
Over the last years, various sentence embedders have been an integral part in the success of current machine learning approaches to Natural Language Processing (NLP). Unfortunately, multiple sources have shown that the bias, inherent in the datasets upon which these embedding methods are trained, is learned by them. A variety of different approaches to remove biases in embeddings exists in the literature. Most of these approaches are applicable to word embeddings and in fewer cases to sentence embeddings. It is problematic that most debiasing approaches are directly transferred from word embeddings, therefore these approaches fail to take into account the nonlinear nature of sentence embedders and the embeddings they produce. It has been shown in literature that bias information is still present if sentence embeddings are debiased using such methods. In this contribution, we explore an approach to remove linear and nonlinear bias information for NLP solutions, without impacting downstream performance. We compare our approach to common debiasing methods on classical bias metrics and on bias metrics which take nonlinear information into account.
[ { "created": "Wed, 27 Mar 2024 13:34:59 GMT", "version": "v1" } ]
2024-03-28
[ [ "Kenneweg", "Philip", "" ], [ "Schröder", "Sarah", "" ], [ "Schulz", "Alexander", "" ], [ "Hammer", "Barbara", "" ] ]
Over the last years, various sentence embedders have been an integral part in the success of current machine learning approaches to Natural Language Processing (NLP). Unfortunately, multiple sources have shown that the bias, inherent in the datasets upon which these embedding methods are trained, is learned by them. A variety of different approaches to remove biases in embeddings exists in the literature. Most of these approaches are applicable to word embeddings and in fewer cases to sentence embeddings. It is problematic that most debiasing approaches are directly transferred from word embeddings, therefore these approaches fail to take into account the nonlinear nature of sentence embedders and the embeddings they produce. It has been shown in literature that bias information is still present if sentence embeddings are debiased using such methods. In this contribution, we explore an approach to remove linear and nonlinear bias information for NLP solutions, without impacting downstream performance. We compare our approach to common debiasing methods on classical bias metrics and on bias metrics which take nonlinear information into account.
2309.02011
Pascal Mattia Esser
Pascal Esser, Satyaki Mukherjee, Debarghya Ghoshdastidar
Representation Learning Dynamics of Self-Supervised Models
null
null
null
null
cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Self-Supervised Learning (SSL) is an important paradigm for learning representations from unlabelled data, and SSL with neural networks has been highly successful in practice. However current theoretical analysis of SSL is mostly restricted to generalisation error bounds. In contrast, learning dynamics often provide a precise characterisation of the behaviour of neural networks based models but, so far, are mainly known in supervised settings. In this paper, we study the learning dynamics of SSL models, specifically representations obtained by minimising contrastive and non-contrastive losses. We show that a naive extension of the dymanics of multivariate regression to SSL leads to learning trivial scalar representations that demonstrates dimension collapse in SSL. Consequently, we formulate SSL objectives with orthogonality constraints on the weights, and derive the exact (network width independent) learning dynamics of the SSL models trained using gradient descent on the Grassmannian manifold. We also argue that the infinite width approximation of SSL models significantly deviate from the neural tangent kernel approximations of supervised models. We numerically illustrate the validity of our theoretical findings, and discuss how the presented results provide a framework for further theoretical analysis of contrastive and non-contrastive SSL.
[ { "created": "Tue, 5 Sep 2023 07:48:45 GMT", "version": "v1" } ]
2023-09-06
[ [ "Esser", "Pascal", "" ], [ "Mukherjee", "Satyaki", "" ], [ "Ghoshdastidar", "Debarghya", "" ] ]
Self-Supervised Learning (SSL) is an important paradigm for learning representations from unlabelled data, and SSL with neural networks has been highly successful in practice. However current theoretical analysis of SSL is mostly restricted to generalisation error bounds. In contrast, learning dynamics often provide a precise characterisation of the behaviour of neural networks based models but, so far, are mainly known in supervised settings. In this paper, we study the learning dynamics of SSL models, specifically representations obtained by minimising contrastive and non-contrastive losses. We show that a naive extension of the dymanics of multivariate regression to SSL leads to learning trivial scalar representations that demonstrates dimension collapse in SSL. Consequently, we formulate SSL objectives with orthogonality constraints on the weights, and derive the exact (network width independent) learning dynamics of the SSL models trained using gradient descent on the Grassmannian manifold. We also argue that the infinite width approximation of SSL models significantly deviate from the neural tangent kernel approximations of supervised models. We numerically illustrate the validity of our theoretical findings, and discuss how the presented results provide a framework for further theoretical analysis of contrastive and non-contrastive SSL.
2408.07966
Shunxin Guo
Shunxin Guo, Hongsong Wang, Shuxia Lin, Zhiqiang Kou, Xin Geng
Addressing Skewed Heterogeneity via Federated Prototype Rectification with Personalization
null
null
null
null
cs.LG cs.DC
http://creativecommons.org/licenses/by/4.0/
Federated learning is an efficient framework designed to facilitate collaborative model training across multiple distributed devices while preserving user data privacy. A significant challenge of federated learning is data-level heterogeneity, i.e., skewed or long-tailed distribution of private data. Although various methods have been proposed to address this challenge, most of them assume that the underlying global data is uniformly distributed across all clients. This paper investigates data-level heterogeneity federated learning with a brief review and redefines a more practical and challenging setting called Skewed Heterogeneous Federated Learning (SHFL). Accordingly, we propose a novel Federated Prototype Rectification with Personalization which consists of two parts: Federated Personalization and Federated Prototype Rectification. The former aims to construct balanced decision boundaries between dominant and minority classes based on private data, while the latter exploits both inter-class discrimination and intra-class consistency to rectify empirical prototypes. Experiments on three popular benchmarks show that the proposed approach outperforms current state-of-the-art methods and achieves balanced performance in both personalization and generalization.
[ { "created": "Thu, 15 Aug 2024 06:26:46 GMT", "version": "v1" } ]
2024-08-16
[ [ "Guo", "Shunxin", "" ], [ "Wang", "Hongsong", "" ], [ "Lin", "Shuxia", "" ], [ "Kou", "Zhiqiang", "" ], [ "Geng", "Xin", "" ] ]
Federated learning is an efficient framework designed to facilitate collaborative model training across multiple distributed devices while preserving user data privacy. A significant challenge of federated learning is data-level heterogeneity, i.e., skewed or long-tailed distribution of private data. Although various methods have been proposed to address this challenge, most of them assume that the underlying global data is uniformly distributed across all clients. This paper investigates data-level heterogeneity federated learning with a brief review and redefines a more practical and challenging setting called Skewed Heterogeneous Federated Learning (SHFL). Accordingly, we propose a novel Federated Prototype Rectification with Personalization which consists of two parts: Federated Personalization and Federated Prototype Rectification. The former aims to construct balanced decision boundaries between dominant and minority classes based on private data, while the latter exploits both inter-class discrimination and intra-class consistency to rectify empirical prototypes. Experiments on three popular benchmarks show that the proposed approach outperforms current state-of-the-art methods and achieves balanced performance in both personalization and generalization.
2309.12708
Yuxiang Yan
Yuxiang Yan, Boda Liu, Jianfei Ai, Qinbu Li, Ru Wan, Jian Pu
PointSSC: A Cooperative Vehicle-Infrastructure Point Cloud Benchmark for Semantic Scene Completion
ICRA2024, oral & poster
null
null
null
cs.CV cs.AI cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Semantic Scene Completion (SSC) aims to jointly generate space occupancies and semantic labels for complex 3D scenes. Most existing SSC models focus on volumetric representations, which are memory-inefficient for large outdoor spaces. Point clouds provide a lightweight alternative but existing benchmarks lack outdoor point cloud scenes with semantic labels. To address this, we introduce PointSSC, the first cooperative vehicle-infrastructure point cloud benchmark for semantic scene completion. These scenes exhibit long-range perception and minimal occlusion. We develop an automated annotation pipeline leveraging Semantic Segment Anything to efficiently assign semantics. To benchmark progress, we propose a LiDAR-based model with a Spatial-Aware Transformer for global and local feature extraction and a Completion and Segmentation Cooperative Module for joint completion and segmentation. PointSSC provides a challenging testbed to drive advances in semantic point cloud completion for real-world navigation. The code and datasets are available at https://github.com/yyxssm/PointSSC.
[ { "created": "Fri, 22 Sep 2023 08:39:16 GMT", "version": "v1" }, { "created": "Thu, 7 Mar 2024 02:50:04 GMT", "version": "v2" } ]
2024-03-08
[ [ "Yan", "Yuxiang", "" ], [ "Liu", "Boda", "" ], [ "Ai", "Jianfei", "" ], [ "Li", "Qinbu", "" ], [ "Wan", "Ru", "" ], [ "Pu", "Jian", "" ] ]
Semantic Scene Completion (SSC) aims to jointly generate space occupancies and semantic labels for complex 3D scenes. Most existing SSC models focus on volumetric representations, which are memory-inefficient for large outdoor spaces. Point clouds provide a lightweight alternative but existing benchmarks lack outdoor point cloud scenes with semantic labels. To address this, we introduce PointSSC, the first cooperative vehicle-infrastructure point cloud benchmark for semantic scene completion. These scenes exhibit long-range perception and minimal occlusion. We develop an automated annotation pipeline leveraging Semantic Segment Anything to efficiently assign semantics. To benchmark progress, we propose a LiDAR-based model with a Spatial-Aware Transformer for global and local feature extraction and a Completion and Segmentation Cooperative Module for joint completion and segmentation. PointSSC provides a challenging testbed to drive advances in semantic point cloud completion for real-world navigation. The code and datasets are available at https://github.com/yyxssm/PointSSC.
1508.05189
Hartmut Klauck
Ralph C. Bottesch, Dmitry Gavinsky, Hartmut Klauck
Correlation in Hard Distributions in Communication Complexity
null
null
null
null
cs.CC quant-ph
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We study the effect that the amount of correlation in a bipartite distribution has on the communication complexity of a problem under that distribution. We introduce a new family of complexity measures that interpolates between the two previously studied extreme cases: the (standard) randomised communication complexity and the case of distributional complexity under product distributions. We give a tight characterisation of the randomised complexity of Disjointness under distributions with mutual information $k$, showing that it is $\Theta(\sqrt{n(k+1)})$ for all $0\leq k\leq n$. This smoothly interpolates between the lower bounds of Babai, Frankl and Simon for the product distribution case ($k=0$), and the bound of Razborov for the randomised case. The upper bounds improve and generalise what was known for product distributions, and imply that any tight bound for Disjointness needs $\Omega(n)$ bits of mutual information in the corresponding distribution. We study the same question in the distributional quantum setting, and show a lower bound of $\Omega((n(k+1))^{1/4})$, and an upper bound, matching up to a logarithmic factor. We show that there are total Boolean functions $f_d$ on $2n$ inputs that have distributional communication complexity $O(\log n)$ under all distributions of information up to $o(n)$, while the (interactive) distributional complexity maximised over all distributions is $\Theta(\log d)$ for $6n\leq d\leq 2^{n/100}$. We show that in the setting of one-way communication under product distributions, the dependence of communication cost on the allowed error $\epsilon$ is multiplicative in $\log(1/\epsilon)$ -- the previous upper bounds had the dependence of more than $1/\epsilon$.
[ { "created": "Fri, 21 Aug 2015 07:05:39 GMT", "version": "v1" } ]
2015-08-25
[ [ "Bottesch", "Ralph C.", "" ], [ "Gavinsky", "Dmitry", "" ], [ "Klauck", "Hartmut", "" ] ]
We study the effect that the amount of correlation in a bipartite distribution has on the communication complexity of a problem under that distribution. We introduce a new family of complexity measures that interpolates between the two previously studied extreme cases: the (standard) randomised communication complexity and the case of distributional complexity under product distributions. We give a tight characterisation of the randomised complexity of Disjointness under distributions with mutual information $k$, showing that it is $\Theta(\sqrt{n(k+1)})$ for all $0\leq k\leq n$. This smoothly interpolates between the lower bounds of Babai, Frankl and Simon for the product distribution case ($k=0$), and the bound of Razborov for the randomised case. The upper bounds improve and generalise what was known for product distributions, and imply that any tight bound for Disjointness needs $\Omega(n)$ bits of mutual information in the corresponding distribution. We study the same question in the distributional quantum setting, and show a lower bound of $\Omega((n(k+1))^{1/4})$, and an upper bound, matching up to a logarithmic factor. We show that there are total Boolean functions $f_d$ on $2n$ inputs that have distributional communication complexity $O(\log n)$ under all distributions of information up to $o(n)$, while the (interactive) distributional complexity maximised over all distributions is $\Theta(\log d)$ for $6n\leq d\leq 2^{n/100}$. We show that in the setting of one-way communication under product distributions, the dependence of communication cost on the allowed error $\epsilon$ is multiplicative in $\log(1/\epsilon)$ -- the previous upper bounds had the dependence of more than $1/\epsilon$.
1802.06157
Matthieu Lequesne
Matthieu Lequesne, Jean-Pierre Tillich
Attack on the Edon-K Key Encapsulation Mechanism
Submitted to ISIT 2018
null
null
null
cs.CR
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
The key encapsulation mechanism Edon-K was proposed in response to the call for post-quantum cryptography standardization issued by the National Institute of Standards and Technologies (NIST). This scheme is inspired by the McEliece scheme but uses another family of codes defined over $\mathbb{F}_{2^{128}}$ instead of $\mathbb{F}_2$ and is not based on the Hamming metric. It allows significantly shorter public keys than the McEliece scheme. In this paper, we give a polynomial time algorithm that recovers the encapsulated secret. This attack makes the scheme insecure for the intended use. We obtain this result by observing that recovering the error in the McEliece scheme corresponding to Edon-K can be viewed as a decoding problem for the rank-metric. We show that the code used in Edon-K is in fact a super-code of a Low Rank Parity Check (LRPC) code of very small rank (1 or 2). A suitable parity-check matrix for the super-code of such low rank can be easily derived from for the public key. We then use this parity-check matrix in a decoding algorithm that was devised for LRPC codes to recover the error. Finally we explain how we decapsulate the secret once we have found the error.
[ { "created": "Fri, 16 Feb 2018 23:06:50 GMT", "version": "v1" } ]
2018-02-20
[ [ "Lequesne", "Matthieu", "" ], [ "Tillich", "Jean-Pierre", "" ] ]
The key encapsulation mechanism Edon-K was proposed in response to the call for post-quantum cryptography standardization issued by the National Institute of Standards and Technologies (NIST). This scheme is inspired by the McEliece scheme but uses another family of codes defined over $\mathbb{F}_{2^{128}}$ instead of $\mathbb{F}_2$ and is not based on the Hamming metric. It allows significantly shorter public keys than the McEliece scheme. In this paper, we give a polynomial time algorithm that recovers the encapsulated secret. This attack makes the scheme insecure for the intended use. We obtain this result by observing that recovering the error in the McEliece scheme corresponding to Edon-K can be viewed as a decoding problem for the rank-metric. We show that the code used in Edon-K is in fact a super-code of a Low Rank Parity Check (LRPC) code of very small rank (1 or 2). A suitable parity-check matrix for the super-code of such low rank can be easily derived from for the public key. We then use this parity-check matrix in a decoding algorithm that was devised for LRPC codes to recover the error. Finally we explain how we decapsulate the secret once we have found the error.
cs/0412095
Mirela Damian
Mirela Damian and Joseph O'Rourke
Partitioning Regular Polygons into Circular Pieces II:Nonconvex Partitions
13 pages, 11 figures
null
null
null
cs.CG cs.DM
null
We explore optimal circular nonconvex partitions of regular k-gons. The circularity of a polygon is measured by its aspect ratio: the ratio of the radii of the smallest circumscribing circle to the largest inscribed disk. An optimal circular partition minimizes the maximum ratio over all pieces in the partition. We show that the equilateral triangle has an optimal 4-piece nonconvex partition, the square an optimal 13-piece nonconvex partition, and the pentagon has an optimal nonconvex partition with more than 20 thousand pieces. For hexagons and beyond, we provide a general algorithm that approaches optimality, but does not achieve it.
[ { "created": "Tue, 21 Dec 2004 05:41:11 GMT", "version": "v1" } ]
2007-05-23
[ [ "Damian", "Mirela", "" ], [ "O'Rourke", "Joseph", "" ] ]
We explore optimal circular nonconvex partitions of regular k-gons. The circularity of a polygon is measured by its aspect ratio: the ratio of the radii of the smallest circumscribing circle to the largest inscribed disk. An optimal circular partition minimizes the maximum ratio over all pieces in the partition. We show that the equilateral triangle has an optimal 4-piece nonconvex partition, the square an optimal 13-piece nonconvex partition, and the pentagon has an optimal nonconvex partition with more than 20 thousand pieces. For hexagons and beyond, we provide a general algorithm that approaches optimality, but does not achieve it.
2109.10750
Hongbo Zhang
Hongbo Zhang, Yunshuang Li, Yipin Guo, Xinyi Chen, Qinyuan Ren
Control of Pneumatic Artificial Muscles with SNN-based Cerebellar-like Model
null
null
null
null
cs.RO
http://creativecommons.org/licenses/by/4.0/
Soft robotics technologies have gained growing interest in recent years, which allows various applications from manufacturing to human-robot interaction. Pneumatic artificial muscle (PAM), a typical soft actuator, has been widely applied to soft robots. The compliance and resilience of soft actuators allow soft robots to behave compliant when interacting with unstructured environments, while the utilization of soft actuators also introduces nonlinearity and uncertainty. Inspired by Cerebellum's vital functions in control of human's physical movement, a neural network model of Cerebellum based on spiking neuron networks (SNNs) is designed. This model is used as a feed-forward controller in controlling a 1-DOF robot arm driven by PAMs. The simulation results show that this Cerebellar-based system achieves good performance and increases the system's response.
[ { "created": "Wed, 22 Sep 2021 14:11:05 GMT", "version": "v1" } ]
2021-09-23
[ [ "Zhang", "Hongbo", "" ], [ "Li", "Yunshuang", "" ], [ "Guo", "Yipin", "" ], [ "Chen", "Xinyi", "" ], [ "Ren", "Qinyuan", "" ] ]
Soft robotics technologies have gained growing interest in recent years, which allows various applications from manufacturing to human-robot interaction. Pneumatic artificial muscle (PAM), a typical soft actuator, has been widely applied to soft robots. The compliance and resilience of soft actuators allow soft robots to behave compliant when interacting with unstructured environments, while the utilization of soft actuators also introduces nonlinearity and uncertainty. Inspired by Cerebellum's vital functions in control of human's physical movement, a neural network model of Cerebellum based on spiking neuron networks (SNNs) is designed. This model is used as a feed-forward controller in controlling a 1-DOF robot arm driven by PAMs. The simulation results show that this Cerebellar-based system achieves good performance and increases the system's response.
cs/0511016
Santo Fortunato Dr
Santo Fortunato, Marian Boguna, Alessandro Flammini, Filippo Menczer
How to make the top ten: Approximating PageRank from in-degree
8 pages, 7 figures, 2 tables
null
null
null
cs.IR physics.soc-ph
null
PageRank has become a key element in the success of search engines, allowing to rank the most important hits in the top screen of results. One key aspect that distinguishes PageRank from other prestige measures such as in-degree is its global nature. From the information provider perspective, this makes it difficult or impossible to predict how their pages will be ranked. Consequently a market has emerged for the optimization of search engine results. Here we study the accuracy with which PageRank can be approximated by in-degree, a local measure made freely available by search engines. Theoretical and empirical analyses lead to conclude that given the weak degree correlations in the Web link graph, the approximation can be relatively accurate, giving service and information providers an effective new marketing tool.
[ { "created": "Thu, 3 Nov 2005 23:01:50 GMT", "version": "v1" } ]
2007-05-23
[ [ "Fortunato", "Santo", "" ], [ "Boguna", "Marian", "" ], [ "Flammini", "Alessandro", "" ], [ "Menczer", "Filippo", "" ] ]
PageRank has become a key element in the success of search engines, allowing to rank the most important hits in the top screen of results. One key aspect that distinguishes PageRank from other prestige measures such as in-degree is its global nature. From the information provider perspective, this makes it difficult or impossible to predict how their pages will be ranked. Consequently a market has emerged for the optimization of search engine results. Here we study the accuracy with which PageRank can be approximated by in-degree, a local measure made freely available by search engines. Theoretical and empirical analyses lead to conclude that given the weak degree correlations in the Web link graph, the approximation can be relatively accurate, giving service and information providers an effective new marketing tool.
2105.03137
Karl-Ludwig Besser
Eduard Jorswieck, Andrew Lonnstrom, Karl-Ludwig Besser, Stefan Rothe, Juergen W. Czarske
Achievable Physical-Layer Secrecy in Multi-Mode Fiber Channels using Artificial Noise
5 pages, 2 figures
null
10.1109/ISWCS49558.2021.9562176
null
cs.IT math.IT
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Reliable and secure communication is an important aspect of modern fiber optic communication. In this work we consider a multi-mode fiber (MMF) channel wiretapped by an eavesdropper. We assume the transmitter knows the legitimate channel, but statistical knowledge of the eavesdropper's channel only. We propose a transmission scheme with artificial noise (AN) for such a channel. In particular, we formulate the corresponding optimization problem which aims to maximize the average secrecy rate and develop an algorithm to solve it. We apply this algorithm to actual measured MMF channels. As real fiber measurements show, for a 55 mode MMF we can achieve positive average secrecy rates with the proper use of AN. Furthermore, the gain compared to standard precoding and power allocation schemes is illustrated.
[ { "created": "Fri, 7 May 2021 09:29:25 GMT", "version": "v1" } ]
2021-10-22
[ [ "Jorswieck", "Eduard", "" ], [ "Lonnstrom", "Andrew", "" ], [ "Besser", "Karl-Ludwig", "" ], [ "Rothe", "Stefan", "" ], [ "Czarske", "Juergen W.", "" ] ]
Reliable and secure communication is an important aspect of modern fiber optic communication. In this work we consider a multi-mode fiber (MMF) channel wiretapped by an eavesdropper. We assume the transmitter knows the legitimate channel, but statistical knowledge of the eavesdropper's channel only. We propose a transmission scheme with artificial noise (AN) for such a channel. In particular, we formulate the corresponding optimization problem which aims to maximize the average secrecy rate and develop an algorithm to solve it. We apply this algorithm to actual measured MMF channels. As real fiber measurements show, for a 55 mode MMF we can achieve positive average secrecy rates with the proper use of AN. Furthermore, the gain compared to standard precoding and power allocation schemes is illustrated.
1602.05568
Mohammad Taha Bahadori
Edward Choi, Mohammad Taha Bahadori, Elizabeth Searles, Catherine Coffey, Jimeng Sun
Multi-layer Representation Learning for Medical Concepts
null
null
null
null
cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Learning efficient representations for concepts has been proven to be an important basis for many applications such as machine translation or document classification. Proper representations of medical concepts such as diagnosis, medication, procedure codes and visits will have broad applications in healthcare analytics. However, in Electronic Health Records (EHR) the visit sequences of patients include multiple concepts (diagnosis, procedure, and medication codes) per visit. This structure provides two types of relational information, namely sequential order of visits and co-occurrence of the codes within each visit. In this work, we propose Med2Vec, which not only learns distributed representations for both medical codes and visits from a large EHR dataset with over 3 million visits, but also allows us to interpret the learned representations confirmed positively by clinical experts. In the experiments, Med2Vec displays significant improvement in key medical applications compared to popular baselines such as Skip-gram, GloVe and stacked autoencoder, while providing clinically meaningful interpretation.
[ { "created": "Wed, 17 Feb 2016 20:55:40 GMT", "version": "v1" } ]
2016-02-18
[ [ "Choi", "Edward", "" ], [ "Bahadori", "Mohammad Taha", "" ], [ "Searles", "Elizabeth", "" ], [ "Coffey", "Catherine", "" ], [ "Sun", "Jimeng", "" ] ]
Learning efficient representations for concepts has been proven to be an important basis for many applications such as machine translation or document classification. Proper representations of medical concepts such as diagnosis, medication, procedure codes and visits will have broad applications in healthcare analytics. However, in Electronic Health Records (EHR) the visit sequences of patients include multiple concepts (diagnosis, procedure, and medication codes) per visit. This structure provides two types of relational information, namely sequential order of visits and co-occurrence of the codes within each visit. In this work, we propose Med2Vec, which not only learns distributed representations for both medical codes and visits from a large EHR dataset with over 3 million visits, but also allows us to interpret the learned representations confirmed positively by clinical experts. In the experiments, Med2Vec displays significant improvement in key medical applications compared to popular baselines such as Skip-gram, GloVe and stacked autoencoder, while providing clinically meaningful interpretation.
2309.08776
Josselin Somerville Roberts
Josselin Somerville Roberts, Julia Di
Projected Task-Specific Layers for Multi-Task Reinforcement Learning
null
ICRA 2024
null
null
cs.LG cs.AI cs.RO
http://creativecommons.org/licenses/by/4.0/
Multi-task reinforcement learning could enable robots to scale across a wide variety of manipulation tasks in homes and workplaces. However, generalizing from one task to another and mitigating negative task interference still remains a challenge. Addressing this challenge by successfully sharing information across tasks will depend on how well the structure underlying the tasks is captured. In this work, we introduce our new architecture, Projected Task-Specific Layers (PTSL), that leverages a common policy with dense task-specific corrections through task-specific layers to better express shared and variable task information. We then show that our model outperforms the state of the art on the MT10 and MT50 benchmarks of Meta-World consisting of 10 and 50 goal-conditioned tasks for a Sawyer arm.
[ { "created": "Fri, 15 Sep 2023 21:42:06 GMT", "version": "v1" }, { "created": "Wed, 6 Mar 2024 18:51:45 GMT", "version": "v2" } ]
2024-03-07
[ [ "Roberts", "Josselin Somerville", "" ], [ "Di", "Julia", "" ] ]
Multi-task reinforcement learning could enable robots to scale across a wide variety of manipulation tasks in homes and workplaces. However, generalizing from one task to another and mitigating negative task interference still remains a challenge. Addressing this challenge by successfully sharing information across tasks will depend on how well the structure underlying the tasks is captured. In this work, we introduce our new architecture, Projected Task-Specific Layers (PTSL), that leverages a common policy with dense task-specific corrections through task-specific layers to better express shared and variable task information. We then show that our model outperforms the state of the art on the MT10 and MT50 benchmarks of Meta-World consisting of 10 and 50 goal-conditioned tasks for a Sawyer arm.
2308.06672
Yong Wang
Yong Wang and Yanzhong Yao and Jiawei Guo and Zhiming Gao
A practical PINN framework for multi-scale problems with multi-magnitude loss terms
null
null
null
null
cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
For multi-scale problems, the conventional physics-informed neural networks (PINNs) face some challenges in obtaining available predictions. In this paper, based on PINNs, we propose a practical deep learning framework for multi-scale problems by reconstructing the loss function and associating it with special neural network architectures. New PINN methods derived from the improved PINN framework differ from the conventional PINN method mainly in two aspects. First, the new methods use a novel loss function by modifying the standard loss function through a (grouping) regularization strategy. The regularization strategy implements a different power operation on each loss term so that all loss terms composing the loss function are of approximately the same order of magnitude, which makes all loss terms be optimized synchronously during the optimization process. Second, for the multi-frequency or high-frequency problems, in addition to using the modified loss function, new methods upgrade the neural network architecture from the common fully-connected neural network to special network architectures such as the Fourier feature architecture, and the integrated architecture developed by us. The combination of the above two techniques leads to a significant improvement in the computational accuracy of multi-scale problems. Several challenging numerical examples demonstrate the effectiveness of the proposed methods. The proposed methods not only significantly outperform the conventional PINN method in terms of computational efficiency and computational accuracy, but also compare favorably with the state-of-the-art methods in the recent literature. The improved PINN framework facilitates better application of PINNs to multi-scale problems.
[ { "created": "Sun, 13 Aug 2023 03:26:01 GMT", "version": "v1" }, { "created": "Sun, 29 Oct 2023 16:22:20 GMT", "version": "v2" } ]
2023-10-31
[ [ "Wang", "Yong", "" ], [ "Yao", "Yanzhong", "" ], [ "Guo", "Jiawei", "" ], [ "Gao", "Zhiming", "" ] ]
For multi-scale problems, the conventional physics-informed neural networks (PINNs) face some challenges in obtaining available predictions. In this paper, based on PINNs, we propose a practical deep learning framework for multi-scale problems by reconstructing the loss function and associating it with special neural network architectures. New PINN methods derived from the improved PINN framework differ from the conventional PINN method mainly in two aspects. First, the new methods use a novel loss function by modifying the standard loss function through a (grouping) regularization strategy. The regularization strategy implements a different power operation on each loss term so that all loss terms composing the loss function are of approximately the same order of magnitude, which makes all loss terms be optimized synchronously during the optimization process. Second, for the multi-frequency or high-frequency problems, in addition to using the modified loss function, new methods upgrade the neural network architecture from the common fully-connected neural network to special network architectures such as the Fourier feature architecture, and the integrated architecture developed by us. The combination of the above two techniques leads to a significant improvement in the computational accuracy of multi-scale problems. Several challenging numerical examples demonstrate the effectiveness of the proposed methods. The proposed methods not only significantly outperform the conventional PINN method in terms of computational efficiency and computational accuracy, but also compare favorably with the state-of-the-art methods in the recent literature. The improved PINN framework facilitates better application of PINNs to multi-scale problems.
1106.3858
Yi Gai
Yi Gai, Hua Liu, and Bhaskar Krishnamachari
A Packet Dropping Mechanism for Efficient Operation of M/M/1 Queues with Selfish Users
This work is an extended version of the conference paper: Y. Gai, H. Liu and B. Krishnamachari, "A packet dropping-based incentive mechanism for M/M/1 queues with selfish users", the 30th IEEE International Conference on Computer Communications (IEEE INFOCOM 2011), China, April, 2011
null
10.1016/j.comnet.2015.12.009
null
cs.GT cs.NI math.OC
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We consider a fundamental game theoretic problem concerning selfish users contributing packets to an M/M/1 queue. In this game, each user controls its own input rate so as to optimize a desired tradeoff between throughput and delay. We first show that the original game has an inefficient Nash Equilibrium (NE), with a Price of Anarchy (PoA) that scales linearly or worse in the number of users. In order to improve the outcome efficiency, we propose an easily implementable mechanism design whereby the server randomly drops packets with a probability that is a function of the total arrival rate. We show that this results in a modified M/M/1 queueing game that is an ordinal potential game with at least one NE. In particular, for a linear packet dropping function, which is similar to the Random Early Detection (RED) algorithm used in Internet Congestion Control, we prove that there is a unique NE. We also show that the simple best response dynamic converges to this unique equilibrium. Finally, for this scheme, we prove that the social welfare (expressed either as the summation of utilities of all players, or as the summation of the logarithm of utilities of all players) at the equilibrium point can be arbitrarily close to the social welfare at the global optimal point, i.e. the PoA can be made arbitrarily close to 1. We also study the impact of arrival rate estimation error on the PoA through simulations.
[ { "created": "Mon, 20 Jun 2011 10:40:21 GMT", "version": "v1" } ]
2016-11-17
[ [ "Gai", "Yi", "" ], [ "Liu", "Hua", "" ], [ "Krishnamachari", "Bhaskar", "" ] ]
We consider a fundamental game theoretic problem concerning selfish users contributing packets to an M/M/1 queue. In this game, each user controls its own input rate so as to optimize a desired tradeoff between throughput and delay. We first show that the original game has an inefficient Nash Equilibrium (NE), with a Price of Anarchy (PoA) that scales linearly or worse in the number of users. In order to improve the outcome efficiency, we propose an easily implementable mechanism design whereby the server randomly drops packets with a probability that is a function of the total arrival rate. We show that this results in a modified M/M/1 queueing game that is an ordinal potential game with at least one NE. In particular, for a linear packet dropping function, which is similar to the Random Early Detection (RED) algorithm used in Internet Congestion Control, we prove that there is a unique NE. We also show that the simple best response dynamic converges to this unique equilibrium. Finally, for this scheme, we prove that the social welfare (expressed either as the summation of utilities of all players, or as the summation of the logarithm of utilities of all players) at the equilibrium point can be arbitrarily close to the social welfare at the global optimal point, i.e. the PoA can be made arbitrarily close to 1. We also study the impact of arrival rate estimation error on the PoA through simulations.
2008.10038
Pengfei Ge
Pengfei Ge, Chuan-Xian Ren, Jiashi Feng, Shuicheng Yan
Dual Adversarial Auto-Encoders for Clustering
null
null
10.1109/TNNLS.2019.2919948
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
As a powerful approach for exploratory data analysis, unsupervised clustering is a fundamental task in computer vision and pattern recognition. Many clustering algorithms have been developed, but most of them perform unsatisfactorily on the data with complex structures. Recently, Adversarial Auto-Encoder (AAE) shows effectiveness on tackling such data by combining Auto-Encoder (AE) and adversarial training, but it cannot effectively extract classification information from the unlabeled data. In this work, we propose Dual Adversarial Auto-encoder (Dual-AAE) which simultaneously maximizes the likelihood function and mutual information between observed examples and a subset of latent variables. By performing variational inference on the objective function of Dual-AAE, we derive a new reconstruction loss which can be optimized by training a pair of Auto-encoders. Moreover, to avoid mode collapse, we introduce the clustering regularization term for the category variable. Experiments on four benchmarks show that Dual-AAE achieves superior performance over state-of-the-art clustering methods. Besides, by adding a reject option, the clustering accuracy of Dual-AAE can reach that of supervised CNN algorithms. Dual-AAE can also be used for disentangling style and content of images without using supervised information.
[ { "created": "Sun, 23 Aug 2020 13:16:34 GMT", "version": "v1" } ]
2020-08-25
[ [ "Ge", "Pengfei", "" ], [ "Ren", "Chuan-Xian", "" ], [ "Feng", "Jiashi", "" ], [ "Yan", "Shuicheng", "" ] ]
As a powerful approach for exploratory data analysis, unsupervised clustering is a fundamental task in computer vision and pattern recognition. Many clustering algorithms have been developed, but most of them perform unsatisfactorily on the data with complex structures. Recently, Adversarial Auto-Encoder (AAE) shows effectiveness on tackling such data by combining Auto-Encoder (AE) and adversarial training, but it cannot effectively extract classification information from the unlabeled data. In this work, we propose Dual Adversarial Auto-encoder (Dual-AAE) which simultaneously maximizes the likelihood function and mutual information between observed examples and a subset of latent variables. By performing variational inference on the objective function of Dual-AAE, we derive a new reconstruction loss which can be optimized by training a pair of Auto-encoders. Moreover, to avoid mode collapse, we introduce the clustering regularization term for the category variable. Experiments on four benchmarks show that Dual-AAE achieves superior performance over state-of-the-art clustering methods. Besides, by adding a reject option, the clustering accuracy of Dual-AAE can reach that of supervised CNN algorithms. Dual-AAE can also be used for disentangling style and content of images without using supervised information.
1607.03830
Jian Du
Jian Du and Yik-Chung Wu
Distributed Clock Skew and Offset Estimation in Wireless Sensor Networks: Asynchronous Algorithm and Convergence Analysis
null
null
null
null
cs.DC cs.IT math.IT
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In this paper, we propose a fully distributed algorithm for joint clock skew and offset estimation in wireless sensor networks based on belief propagation. In the proposed algorithm, each node can estimate its clock skew and offset in a completely distributed and asynchronous way: some nodes may update their estimates more frequently than others using outdated message from neighboring nodes. In addition, the proposed algorithm is robust to random packet loss. Such algorithm does not require any centralized information processing or coordination, and is scalable with network size. The proposed algorithm represents a unified framework that encompasses both classes of synchronous and asynchronous algorithms for network-wide clock synchronization. It is shown analytically that the proposed asynchronous algorithm converges to the optimal estimates with estimation mean-square-error at each node approaching the centralized Cram\'er-Rao bound under any network topology. Simulation results further show that {the convergence speed is faster than that corresponding to a synchronous algorithm}.
[ { "created": "Sun, 10 Jul 2016 15:38:22 GMT", "version": "v1" } ]
2016-07-14
[ [ "Du", "Jian", "" ], [ "Wu", "Yik-Chung", "" ] ]
In this paper, we propose a fully distributed algorithm for joint clock skew and offset estimation in wireless sensor networks based on belief propagation. In the proposed algorithm, each node can estimate its clock skew and offset in a completely distributed and asynchronous way: some nodes may update their estimates more frequently than others using outdated message from neighboring nodes. In addition, the proposed algorithm is robust to random packet loss. Such algorithm does not require any centralized information processing or coordination, and is scalable with network size. The proposed algorithm represents a unified framework that encompasses both classes of synchronous and asynchronous algorithms for network-wide clock synchronization. It is shown analytically that the proposed asynchronous algorithm converges to the optimal estimates with estimation mean-square-error at each node approaching the centralized Cram\'er-Rao bound under any network topology. Simulation results further show that {the convergence speed is faster than that corresponding to a synchronous algorithm}.
1912.01438
Zirui Wang
Zirui Wang, Shuda Li, Henry Howard-Jenkins, Victor Adrian Prisacariu, Min Chen
FlowNet3D++: Geometric Losses For Deep Scene Flow Estimation
WACV 2020
null
null
null
cs.CV cs.RO
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We present FlowNet3D++, a deep scene flow estimation network. Inspired by classical methods, FlowNet3D++ incorporates geometric constraints in the form of point-to-plane distance and angular alignment between individual vectors in the flow field, into FlowNet3D. We demonstrate that the addition of these geometric loss terms improves the previous state-of-art FlowNet3D accuracy from 57.85% to 63.43%. To further demonstrate the effectiveness of our geometric constraints, we propose a benchmark for flow estimation on the task of dynamic 3D reconstruction, thus providing a more holistic and practical measure of performance than the breakdown of individual metrics previously used to evaluate scene flow. This is made possible through the contribution of a novel pipeline to integrate point-based scene flow predictions into a global dense volume. FlowNet3D++ achieves up to a 15.0% reduction in reconstruction error over FlowNet3D, and up to a 35.2% improvement over KillingFusion alone. We will release our scene flow estimation code later.
[ { "created": "Tue, 3 Dec 2019 14:53:56 GMT", "version": "v1" }, { "created": "Tue, 10 Dec 2019 11:16:06 GMT", "version": "v2" }, { "created": "Mon, 26 Apr 2021 14:00:11 GMT", "version": "v3" } ]
2021-04-27
[ [ "Wang", "Zirui", "" ], [ "Li", "Shuda", "" ], [ "Howard-Jenkins", "Henry", "" ], [ "Prisacariu", "Victor Adrian", "" ], [ "Chen", "Min", "" ] ]
We present FlowNet3D++, a deep scene flow estimation network. Inspired by classical methods, FlowNet3D++ incorporates geometric constraints in the form of point-to-plane distance and angular alignment between individual vectors in the flow field, into FlowNet3D. We demonstrate that the addition of these geometric loss terms improves the previous state-of-art FlowNet3D accuracy from 57.85% to 63.43%. To further demonstrate the effectiveness of our geometric constraints, we propose a benchmark for flow estimation on the task of dynamic 3D reconstruction, thus providing a more holistic and practical measure of performance than the breakdown of individual metrics previously used to evaluate scene flow. This is made possible through the contribution of a novel pipeline to integrate point-based scene flow predictions into a global dense volume. FlowNet3D++ achieves up to a 15.0% reduction in reconstruction error over FlowNet3D, and up to a 35.2% improvement over KillingFusion alone. We will release our scene flow estimation code later.
1901.08479
Jaehoon Cha
Jaehoon Cha, Kyeong Soo Kim, Sanghyuk Lee
On the Transformation of Latent Space in Autoencoders
null
null
null
null
cs.LG stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Noting the importance of the latent variables in inference and learning, we propose a novel framework for autoencoders based on the homeomorphic transformation of latent variables, which could reduce the distance between vectors in the transformed space, while preserving the topological properties of the original space, and investigate the effect of the latent space transformation on learning generative models and denoising corrupted data. The experimental results demonstrate that our generative and denoising models based on the proposed framework can provide better performance than conventional variational and denoising autoencoders due to the transformation, where we evaluate the performance of generative and denoising models in terms of the Hausdorff distance between the sets of training and processed i.e., either generated or denoised images, which can objectively measure their differences, as well as through direct comparison of the visual characteristics of the processed images.
[ { "created": "Thu, 24 Jan 2019 16:13:24 GMT", "version": "v1" }, { "created": "Mon, 3 Jun 2019 14:35:57 GMT", "version": "v2" } ]
2019-06-04
[ [ "Cha", "Jaehoon", "" ], [ "Kim", "Kyeong Soo", "" ], [ "Lee", "Sanghyuk", "" ] ]
Noting the importance of the latent variables in inference and learning, we propose a novel framework for autoencoders based on the homeomorphic transformation of latent variables, which could reduce the distance between vectors in the transformed space, while preserving the topological properties of the original space, and investigate the effect of the latent space transformation on learning generative models and denoising corrupted data. The experimental results demonstrate that our generative and denoising models based on the proposed framework can provide better performance than conventional variational and denoising autoencoders due to the transformation, where we evaluate the performance of generative and denoising models in terms of the Hausdorff distance between the sets of training and processed i.e., either generated or denoised images, which can objectively measure their differences, as well as through direct comparison of the visual characteristics of the processed images.
1708.08552
Xuanqing Liu
Xuanqing Liu, Cho-Jui Hsieh, Jason D. Lee and Yuekai Sun
An inexact subsampled proximal Newton-type method for large-scale machine learning
null
null
null
null
cs.LG cs.NA stat.ML
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We propose a fast proximal Newton-type algorithm for minimizing regularized finite sums that returns an $\epsilon$-suboptimal point in $\tilde{\mathcal{O}}(d(n + \sqrt{\kappa d})\log(\frac{1}{\epsilon}))$ FLOPS, where $n$ is number of samples, $d$ is feature dimension, and $\kappa$ is the condition number. As long as $n > d$, the proposed method is more efficient than state-of-the-art accelerated stochastic first-order methods for non-smooth regularizers which requires $\tilde{\mathcal{O}}(d(n + \sqrt{\kappa n})\log(\frac{1}{\epsilon}))$ FLOPS. The key idea is to form the subsampled Newton subproblem in a way that preserves the finite sum structure of the objective, thereby allowing us to leverage recent developments in stochastic first-order methods to solve the subproblem. Experimental results verify that the proposed algorithm outperforms previous algorithms for $\ell_1$-regularized logistic regression on real datasets.
[ { "created": "Mon, 28 Aug 2017 22:47:48 GMT", "version": "v1" } ]
2017-08-30
[ [ "Liu", "Xuanqing", "" ], [ "Hsieh", "Cho-Jui", "" ], [ "Lee", "Jason D.", "" ], [ "Sun", "Yuekai", "" ] ]
We propose a fast proximal Newton-type algorithm for minimizing regularized finite sums that returns an $\epsilon$-suboptimal point in $\tilde{\mathcal{O}}(d(n + \sqrt{\kappa d})\log(\frac{1}{\epsilon}))$ FLOPS, where $n$ is number of samples, $d$ is feature dimension, and $\kappa$ is the condition number. As long as $n > d$, the proposed method is more efficient than state-of-the-art accelerated stochastic first-order methods for non-smooth regularizers which requires $\tilde{\mathcal{O}}(d(n + \sqrt{\kappa n})\log(\frac{1}{\epsilon}))$ FLOPS. The key idea is to form the subsampled Newton subproblem in a way that preserves the finite sum structure of the objective, thereby allowing us to leverage recent developments in stochastic first-order methods to solve the subproblem. Experimental results verify that the proposed algorithm outperforms previous algorithms for $\ell_1$-regularized logistic regression on real datasets.
2405.00028
Pavan L. Veluvali
Pavan L. Veluvali, Jan Heiland, Peter Benner
MaRDIFlow: A CSE workflow framework for abstracting meta-data from FAIR computational experiments
13 pages, 7 figures
null
null
null
cs.DC
http://creativecommons.org/licenses/by/4.0/
Numerical algorithms and computational tools are instrumental in navigating and addressing complex simulation and data processing tasks. The exponential growth of metadata and parameter-driven simulations has led to an increasing demand for automated workflows that can replicate computational experiments across platforms. In general, a computational workflow is defined as a sequential description for accomplishing a scientific objective, often described by tasks and their associated data dependencies. If characterized through input-output relation, workflow components can be structured to allow interchangeable utilization of individual tasks and their accompanying metadata. In the present work, we develop a novel computational framework, namely, MaRDIFlow, that focuses on the automation of abstracting meta-data embedded in an ontology of mathematical objects. This framework also effectively addresses the inherent execution and environmental dependencies by incorporating them into multi-layered descriptions. Additionally, we demonstrate a working prototype with example use cases and methodically integrate them into our workflow tool and data provenance framework. Furthermore, we show how to best apply the FAIR principles to computational workflows, such that abstracted components are Findable, Accessible, Interoperable, and Reusable in nature.
[ { "created": "Wed, 28 Feb 2024 16:13:17 GMT", "version": "v1" } ]
2024-05-02
[ [ "Veluvali", "Pavan L.", "" ], [ "Heiland", "Jan", "" ], [ "Benner", "Peter", "" ] ]
Numerical algorithms and computational tools are instrumental in navigating and addressing complex simulation and data processing tasks. The exponential growth of metadata and parameter-driven simulations has led to an increasing demand for automated workflows that can replicate computational experiments across platforms. In general, a computational workflow is defined as a sequential description for accomplishing a scientific objective, often described by tasks and their associated data dependencies. If characterized through input-output relation, workflow components can be structured to allow interchangeable utilization of individual tasks and their accompanying metadata. In the present work, we develop a novel computational framework, namely, MaRDIFlow, that focuses on the automation of abstracting meta-data embedded in an ontology of mathematical objects. This framework also effectively addresses the inherent execution and environmental dependencies by incorporating them into multi-layered descriptions. Additionally, we demonstrate a working prototype with example use cases and methodically integrate them into our workflow tool and data provenance framework. Furthermore, we show how to best apply the FAIR principles to computational workflows, such that abstracted components are Findable, Accessible, Interoperable, and Reusable in nature.
2103.13814
Ni Xiao
Ni Xiao and Lei Zhang
Dynamic Weighted Learning for Unsupervised Domain Adaptation
This paper has been accepted by CVPR2021
null
null
null
cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Unsupervised domain adaptation (UDA) aims to improve the classification performance on an unlabeled target domain by leveraging information from a fully labeled source domain. Recent approaches explore domain-invariant and class-discriminant representations to tackle this task. These methods, however, ignore the interaction between domain alignment learning and class discrimination learning. As a result, the missing or inadequate tradeoff between domain alignment and class discrimination are prone to the problem of negative transfer. In this paper, we propose Dynamic Weighted Learning (DWL) to avoid the discriminability vanishing problem caused by excessive alignment learning and domain misalignment problem caused by excessive discriminant learning. Technically, DWL dynamically weights the learning losses of alignment and discriminability by introducing the degree of alignment and discriminability. Besides, the problem of sample imbalance across domains is first considered in our work, and we solve the problem by weighing the samples to guarantee information balance across domains. Extensive experiments demonstrate that DWL has an excellent performance in several benchmark datasets.
[ { "created": "Mon, 22 Mar 2021 16:11:04 GMT", "version": "v1" } ]
2021-03-26
[ [ "Xiao", "Ni", "" ], [ "Zhang", "Lei", "" ] ]
Unsupervised domain adaptation (UDA) aims to improve the classification performance on an unlabeled target domain by leveraging information from a fully labeled source domain. Recent approaches explore domain-invariant and class-discriminant representations to tackle this task. These methods, however, ignore the interaction between domain alignment learning and class discrimination learning. As a result, the missing or inadequate tradeoff between domain alignment and class discrimination are prone to the problem of negative transfer. In this paper, we propose Dynamic Weighted Learning (DWL) to avoid the discriminability vanishing problem caused by excessive alignment learning and domain misalignment problem caused by excessive discriminant learning. Technically, DWL dynamically weights the learning losses of alignment and discriminability by introducing the degree of alignment and discriminability. Besides, the problem of sample imbalance across domains is first considered in our work, and we solve the problem by weighing the samples to guarantee information balance across domains. Extensive experiments demonstrate that DWL has an excellent performance in several benchmark datasets.
1603.01954
Nan Wu
Nan Wu, Zheyu Liu, Fei Qiao, Xiaojun Guo, Qi Wei, Yuan Xie, Huazhong Yang
A Real-Time and Energy-Efficient Implementation of Difference-of-Gaussian with Flexible Thin-Film Transistors
null
null
null
null
cs.ET
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
With many advantageous features, softness and better biocompatibility, flexible electronic devices have developed rapidly and increasingly attracted attention. Many currently applications with flexible devices are sensors and drivers, while there is nearly no utilization aiming at complex computation since flexible devices have lower electron mobility, simple structure and large process variation. In this paper, we proposed an innovative method that enabled flexible devices to implement real-time and energy-efficient Difference-of-Gaussian, which illustrated feasibility and potentials for them to achieve complicated real-time computation in future generation products.
[ { "created": "Mon, 7 Mar 2016 06:34:19 GMT", "version": "v1" } ]
2016-03-08
[ [ "Wu", "Nan", "" ], [ "Liu", "Zheyu", "" ], [ "Qiao", "Fei", "" ], [ "Guo", "Xiaojun", "" ], [ "Wei", "Qi", "" ], [ "Xie", "Yuan", "" ], [ "Yang", "Huazhong", "" ] ]
With many advantageous features, softness and better biocompatibility, flexible electronic devices have developed rapidly and increasingly attracted attention. Many currently applications with flexible devices are sensors and drivers, while there is nearly no utilization aiming at complex computation since flexible devices have lower electron mobility, simple structure and large process variation. In this paper, we proposed an innovative method that enabled flexible devices to implement real-time and energy-efficient Difference-of-Gaussian, which illustrated feasibility and potentials for them to achieve complicated real-time computation in future generation products.
0804.4753
Icius Committee
C. E. Huang, and J. S. Chen
Wavelet Based Iterative Learning Control with Fuzzy PD Feedback for Position Tracking of A Pneumatic Servo System
Uploaded by ICIUS2007 Conference Organizer on behalf of the author(s). 8 pages, 9 figures, 1 tables
Proceedings of the International Conference on Intelligent Unmanned System (ICIUS 2007), Bali, Indonesia, October 24-25, 2007, Paper No. ICIUS2007-A001
null
null
cs.RO
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
In this paper, a wavelet-based iterative learning control (WILC) scheme with Fuzzy PD feedback is presented for a pneumatic control system with nonsmooth nonlinearities and uncertain parameters. The wavelet transform is employed to extract the learnable dynamics from measured output signal before it can be used to update the control profile. The wavelet transform is adopted to decompose the original signal into many low-resolution signals that contain the learnable and unlearnable parts. The desired control profile is then compared with the learnable part of the transformed signal. Thus, the effects from unlearnable dynamics on the controlled system can be attenuated by a Fuzzy PD feedback controller. As for the rules of Fuzzy PD controller in the feedback loop, a genetic algorithm (GA) is employed to search for the inference rules of optimization. A proportional-valve controlled pneumatic cylinder actuator system is used as the control target for simulation. Simulation results have shown a much-improved positiontracking performance.
[ { "created": "Wed, 30 Apr 2008 08:22:19 GMT", "version": "v1" } ]
2008-05-01
[ [ "Huang", "C. E.", "" ], [ "Chen", "J. S.", "" ] ]
In this paper, a wavelet-based iterative learning control (WILC) scheme with Fuzzy PD feedback is presented for a pneumatic control system with nonsmooth nonlinearities and uncertain parameters. The wavelet transform is employed to extract the learnable dynamics from measured output signal before it can be used to update the control profile. The wavelet transform is adopted to decompose the original signal into many low-resolution signals that contain the learnable and unlearnable parts. The desired control profile is then compared with the learnable part of the transformed signal. Thus, the effects from unlearnable dynamics on the controlled system can be attenuated by a Fuzzy PD feedback controller. As for the rules of Fuzzy PD controller in the feedback loop, a genetic algorithm (GA) is employed to search for the inference rules of optimization. A proportional-valve controlled pneumatic cylinder actuator system is used as the control target for simulation. Simulation results have shown a much-improved positiontracking performance.
1209.2079
Ilgin \c{S}afak
Ilg{\i}n \c{S}afak, Emre Akta\c{s} and Ali \"Ozg\"ur Y{\i}lmaz
Error Rate Analysis of GF(q) Network Coded Detect-and-Forward Wireless Relay Networks Using Equivalent Relay Channel Models
28 pages, 10 figures. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible
null
10.1109/TWC.2013.051613.121309
null
cs.IT math.IT
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
This paper investigates simple means of analyzing the error rate performance of a general q-ary Galois Field network coded detect-and-forward cooperative relay network with known relay error statistics at the destination. Equivalent relay channels are used in obtaining an approximate error rate of the relay network, from which the diversity order is found. Error rate analyses using equivalent relay channel models are shown to be closely matched with simulation results. Using the equivalent relay channels, low complexity receivers are developed whose performances are close to that of the optimal maximum likelihood receiver.
[ { "created": "Mon, 10 Sep 2012 18:04:45 GMT", "version": "v1" }, { "created": "Tue, 5 Feb 2013 21:52:38 GMT", "version": "v2" } ]
2013-11-25
[ [ "Şafak", "Ilgın", "" ], [ "Aktaş", "Emre", "" ], [ "Yılmaz", "Ali Özgür", "" ] ]
This paper investigates simple means of analyzing the error rate performance of a general q-ary Galois Field network coded detect-and-forward cooperative relay network with known relay error statistics at the destination. Equivalent relay channels are used in obtaining an approximate error rate of the relay network, from which the diversity order is found. Error rate analyses using equivalent relay channel models are shown to be closely matched with simulation results. Using the equivalent relay channels, low complexity receivers are developed whose performances are close to that of the optimal maximum likelihood receiver.
2301.10966
Xuan Quang Ngo
Thai Nguyen Chau, Xuan Quang Ngo, Van Tu Duong, Trong Trung Nguyen, Huy Hung Nguyen, Tan Tien Nguyen
Design of Mobile Manipulator for Fire Extinguisher Testing. Part II: Design and Simulation
10 pages, 15 figures, the 7th International Conference on Advanced Engineering, Theory and Applications
null
null
null
cs.RO cs.SY eess.SY
http://creativecommons.org/licenses/by/4.0/
All flames are extinguished as early as possible, or fire services have to deal with major conflagrations. This leads to the fact that the quality of fire extinguishers has become a very sensitive and important issue in firefighting. Inspired by the development of automatic fire fighting systems, this paper presents a mobile manipulator to evaluate the power of fire extinguishers, which is designed according to the standard of fire extinguishers named as ISO 7165:2009 and ISO 11601:2008. A detailed discussion on key specifications solutions and mechanical design of the chassis of the mobile manipulator has been presented in Part I: Key Specifications and Conceptual Design. The focus of this part is on the rest of the mechanical design and controller de-sign of the mobile manipulator.
[ { "created": "Thu, 26 Jan 2023 07:19:20 GMT", "version": "v1" } ]
2023-01-31
[ [ "Chau", "Thai Nguyen", "" ], [ "Ngo", "Xuan Quang", "" ], [ "Duong", "Van Tu", "" ], [ "Nguyen", "Trong Trung", "" ], [ "Nguyen", "Huy Hung", "" ], [ "Nguyen", "Tan Tien", "" ] ]
All flames are extinguished as early as possible, or fire services have to deal with major conflagrations. This leads to the fact that the quality of fire extinguishers has become a very sensitive and important issue in firefighting. Inspired by the development of automatic fire fighting systems, this paper presents a mobile manipulator to evaluate the power of fire extinguishers, which is designed according to the standard of fire extinguishers named as ISO 7165:2009 and ISO 11601:2008. A detailed discussion on key specifications solutions and mechanical design of the chassis of the mobile manipulator has been presented in Part I: Key Specifications and Conceptual Design. The focus of this part is on the rest of the mechanical design and controller de-sign of the mobile manipulator.
0904.1645
Cedric Chauve
Cedric Chauve, A\"ida Ouangraoua (LaBRI)
A 3-approximation algorithm for computing a parsimonious first speciation in the gene duplication model
null
null
null
null
cs.DM cs.DS q-bio.QM
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We consider the following problem: from a given set of gene families trees on a set of genomes, find a first speciation, that splits these genomes into two subsets, that minimizes the number of gene duplications that happened before this speciation. We call this problem the Minimum Duplication Bipartition Problem. Using a generalization of the Minimum Edge-Cut Problem, known as Submodular Function Minimization, we propose a polynomial time and space 3-approximation algorithm for the Minimum Duplication Bipartition Problem.
[ { "created": "Fri, 10 Apr 2009 06:28:07 GMT", "version": "v1" }, { "created": "Wed, 15 Apr 2009 06:29:22 GMT", "version": "v2" } ]
2009-04-15
[ [ "Chauve", "Cedric", "", "LaBRI" ], [ "Ouangraoua", "Aïda", "", "LaBRI" ] ]
We consider the following problem: from a given set of gene families trees on a set of genomes, find a first speciation, that splits these genomes into two subsets, that minimizes the number of gene duplications that happened before this speciation. We call this problem the Minimum Duplication Bipartition Problem. Using a generalization of the Minimum Edge-Cut Problem, known as Submodular Function Minimization, we propose a polynomial time and space 3-approximation algorithm for the Minimum Duplication Bipartition Problem.
1702.08726
Lenz Belzner
Lenz Belzner, Thomas Gabor
Stacked Thompson Bandits
Accepted at SEsCPS @ ICSE 2017
null
null
null
cs.SE cs.AI cs.SY
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We introduce Stacked Thompson Bandits (STB) for efficiently generating plans that are likely to satisfy a given bounded temporal logic requirement. STB uses a simulation for evaluation of plans, and takes a Bayesian approach to using the resulting information to guide its search. In particular, we show that stacking multiarmed bandits and using Thompson sampling to guide the action selection process for each bandit enables STB to generate plans that satisfy requirements with a high probability while only searching a fraction of the search space.
[ { "created": "Tue, 28 Feb 2017 10:19:30 GMT", "version": "v1" } ]
2017-03-01
[ [ "Belzner", "Lenz", "" ], [ "Gabor", "Thomas", "" ] ]
We introduce Stacked Thompson Bandits (STB) for efficiently generating plans that are likely to satisfy a given bounded temporal logic requirement. STB uses a simulation for evaluation of plans, and takes a Bayesian approach to using the resulting information to guide its search. In particular, we show that stacking multiarmed bandits and using Thompson sampling to guide the action selection process for each bandit enables STB to generate plans that satisfy requirements with a high probability while only searching a fraction of the search space.
2104.14544
Deqing Sun
Deqing Sun, Daniel Vlasic, Charles Herrmann, Varun Jampani, Michael Krainin, Huiwen Chang, Ramin Zabih, William T. Freeman, Ce Liu
AutoFlow: Learning a Better Training Set for Optical Flow
CVPR 2021
null
null
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Synthetic datasets play a critical role in pre-training CNN models for optical flow, but they are painstaking to generate and hard to adapt to new applications. To automate the process, we present AutoFlow, a simple and effective method to render training data for optical flow that optimizes the performance of a model on a target dataset. AutoFlow takes a layered approach to render synthetic data, where the motion, shape, and appearance of each layer are controlled by learnable hyperparameters. Experimental results show that AutoFlow achieves state-of-the-art accuracy in pre-training both PWC-Net and RAFT. Our code and data are available at https://autoflow-google.github.io .
[ { "created": "Thu, 29 Apr 2021 17:55:23 GMT", "version": "v1" } ]
2021-04-30
[ [ "Sun", "Deqing", "" ], [ "Vlasic", "Daniel", "" ], [ "Herrmann", "Charles", "" ], [ "Jampani", "Varun", "" ], [ "Krainin", "Michael", "" ], [ "Chang", "Huiwen", "" ], [ "Zabih", "Ramin", "" ], [ "Freeman", "William T.", "" ], [ "Liu", "Ce", "" ] ]
Synthetic datasets play a critical role in pre-training CNN models for optical flow, but they are painstaking to generate and hard to adapt to new applications. To automate the process, we present AutoFlow, a simple and effective method to render training data for optical flow that optimizes the performance of a model on a target dataset. AutoFlow takes a layered approach to render synthetic data, where the motion, shape, and appearance of each layer are controlled by learnable hyperparameters. Experimental results show that AutoFlow achieves state-of-the-art accuracy in pre-training both PWC-Net and RAFT. Our code and data are available at https://autoflow-google.github.io .
2003.13141
Jie Chen
Jie Chen, Zhiheng Li, Jiebo Luo, and Chenliang Xu
Learning a Weakly-Supervised Video Actor-Action Segmentation Model with a Wise Selection
11 pages, 8 figures, cvpr-2020 supplementary video: https://youtu.be/CX1hEOV9tlo
null
null
null
cs.CV
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
We address weakly-supervised video actor-action segmentation (VAAS), which extends general video object segmentation (VOS) to additionally consider action labels of the actors. The most successful methods on VOS synthesize a pool of pseudo-annotations (PAs) and then refine them iteratively. However, they face challenges as to how to select from a massive amount of PAs high-quality ones, how to set an appropriate stop condition for weakly-supervised training, and how to initialize PAs pertaining to VAAS. To overcome these challenges, we propose a general Weakly-Supervised framework with a Wise Selection of training samples and model evaluation criterion (WS^2). Instead of blindly trusting quality-inconsistent PAs, WS^2 employs a learning-based selection to select effective PAs and a novel region integrity criterion as a stopping condition for weakly-supervised training. In addition, a 3D-Conv GCAM is devised to adapt to the VAAS task. Extensive experiments show that WS^2 achieves state-of-the-art performance on both weakly-supervised VOS and VAAS tasks and is on par with the best fully-supervised method on VAAS.
[ { "created": "Sun, 29 Mar 2020 21:15:18 GMT", "version": "v1" } ]
2020-03-31
[ [ "Chen", "Jie", "" ], [ "Li", "Zhiheng", "" ], [ "Luo", "Jiebo", "" ], [ "Xu", "Chenliang", "" ] ]
We address weakly-supervised video actor-action segmentation (VAAS), which extends general video object segmentation (VOS) to additionally consider action labels of the actors. The most successful methods on VOS synthesize a pool of pseudo-annotations (PAs) and then refine them iteratively. However, they face challenges as to how to select from a massive amount of PAs high-quality ones, how to set an appropriate stop condition for weakly-supervised training, and how to initialize PAs pertaining to VAAS. To overcome these challenges, we propose a general Weakly-Supervised framework with a Wise Selection of training samples and model evaluation criterion (WS^2). Instead of blindly trusting quality-inconsistent PAs, WS^2 employs a learning-based selection to select effective PAs and a novel region integrity criterion as a stopping condition for weakly-supervised training. In addition, a 3D-Conv GCAM is devised to adapt to the VAAS task. Extensive experiments show that WS^2 achieves state-of-the-art performance on both weakly-supervised VOS and VAAS tasks and is on par with the best fully-supervised method on VAAS.
2210.02719
Chenxi Sun
Chenxi Sun and Hongyan Li and Moxian Song and Derun Cai and Baofeng Zhang and Shenda Hong
Continuous Diagnosis and Prognosis by Controlling the Update Process of Deep Neural Networks
41 pages, 15 figures
null
null
null
cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Continuous diagnosis and prognosis are essential for intensive care patients. It can provide more opportunities for timely treatment and rational resource allocation, especially for sepsis, a main cause of death in ICU, and COVID-19, a new worldwide epidemic. Although deep learning methods have shown their great superiority in many medical tasks, they tend to catastrophically forget, over fit, and get results too late when performing diagnosis and prognosis in the continuous mode. In this work, we summarized the three requirements of this task, proposed a new concept, continuous classification of time series (CCTS), and designed a novel model training method, restricted update strategy of neural networks (RU). In the context of continuous prognosis, our method outperformed all baselines and achieved the average accuracy of 90%, 97%, and 85% on sepsis prognosis, COVID-19 mortality prediction, and eight diseases classification. Superiorly, our method can also endow deep learning with interpretability, having the potential to explore disease mechanisms and provide a new horizon for medical research. We have achieved disease staging for sepsis and COVID-19, discovering four stages and three stages with their typical biomarkers respectively. Further, our method is a data-agnostic and model-agnostic plug-in, it can be used to continuously prognose other diseases with staging and even implement CCTS in other fields.
[ { "created": "Thu, 6 Oct 2022 07:06:45 GMT", "version": "v1" } ]
2022-10-07
[ [ "Sun", "Chenxi", "" ], [ "Li", "Hongyan", "" ], [ "Song", "Moxian", "" ], [ "Cai", "Derun", "" ], [ "Zhang", "Baofeng", "" ], [ "Hong", "Shenda", "" ] ]
Continuous diagnosis and prognosis are essential for intensive care patients. It can provide more opportunities for timely treatment and rational resource allocation, especially for sepsis, a main cause of death in ICU, and COVID-19, a new worldwide epidemic. Although deep learning methods have shown their great superiority in many medical tasks, they tend to catastrophically forget, over fit, and get results too late when performing diagnosis and prognosis in the continuous mode. In this work, we summarized the three requirements of this task, proposed a new concept, continuous classification of time series (CCTS), and designed a novel model training method, restricted update strategy of neural networks (RU). In the context of continuous prognosis, our method outperformed all baselines and achieved the average accuracy of 90%, 97%, and 85% on sepsis prognosis, COVID-19 mortality prediction, and eight diseases classification. Superiorly, our method can also endow deep learning with interpretability, having the potential to explore disease mechanisms and provide a new horizon for medical research. We have achieved disease staging for sepsis and COVID-19, discovering four stages and three stages with their typical biomarkers respectively. Further, our method is a data-agnostic and model-agnostic plug-in, it can be used to continuously prognose other diseases with staging and even implement CCTS in other fields.
2405.04271
Arne Rubehn
Arne Rubehn, Jessica Nieder, Robert Forkel, Johann-Mattis List
Generating Feature Vectors from Phonetic Transcriptions in Cross-Linguistic Data Formats
To appear in the Proceedings of the 2024 Meeting of the Society for Computation in Linguistics (SCiL)
null
null
null
cs.CL
http://creativecommons.org/licenses/by/4.0/
When comparing speech sounds across languages, scholars often make use of feature representations of individual sounds in order to determine fine-grained sound similarities. Although binary feature systems for large numbers of speech sounds have been proposed, large-scale computational applications often face the challenges that the proposed feature systems -- even if they list features for several thousand sounds -- only cover a smaller part of the numerous speech sounds reflected in actual cross-linguistic data. In order to address the problem of missing data for attested speech sounds, we propose a new approach that can create binary feature vectors dynamically for all sounds that can be represented in the the standardized version of the International Phonetic Alphabet proposed by the Cross-Linguistic Transcription Systems (CLTS) reference catalog. Since CLTS is actively used in large data collections, covering more than 2,000 distinct language varieties, our procedure for the generation of binary feature vectors provides immediate access to a very large collection of multilingual wordlists. Testing our feature system in different ways on different datasets proves that the system is not only useful to provide a straightforward means to compare the similarity of speech sounds, but also illustrates its potential to be used in future cross-linguistic machine learning applications.
[ { "created": "Tue, 7 May 2024 12:40:59 GMT", "version": "v1" } ]
2024-05-08
[ [ "Rubehn", "Arne", "" ], [ "Nieder", "Jessica", "" ], [ "Forkel", "Robert", "" ], [ "List", "Johann-Mattis", "" ] ]
When comparing speech sounds across languages, scholars often make use of feature representations of individual sounds in order to determine fine-grained sound similarities. Although binary feature systems for large numbers of speech sounds have been proposed, large-scale computational applications often face the challenges that the proposed feature systems -- even if they list features for several thousand sounds -- only cover a smaller part of the numerous speech sounds reflected in actual cross-linguistic data. In order to address the problem of missing data for attested speech sounds, we propose a new approach that can create binary feature vectors dynamically for all sounds that can be represented in the the standardized version of the International Phonetic Alphabet proposed by the Cross-Linguistic Transcription Systems (CLTS) reference catalog. Since CLTS is actively used in large data collections, covering more than 2,000 distinct language varieties, our procedure for the generation of binary feature vectors provides immediate access to a very large collection of multilingual wordlists. Testing our feature system in different ways on different datasets proves that the system is not only useful to provide a straightforward means to compare the similarity of speech sounds, but also illustrates its potential to be used in future cross-linguistic machine learning applications.
2310.12505
Boyi Deng
Boyi Deng, Wenjie Wang, Fuli Feng, Yang Deng, Qifan Wang, Xiangnan He
Attack Prompt Generation for Red Teaming and Defending Large Language Models
Accepted to EMNLP 2023 (Findings)
null
null
null
cs.CL cs.CR cs.LG
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Large language models (LLMs) are susceptible to red teaming attacks, which can induce LLMs to generate harmful content. Previous research constructs attack prompts via manual or automatic methods, which have their own limitations on construction cost and quality. To address these issues, we propose an integrated approach that combines manual and automatic methods to economically generate high-quality attack prompts. Specifically, considering the impressive capabilities of newly emerged LLMs, we propose an attack framework to instruct LLMs to mimic human-generated prompts through in-context learning. Furthermore, we propose a defense framework that fine-tunes victim LLMs through iterative interactions with the attack framework to enhance their safety against red teaming attacks. Extensive experiments on different LLMs validate the effectiveness of our proposed attack and defense frameworks. Additionally, we release a series of attack prompts datasets named SAP with varying sizes, facilitating the safety evaluation and enhancement of more LLMs. Our code and dataset is available on https://github.com/Aatrox103/SAP .
[ { "created": "Thu, 19 Oct 2023 06:15:05 GMT", "version": "v1" } ]
2023-10-20
[ [ "Deng", "Boyi", "" ], [ "Wang", "Wenjie", "" ], [ "Feng", "Fuli", "" ], [ "Deng", "Yang", "" ], [ "Wang", "Qifan", "" ], [ "He", "Xiangnan", "" ] ]
Large language models (LLMs) are susceptible to red teaming attacks, which can induce LLMs to generate harmful content. Previous research constructs attack prompts via manual or automatic methods, which have their own limitations on construction cost and quality. To address these issues, we propose an integrated approach that combines manual and automatic methods to economically generate high-quality attack prompts. Specifically, considering the impressive capabilities of newly emerged LLMs, we propose an attack framework to instruct LLMs to mimic human-generated prompts through in-context learning. Furthermore, we propose a defense framework that fine-tunes victim LLMs through iterative interactions with the attack framework to enhance their safety against red teaming attacks. Extensive experiments on different LLMs validate the effectiveness of our proposed attack and defense frameworks. Additionally, we release a series of attack prompts datasets named SAP with varying sizes, facilitating the safety evaluation and enhancement of more LLMs. Our code and dataset is available on https://github.com/Aatrox103/SAP .
2210.10515
Pouria Mehrabi
Pouria Mehrabi, Hamid D. Taghirad
A Segment-Wise Gaussian Process-Based Ground Segmentation With Local Smoothness Estimation
null
null
null
null
cs.LG cs.SY eess.SY
http://creativecommons.org/licenses/by-nc-nd/4.0/
Both in terrestrial and extraterrestrial environments, the precise and informative model of the ground and the surface ahead is crucial for navigation and obstacle avoidance. The ground surface is not always flat and it may be sloped, bumpy and rough specially in off-road terrestrial scenes. In bumpy and rough scenes the functional relationship of the surface-related features may vary in different areas of the ground, as the structure of the ground surface may vary suddenly and further the measured point cloud of the ground does not bear smoothness. Thus, the ground-related features must be obtained based on local estimates or even point estimates. To tackle this problem, the segment-wise GP-based ground segmentation method with local smoothness estimation is proposed. This method is an extension to our previous method in which a realistic measurement of the length-scale values were provided for the covariance kernel in each line-segment to give precise estimation of the ground for sloped terrains. In this extension, the value of the length-scale is estimated locally for each data point which makes it much more precise for the rough scenes while being not computationally complex and more robust to under-segmentation, sparsity and under-represent-ability. The segment-wise task is performed to estimate a partial continuous model of the ground for each radial range segment. Simulation results show the effectiveness of the proposed method to give a continuous and precise estimation of the ground surface in rough and bumpy scenes while being fast enough for real-world applications.
[ { "created": "Wed, 19 Oct 2022 12:42:21 GMT", "version": "v1" } ]
2022-10-20
[ [ "Mehrabi", "Pouria", "" ], [ "Taghirad", "Hamid D.", "" ] ]
Both in terrestrial and extraterrestrial environments, the precise and informative model of the ground and the surface ahead is crucial for navigation and obstacle avoidance. The ground surface is not always flat and it may be sloped, bumpy and rough specially in off-road terrestrial scenes. In bumpy and rough scenes the functional relationship of the surface-related features may vary in different areas of the ground, as the structure of the ground surface may vary suddenly and further the measured point cloud of the ground does not bear smoothness. Thus, the ground-related features must be obtained based on local estimates or even point estimates. To tackle this problem, the segment-wise GP-based ground segmentation method with local smoothness estimation is proposed. This method is an extension to our previous method in which a realistic measurement of the length-scale values were provided for the covariance kernel in each line-segment to give precise estimation of the ground for sloped terrains. In this extension, the value of the length-scale is estimated locally for each data point which makes it much more precise for the rough scenes while being not computationally complex and more robust to under-segmentation, sparsity and under-represent-ability. The segment-wise task is performed to estimate a partial continuous model of the ground for each radial range segment. Simulation results show the effectiveness of the proposed method to give a continuous and precise estimation of the ground surface in rough and bumpy scenes while being fast enough for real-world applications.
2405.02150
Manoel Horta Ribeiro
Giuseppe Russo Latona, Manoel Horta Ribeiro, Tim R. Davidson, Veniamin Veselovsky, Robert West
The AI Review Lottery: Widespread AI-Assisted Peer Reviews Boost Paper Scores and Acceptance Rates
Manoel Horta Ribeiro, Tim R. Davidson, and Veniamin Veselovsky contributed equally to this work
null
null
null
cs.CY
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Journals and conferences worry that peer reviews assisted by artificial intelligence (AI), in particular, large language models (LLMs), may negatively influence the validity and fairness of the peer-review system, a cornerstone of modern science. In this work, we address this concern with a quasi-experimental study of the prevalence and impact of AI-assisted peer reviews in the context of the 2024 International Conference on Learning Representations (ICLR), a large and prestigious machine-learning conference. Our contributions are threefold. Firstly, we obtain a lower bound for the prevalence of AI-assisted reviews at ICLR 2024 using the GPTZero LLM detector, estimating that at least $15.8\%$ of reviews were written with AI assistance. Secondly, we estimate the impact of AI-assisted reviews on submission scores. Considering pairs of reviews with different scores assigned to the same paper, we find that in $53.4\%$ of pairs the AI-assisted review scores higher than the human review ($p = 0.002$; relative difference in probability of scoring higher: $+14.4\%$ in favor of AI-assisted reviews). Thirdly, we assess the impact of receiving an AI-assisted peer review on submission acceptance. In a matched study, submissions near the acceptance threshold that received an AI-assisted peer review were $4.9$ percentage points ($p = 0.024$) more likely to be accepted than submissions that did not. Overall, we show that AI-assisted reviews are consequential to the peer-review process and offer a discussion on future implications of current trends
[ { "created": "Fri, 3 May 2024 14:56:43 GMT", "version": "v1" } ]
2024-05-06
[ [ "Latona", "Giuseppe Russo", "" ], [ "Ribeiro", "Manoel Horta", "" ], [ "Davidson", "Tim R.", "" ], [ "Veselovsky", "Veniamin", "" ], [ "West", "Robert", "" ] ]
Journals and conferences worry that peer reviews assisted by artificial intelligence (AI), in particular, large language models (LLMs), may negatively influence the validity and fairness of the peer-review system, a cornerstone of modern science. In this work, we address this concern with a quasi-experimental study of the prevalence and impact of AI-assisted peer reviews in the context of the 2024 International Conference on Learning Representations (ICLR), a large and prestigious machine-learning conference. Our contributions are threefold. Firstly, we obtain a lower bound for the prevalence of AI-assisted reviews at ICLR 2024 using the GPTZero LLM detector, estimating that at least $15.8\%$ of reviews were written with AI assistance. Secondly, we estimate the impact of AI-assisted reviews on submission scores. Considering pairs of reviews with different scores assigned to the same paper, we find that in $53.4\%$ of pairs the AI-assisted review scores higher than the human review ($p = 0.002$; relative difference in probability of scoring higher: $+14.4\%$ in favor of AI-assisted reviews). Thirdly, we assess the impact of receiving an AI-assisted peer review on submission acceptance. In a matched study, submissions near the acceptance threshold that received an AI-assisted peer review were $4.9$ percentage points ($p = 0.024$) more likely to be accepted than submissions that did not. Overall, we show that AI-assisted reviews are consequential to the peer-review process and offer a discussion on future implications of current trends