aid stringlengths 9 15 | mid stringlengths 7 10 | abstract stringlengths 78 2.56k | related_work stringlengths 92 1.77k | ref_abstract dict |
|---|---|---|---|---|
1611.08307 | 2557805692 | To enhance developer productivity, all modern integrated development environments (IDEs) include code suggestion functionality that proposes likely next tokens at the cursor. While current IDEs work well for statically-typed languages, their reliance on type annotations means that they do not provide the same level of support for dynamic programming languages as for statically-typed languages. Moreover, suggestion engines in modern IDEs do not propose expressions or multi-statement idiomatic code. Recent work has shown that language models can improve code suggestion systems by learning from software repositories. This paper introduces a neural language model with a sparse pointer network aimed at capturing very long range dependencies. We release a large-scale code suggestion corpus of 41M lines of Python code crawled from GitHub. On this corpus, we found standard neural language models to perform well at suggesting local phenomena, but struggle to refer to identifiers that are introduced many tokens in the past. By augmenting a neural language model with a pointer network specialized in referring to predefined classes of identifiers, we obtain a much lower perplexity and a 5 percentage points increase in accuracy for code suggestion compared to an LSTM baseline. In fact, this increase in code suggestion accuracy is due to a 13 times more accurate prediction of identifiers. Furthermore, a qualitative analysis shows this model indeed captures interesting long-range dependencies, like referring to a class member defined over 60 tokens in the past. | @cite_0 recently used a pointer network to generate code from natural language descriptions. Our use of a controller for deciding whether to generate from a language model or copy an identifier using a sparse pointer network is inspired by their latent code predictor. However, their inputs (textual descriptions) are short whereas code suggestion requires capturing very long-range dependencies that we addressed by a filtered view on the memory of previous identifier representations. | {
"cite_N": [
"@cite_0"
],
"mid": [
"2304240348"
],
"abstract": [
"Many language generation tasks require the production of text conditioned on both structured and unstructured inputs. We present a novel neural network architecture which generates an output sequence conditioned on an arbitrary number of input functions. Crucially, our approach allows both the choice of conditioning context and the granularity of generation, for example characters or tokens, to be marginalised, thus permitting scalable and effective training. Using this framework, we address the problem of generating programming code from a mixed natural language and structured specification. We create two new data sets for this paradigm derived from the collectible trading card games Magic the Gathering and Hearthstone. On these, and a third preexisting corpus, we demonstrate that marginalising multiple predictors allows our model to outperform strong benchmarks."
]
} |
1611.08350 | 2952731221 | This paper introduces a learning scheme to construct a Hilbert space (i.e., a vector space along its inner product) to address both unsupervised and semi-supervised domain adaptation problems. This is achieved by learning projections from each domain to a latent space along the Mahalanobis metric of the latent space to simultaneously minimizing a notion of domain variance while maximizing a measure of discriminatory power. In particular, we make use of the Riemannian optimization techniques to match statistical properties (e.g., first and second order statistics) between samples projected into the latent space from different domains. Upon availability of class labels, we further deem samples sharing the same label to form more compact clusters while pulling away samples coming from different classes.We extensively evaluate and contrast our proposal against state-of-the-art methods for the task of visual domain adaptation using both handcrafted and deep-net features. Our experiments show that even with a simple nearest neighbor classifier, the proposed method can outperform several state-of-the-art methods benefitting from more involved classification schemes. | The literature on domain adaptation spans a very broad range (see @cite_17 for a recent survey). Our solution falls under the category of domain adaptation by subspace learning (DA-SL). As such, we confine our review only to methods under the umbrella of DA-SL. | {
"cite_N": [
"@cite_17"
],
"mid": [
"1982696459"
],
"abstract": [
"In pattern recognition and computer vision, one is often faced with scenarios where the training data used to learn a model have different distribution from the data on which the model is applied. Regardless of the cause, any distributional change that occurs after learning a classifier can degrade its performance at test time. Domain adaptation tries to mitigate this degradation. In this article, we provide a survey of domain adaptation methods for visual recognition. We discuss the merits and drawbacks of existing domain adaptation approaches and identify promising avenues for research in this rapidly evolving field."
]
} |
1611.08350 | 2952731221 | This paper introduces a learning scheme to construct a Hilbert space (i.e., a vector space along its inner product) to address both unsupervised and semi-supervised domain adaptation problems. This is achieved by learning projections from each domain to a latent space along the Mahalanobis metric of the latent space to simultaneously minimizing a notion of domain variance while maximizing a measure of discriminatory power. In particular, we make use of the Riemannian optimization techniques to match statistical properties (e.g., first and second order statistics) between samples projected into the latent space from different domains. Upon availability of class labels, we further deem samples sharing the same label to form more compact clusters while pulling away samples coming from different classes.We extensively evaluate and contrast our proposal against state-of-the-art methods for the task of visual domain adaptation using both handcrafted and deep-net features. Our experiments show that even with a simple nearest neighbor classifier, the proposed method can outperform several state-of-the-art methods benefitting from more involved classification schemes. | One notable example of constructing a latent space is the work of @cite_18 . In particular, the authors propose to use two fixed and predefined transformations to project source and target data to a common and higher-dimensional space. As a requirement, the method only accepts domains with the same dimensionality and hence cannot be directly used to adapt heterogeneous domains. | {
"cite_N": [
"@cite_18"
],
"mid": [
"2097908878"
],
"abstract": [
"In this work, we propose a semisupervised extension to a well-known supervised domain adaptation approach (EA) (Daume III, 2007). Our proposed approach (EA++) builds on the notion of augmented space (introduced in EA) and harnesses unlabeled data in target domain to ameliorate the transfer of information from source to target. This semisupervised approach to domain adaptation is extremely simple to implement, and can be applied as a pre-processing step to any supervised learner. Experimental results on sequential labeling tasks demonstrate the efficacy of the proposed method."
]
} |
1611.08350 | 2952731221 | This paper introduces a learning scheme to construct a Hilbert space (i.e., a vector space along its inner product) to address both unsupervised and semi-supervised domain adaptation problems. This is achieved by learning projections from each domain to a latent space along the Mahalanobis metric of the latent space to simultaneously minimizing a notion of domain variance while maximizing a measure of discriminatory power. In particular, we make use of the Riemannian optimization techniques to match statistical properties (e.g., first and second order statistics) between samples projected into the latent space from different domains. Upon availability of class labels, we further deem samples sharing the same label to form more compact clusters while pulling away samples coming from different classes.We extensively evaluate and contrast our proposal against state-of-the-art methods for the task of visual domain adaptation using both handcrafted and deep-net features. Our experiments show that even with a simple nearest neighbor classifier, the proposed method can outperform several state-of-the-art methods benefitting from more involved classification schemes. | @cite_7 observed that the geodesic connecting the source and target subspaces conveys useful information for DA and proposed the Sampling Geodesic Flow (SGF) method . The Geodesic Flow Kernel (GFK) is an improvement over the SGF technique where instead of sampling a few points on the geodesic, the whole curve is used for domain adaptation . In both methods, the domain subspaces are fixed and obtained by Principal Component Analysis (PCA) or Partial Least Square regression (PLS) . In contrast to our solution, in SGF and GFK learning the domain subspaces is disjoint from the knowledge transfer algorithm. In our experiments, we will see that the subspaces determined by our method can even boost the performance of GFK, showing the importance of joint learning of domain subspaces and the knowledge transfer scheme. @cite_23 dictionary learning is used for interpolating the intermediate subspaces. | {
"cite_N": [
"@cite_23",
"@cite_7"
],
"mid": [
"2132175002",
"2128053425"
],
"abstract": [
"Domain adaptation addresses the problem where data instances of a source domain have different distributions from that of a target domain, which occurs frequently in many real life scenarios. This work focuses on unsupervised domain adaptation, where labeled data are only available in the source domain. We propose to interpolate subspaces through dictionary learning to link the source and target domains. These subspaces are able to capture the intrinsic domain shift and form a shared feature representation for cross domain recognition. Further, we introduce a quantitative measure to characterize the shift between two domains, which enables us to select the optimal domain to adapt to the given multiple source domains. We present experiments on face recognition across pose, illumination and blur variations, cross dataset object recognition, and report improved performance over the state of the art.",
"Adapting the classifier trained on a source domain to recognize instances from a new target domain is an important problem that is receiving recent attention. In this paper, we present one of the first studies on unsupervised domain adaptation in the context of object recognition, where we have labeled data only from the source domain (and therefore do not have correspondences between object categories across domains). Motivated by incremental learning, we create intermediate representations of data between the two domains by viewing the generative subspaces (of same dimension) created from these domains as points on the Grassmann manifold, and sampling points along the geodesic between them to obtain subspaces that provide a meaningful description of the underlying domain shift. We then obtain the projections of labeled source domain data onto these subspaces, from which a discriminative classifier is learnt to classify projected data from the target domain. We discuss extensions of our approach for semi-supervised adaptation, and for cases with multiple source and target domains, and report competitive results on standard datasets."
]
} |
1611.08350 | 2952731221 | This paper introduces a learning scheme to construct a Hilbert space (i.e., a vector space along its inner product) to address both unsupervised and semi-supervised domain adaptation problems. This is achieved by learning projections from each domain to a latent space along the Mahalanobis metric of the latent space to simultaneously minimizing a notion of domain variance while maximizing a measure of discriminatory power. In particular, we make use of the Riemannian optimization techniques to match statistical properties (e.g., first and second order statistics) between samples projected into the latent space from different domains. Upon availability of class labels, we further deem samples sharing the same label to form more compact clusters while pulling away samples coming from different classes.We extensively evaluate and contrast our proposal against state-of-the-art methods for the task of visual domain adaptation using both handcrafted and deep-net features. Our experiments show that even with a simple nearest neighbor classifier, the proposed method can outperform several state-of-the-art methods benefitting from more involved classification schemes. | Domain adaptation by fixing the subspace representation of one of the domains is a popular theme in many recent works, as it simplifies the learning scheme. Examples are the max-margin adaptation , the metric similarity learning of @cite_2 and its kernel extension , the landmark approach of @cite_19 , the alignment technique of @cite_5 @cite_12 , correlation matching of @cite_4 and methods that use maximum mean discrepancy (MMD) for domain adaptation . | {
"cite_N": [
"@cite_4",
"@cite_19",
"@cite_2",
"@cite_5",
"@cite_12"
],
"mid": [
"",
"2422697180",
"1722318740",
"2104068492",
"2073309755"
],
"abstract": [
"",
"While domain adaptation (DA) aims to associate the learning tasks across data domains, heterogeneous domain adaptation (HDA) particularly deals with learning from cross-domain data which are of different types of features. In other words, for HDA, data from source and target domains are observed in separate feature spaces and thus exhibit distinct distributions. In this paper, we propose a novel learning algorithm of Cross-Domain Landmark Selection (CDLS) for solving the above task. With the goal of deriving a domain-invariant feature subspace for HDA, our CDLS is able to identify representative cross-domain data, including the unlabeled ones in the target domain, for performing adaptation. In addition, the adaptation capabilities of such cross-domain landmarks can be determined accordingly. This is the reason why our CDLS is able to achieve promising HDA performance when comparing to state-of-the-art HDA methods. We conduct classification experiments using data across different features, domains, and modalities. The effectiveness of our proposed method can be successfully verified.",
"Domain adaptation is an important emerging topic in computer vision. In this paper, we present one of the first studies of domain shift in the context of object recognition. We introduce a method that adapts object models acquired in a particular visual domain to new imaging conditions by learning a transformation that minimizes the effect of domain-induced changes in the feature distribution. The transformation is learned in a supervised manner and can be applied to categories for which there are no labeled examples in the new domain. While we focus our evaluation on object recognition tasks, the transform-based adaptation technique we develop is general and could be applied to nonimage data. Another contribution is a new multi-domain object database, freely available for download. We experimentally demonstrate the ability of our method to improve recognition on categories with few or no target domain labels and moderate to large changes in the imaging conditions.",
"In this paper, we introduce a new domain adaptation (DA) algorithm where the source and target domains are represented by subspaces described by eigenvectors. In this context, our method seeks a domain adaptation solution by learning a mapping function which aligns the source subspace with the target one. We show that the solution of the corresponding optimization problem can be obtained in a simple closed form, leading to an extremely fast algorithm. We use a theoretical result to tune the unique hyper parameter corresponding to the size of the subspaces. We run our method on various datasets and show that, despite its intrinsic simplicity, it outperforms state of the art DA methods.",
"article i nfo Partial Least Squares (PLS) methods are particularly suited to the analysis of relationships between measures of brain activity and of behavior or experimental design. In neuroimaging, PLS refers to two related methods: (1) symmetric PLS or Partial Least Squares Correlation (PLSC), and (2) asymmetric PLS or Partial Least Squares Regression (PLSR). The most popular (by far) version of PLS for neuroimaging is PLSC. It exists in several varieties based on the type of data that are related to brain activity: behavior PLSC analyzes the relationship between brain activity and behavioral data, task PLSC analyzes how brain activity relates to pre- defined categories or experimental design, seed PLSC analyzes the pattern of connectivity between brain regions, and multi-block or multi-table PLSC integrates one or more of these varieties in a common analysis. PLSR, in contrast to PLSC, is a predictive technique which, typically, predicts behavior (or design) from brain activity. For both PLS methods, statistical inferences are implemented using cross-validation techniques to identify significant patterns of voxel activation. This paper presents both PLS methods and illustrates them with small numerical examples and typical applications in neuroimaging."
]
} |
1611.08350 | 2952731221 | This paper introduces a learning scheme to construct a Hilbert space (i.e., a vector space along its inner product) to address both unsupervised and semi-supervised domain adaptation problems. This is achieved by learning projections from each domain to a latent space along the Mahalanobis metric of the latent space to simultaneously minimizing a notion of domain variance while maximizing a measure of discriminatory power. In particular, we make use of the Riemannian optimization techniques to match statistical properties (e.g., first and second order statistics) between samples projected into the latent space from different domains. Upon availability of class labels, we further deem samples sharing the same label to form more compact clusters while pulling away samples coming from different classes.We extensively evaluate and contrast our proposal against state-of-the-art methods for the task of visual domain adaptation using both handcrafted and deep-net features. Our experiments show that even with a simple nearest neighbor classifier, the proposed method can outperform several state-of-the-art methods benefitting from more involved classification schemes. | In contrast to the above methods, some studies opt to learn the domain representation along the knowledge transfer method jointly. Two representative works are the HeMap and manifold alignment . The HeMap learns two projections to minimize the instance discrepancies . The problem is however formulated such that equal number of source and target instances is required to perform the training. The manifold alignment algorithm of @cite_22 attempts to preserve the label structure in the latent space. However, it is essential for the algorithm to have access to labeled data in both source and target domains. | {
"cite_N": [
"@cite_22"
],
"mid": [
"2953265577"
],
"abstract": [
"We propose a new learning method for heterogeneous domain adaptation (HDA), in which the data from the source domain and the target domain are represented by heterogeneous features with different dimensions. Using two different projection matrices, we first transform the data from two domains into a common subspace in order to measure the similarity between the data from two domains. We then propose two new feature mapping functions to augment the transformed data with their original features and zeros. The existing learning methods (e.g., SVM and SVR) can be readily incorporated with our newly proposed augmented feature representations to effectively utilize the data from both domains for HDA. Using the hinge loss function in SVM as an example, we introduce the detailed objective function in our method called Heterogeneous Feature Augmentation (HFA) for a linear case and also describe its kernelization in order to efficiently cope with the data with very high dimensions. Moreover, we also develop an alternating optimization algorithm to effectively solve the nontrivial optimization problem in our HFA method. Comprehensive experiments on two benchmark datasets clearly demonstrate that HFA outperforms the existing HDA methods."
]
} |
1611.08269 | 2557430994 | To cope with the massive growth of semantic data streams, several RDF Stream Processing (RSP) engines have been implemented. The efficiency of their throughput, latency and memory consumption can be evaluated using available benchmarks such as LSBench and City- Bench. Nevertheless, these benchmarks lack an in-depth performance evaluation as some measurement metrics have not been considered. The main goal of this paper is to analyze the performance of two popular RSP engines, namely C-SPARQL and CQELS, when varying a set of performance metrics. More precisely, we evaluate the impact of stream rate, number of streams and window size on execution time as well as on memory consumption. | supports time-driven query execution and extends the standard SPARQL query language with keywords such as RANGE and STEP . The RANGE keyword defines the time-based window (e.g., RANGE 5m means a window of 5 minutes), and the STEP keyword indicates the frequency at which the query should be executed. Standard SPARQL 1.1 operators can be used over the data within the window such as aggregation, ordering and comparison. C-SPARQL streams out the whole output at each query execution, which refers to Rstream operator among the different streaming operators @cite_4 (e.g., Rstream, Istream, Dstream). | {
"cite_N": [
"@cite_4"
],
"mid": [
"1582747324"
],
"abstract": [
"Despite the recent surge of research in query processing over data streams, little attention has been devoted to defining precise semantics for continuous queries over streams. We first present an abstract semantics based on several building blocks: formal definitions for streams and relations, mappings among them, and any relational query language. From these basics we define a precise interpretation for continuous queries over streams and relations. We then propose a concrete language, CQL (for Continuous Query Language), which instantiates the abstract semantics using SQL as the relational query language and window specifications derived from SQL-99 to map from streams to relations. We have implemented most of the CQL language in a Data Stream Management System at Stanford, and we have developed a public repository of data stream applications that includes a wide variety of queries expressed in CQL."
]
} |
1611.08135 | 2558550198 | Community based question answering platforms have attracted substantial users to share knowledge and learn from each other. As the rapid enlargement of CQA platforms, quantities of overlapped questions emerge, which makes users confounded to select a proper reference. It is urgent for us to take effective automated algorithms to reuse historical questions with corresponding answers. In this paper we focus on the problem with question retrieval, which aims to match historical questions that are relevant or semantically equivalent to resolve one s query directly. The challenges in this task are the lexical gaps between questions for the word ambiguity and word mismatch problem. Furthermore, limited words in queried sentences cause sparsity of word features. To alleviate these challenges, we propose a novel framework named HNIL which encodes not only the question contents but also the askers social interactions to enhance the question embedding performance. More specifically, we apply random walk based learning method with recurrent neural network to match the similarities between askers question and historical questions proposed by other users. Extensive experiments on a large scale dataset from a real world CQA site show that employing the heterogeneous social network information outperforms the other state of the art solutions in this task. | The first approach is the most widely considered in exploring question retrieval problems. It considers the metadata of questions by taking question categories and labels into consideration. @cite_30 , @cite_20 , @cite_22 embodied three language models to exploit question categories smoothing for estimating questions similarities under the same category. @cite_8 , @cite_19 , @cite_2 proposed several methods in employing category side-information. @cite_8 they leveraged user chosen category and filter irrelevant questions under leaf categories. @cite_19 they developed group non-negative matrix factorization with learning the category-specific topics for each category as well as shared topics across all categories. @cite_2 also employed fisher kernel to aggregate word embedding vectors from variable size into fixed-length, thus learnt a continuous word embedding model. | {
"cite_N": [
"@cite_30",
"@cite_22",
"@cite_8",
"@cite_19",
"@cite_2",
"@cite_20"
],
"mid": [
"2036089660",
"2120284981",
"2048370214",
"2251251137",
"2252143362",
"1975809876"
],
"abstract": [
"Community Question Answering (CQA) has emerged as a popular type of service meeting a wide range of information needs. Such services enable users to ask and answer questions and to access existing question-answer pairs. CQA archives contain very large volumes of valuable user-generated content and have become important information resources on the Web. To make the body of knowledge accumulated in CQA archives accessible, effective and efficient question search is required. Question search in a CQA archive aims to retrieve historical questions that are relevant to new questions posed by users. This paper proposes a category-based framework for search in CQA archives. The framework embodies several new techniques that use language models to exploit categories of questions for improving question-answer search. Experiments conducted on real data from Yahoo! Answers demonstrate that the proposed techniques are effective and efficient and are capable of outperforming baseline methods significantly.",
"Community Question Answering (CQA) is a popular type of service where users ask questions and where answers are obtained from other users or from historical question-answer pairs. CQA archives contain large volumes of questions organized into a hierarchy of categories. As an essential function of CQA services, question retrieval in a CQA archive aims to retrieve historical question-answer pairs that are relevant to a query question. This article presents several new approaches to exploiting the category information of questions for improving the performance of question retrieval, and it applies these approaches to existing question retrieval models, including a state-of-the-art question retrieval model. Experiments conducted on real CQA data demonstrate that the proposed techniques are effective and efficient and are capable of outperforming a variety of baseline methods significantly.",
"Community question answering (cQA) has become an important service due to the popularity of cQA archives on the web. This paper is concerned with the problem of question search. Question search in cQA aims to find the historical questions that are semantically equivalent or similar to the queried questions. In this paper, we propose a faster and better retrieval model for question search by leveraging user chosen category. After introducing the question category, we can filter certain amount of irrelevant historical questions under a wide range of leaf categories. Experimental results conducted on real cQA data set demonstrate that the proposed techniques are more effective and efficient than a variety of baseline methods.",
"Community question answering (CQA) has become an important service due to the popularity of CQA archives on the web. A distinctive feature is that CQA services usually organize questions into a hierarchy of natural categories. In this paper, we focus on the problem of question retrieval and propose a novel approach, called group non-negative matrix factorization with natural categories (GNMFNC). This is achieved by learning the category-specific topics for each category as well as shared topics across all categories via a group non-negative matrix factorization framework. We derive an efficient algorithm for learning the factorization, analyze its complexity, and provide proof of convergence. Experiments are carried out on a real world CQA data set from Yahoo! Answers. The results show that our proposed approach significantly outperforms various baseline methods and achieves the state-of-the-art performance for question retrieval.",
"Community question answering (cQA) has become an important issue due to the popularity of cQA archives on the web. This paper is concerned with the problem of question retrieval. Question retrieval in cQA archives aims to find the existing questions that are semantically equivalent or relevant to the queried questions. However, the lexical gap problem brings about new challenge for question retrieval in cQA. In this paper, we propose to learn continuous word embeddings with metadata of category information within cQA pages for question retrieval. To deal with the variable size of word embedding vectors, we employ the framework of fisher kernel to aggregated them into the fixedlength vectors. Experimental results on large-scale real world cQA data set show that our approach can significantly outperform state-of-the-art translation models and topic-based models for question re-",
"Community Question Answering (CQA) has emerged as a popular type of service where users ask and answer questions and access historical question-answer pairs. CQA archives contain very large volumes of questions organized into a hierarchy of categories. As an essential function of CQA services, question retrieval in a CQA archive aims to retrieve historical question-answer pairs that are relevant to a query question. In this paper, we present a new approach to exploiting category information of questions for improving the performance of question retrieval, and we apply the approach to existing question retrieval models, including a state-of-the-art question retrieval model. Experiments conducted on real CQA data demonstrate that the proposed techniques are capable of outperforming a variety of baseline methods significantly."
]
} |
1611.08135 | 2558550198 | Community based question answering platforms have attracted substantial users to share knowledge and learn from each other. As the rapid enlargement of CQA platforms, quantities of overlapped questions emerge, which makes users confounded to select a proper reference. It is urgent for us to take effective automated algorithms to reuse historical questions with corresponding answers. In this paper we focus on the problem with question retrieval, which aims to match historical questions that are relevant or semantically equivalent to resolve one s query directly. The challenges in this task are the lexical gaps between questions for the word ambiguity and word mismatch problem. Furthermore, limited words in queried sentences cause sparsity of word features. To alleviate these challenges, we propose a novel framework named HNIL which encodes not only the question contents but also the askers social interactions to enhance the question embedding performance. More specifically, we apply random walk based learning method with recurrent neural network to match the similarities between askers question and historical questions proposed by other users. Extensive experiments on a large scale dataset from a real world CQA site show that employing the heterogeneous social network information outperforms the other state of the art solutions in this task. | The second approach, translation-model based method learns the pair relevance of question-answer data to bridging Lexical gaps between queries and questions, or questions and answers. @cite_12 discussed a method that refers to the similarities between answers to estimate question semantically similar probabilities. @cite_18 combined the question part with a query likelihood approach by incorporating word-to-word translation probabilities. @cite_23 investigated empirical methods to eliminate non-topical or unimportant words in order to construct compact translation models for retrieval purposes. Apart from word-level translation method, @cite_26 learnt a phrase-based translation model which aims to capture question contextual information rather than word-based in isolation. | {
"cite_N": [
"@cite_18",
"@cite_26",
"@cite_23",
"@cite_12"
],
"mid": [
"2147308966",
"2102394389",
"2056260421",
"2162355876"
],
"abstract": [
"Retrieval in a question and answer archive involves finding good answers for a user's question. In contrast to typical document retrieval, a retrieval model for this task can exploit question similarity as well as ranking the associated answers. In this paper, we propose a retrieval model that combines a translation-based language model for the question part with a query likelihood approach for the answer part. The proposed model incorporates word-to-word translation probabilities learned through exploiting different sources of information. Experiments show that the proposed translation based language model for the question part outperforms baseline methods significantly. By combining with the query likelihood language model for the answer part, substantial additional effectiveness improvements are obtained.",
"Community-based question answer (Q&A) has become an important issue due to the popularity of Q&A archives on the web. This paper is concerned with the problem of question retrieval. Question retrieval in Q&A archives aims to find historical questions that are semantically equivalent or relevant to the queried questions. In this paper, we propose a novel phrase-based translation model for question retrieval. Compared to the traditional word-based translation models, the phrase-based translation model is more effective because it captures contextual information in modeling the translation of phrases as a whole, rather than translating single words in isolation. Experiments conducted on real Q&A data demonstrate that our proposed phrase-based translation model significantly outperforms the state-of-the-art word-based translation model.",
"Lexical gaps between queries and questions (documents) have been a major issue in question retrieval on large online question and answer (Q&A) collections. Previous studies address the issue by implicitly expanding queries with the help of translation models pre-constructed using statistical techniques. However, since it is possible for unimportant words (e.g., non-topical words, common words) to be included in the translation models, a lack of noise control on the models can cause degradation of retrieval performance. This paper investigates a number of empirical methods for eliminating unimportant words in order to construct compact translation models for retrieval purposes. Experiments conducted on a real world Q&A collection show that substantial improvements in retrieval performance can be achieved by using compact translation models.",
"There has recently been a significant increase in the number of community-based question and answer services on the Web where people answer other peoples' questions. These services rapidly build up large archives of questions and answers, and these archives are a valuable linguistic resource. One of the major tasks in a question and answer service is to find questions in the archive that a semantically similar to a user's question. This enables high quality answers from the archive to be retrieved and removes the time lag associated with a community-based system. In this paper, we discuss methods for question retrieval that are based on using the similarity between answers in the archive to estimate probabilities for a translation-based retrieval model. We show that with this model it is possible to find semantically similar questions with relatively little word overlap."
]
} |
1611.08135 | 2558550198 | Community based question answering platforms have attracted substantial users to share knowledge and learn from each other. As the rapid enlargement of CQA platforms, quantities of overlapped questions emerge, which makes users confounded to select a proper reference. It is urgent for us to take effective automated algorithms to reuse historical questions with corresponding answers. In this paper we focus on the problem with question retrieval, which aims to match historical questions that are relevant or semantically equivalent to resolve one s query directly. The challenges in this task are the lexical gaps between questions for the word ambiguity and word mismatch problem. Furthermore, limited words in queried sentences cause sparsity of word features. To alleviate these challenges, we propose a novel framework named HNIL which encodes not only the question contents but also the askers social interactions to enhance the question embedding performance. More specifically, we apply random walk based learning method with recurrent neural network to match the similarities between askers question and historical questions proposed by other users. Extensive experiments on a large scale dataset from a real world CQA site show that employing the heterogeneous social network information outperforms the other state of the art solutions in this task. | The third approach topic-modelling based approaches also arised many attentions for we can compare the latent similarity of questions without being constrainted by the queried sentences forms. @cite_28 identified the question topic and focus into a consisting data structure. @cite_0 assumed that questions and answers share some common latent topics and through this way the model can match questions on a topic level. Chen' @cite_15 assumption is quite similar to @cite_0 . | {
"cite_N": [
"@cite_28",
"@cite_15",
"@cite_0"
],
"mid": [
"2156938860",
"1994842363",
"1970875273"
],
"abstract": [
"This paper is concerned with the problem of question search. In question search, given a question as query, we are to return questions semantically equivalent or close to the queried question. In this paper, we propose to conduct question search by identifying question topic and question focus. More specifically, we first summarize questions in a data structure consisting of question topic and question focus. Then we model question topic and question focus in a language modeling framework for search. We also propose to use the MDLbased tree cut model for identifying question topic and question focus automatically. Experimental results indicate that our approach of identifying question topic and question focus for search significantly outperforms the baseline methods such as Vector Space Model (VSM) and Language Model for Information Retrieval (LMIR).",
"The major challenge for Question Retrieval (QR) in Community Question Answering (CQA) is the lexical gap between the queried question and the historical questions. This paper proposes a novel Question-Answer Topic Model (QATM) to learn the latent topics aligned across the question-answer pairs to alleviate the lexical gap problem, with the assumption that a question and its paired answer share the same topic distribution. Experiments conducted on a real world CQA dataset from Yahoo! Answers show that combining both parts properly can get more knowledge than each part or both parts in a simple mixing way and combining our QATM with the state-of-the-art translation-based language model, where the topic and translation information is learned from the question-answer pairs at two different grained semantic levels respectively, can significantly improve the QR performance.",
"This paper studies the problem of question retrieval in community question answering (CQA). To bridge lexical gaps in questions, which is regarded as the biggest challenge in retrieval, state-of-the-art methods learn translation models using answers under an assumption that they are parallel texts. In practice, however, questions and answers are far from \"parallel\". Indeed, they are heterogeneous for both the literal level and user behaviors. There are a particularly large number of low quality answers, to which the performance of translation models is vulnerable. To address these problems, we propose a supervised question-answer topic modeling approach. The approach assumes that questions and answers share some common latent topics and are generated in a \"question language\" and \"answer language\" respectively following the topics. The topics also determine an answer quality signal. Compared with translation models, our approach not only comprehensively models user behaviors on CQA portals, but also highlights the instinctive heterogeneity of questions and answers. More importantly, it takes answer quality into account and performs robustly against noise in answers. With the topic modeling approach, we propose a topic-based language model, which matches questions not only on a term level but also on a topic level. We conducted experiments on large scale data from Yahoo! Answers and Baidu Knows. Experimental results show that the proposed model can significantly outperform state-of-the-art retrieval models in CQA."
]
} |
1611.08135 | 2558550198 | Community based question answering platforms have attracted substantial users to share knowledge and learn from each other. As the rapid enlargement of CQA platforms, quantities of overlapped questions emerge, which makes users confounded to select a proper reference. It is urgent for us to take effective automated algorithms to reuse historical questions with corresponding answers. In this paper we focus on the problem with question retrieval, which aims to match historical questions that are relevant or semantically equivalent to resolve one s query directly. The challenges in this task are the lexical gaps between questions for the word ambiguity and word mismatch problem. Furthermore, limited words in queried sentences cause sparsity of word features. To alleviate these challenges, we propose a novel framework named HNIL which encodes not only the question contents but also the askers social interactions to enhance the question embedding performance. More specifically, we apply random walk based learning method with recurrent neural network to match the similarities between askers question and historical questions proposed by other users. Extensive experiments on a large scale dataset from a real world CQA site show that employing the heterogeneous social network information outperforms the other state of the art solutions in this task. | The fourth approaches leverage neural network to model questions embeddings. As the flourish of deep learning especially in natural language processing, researchers bagan to incorporate the neural network into learning to rank frameworks. @cite_17 learnd the semantic representation of queries and answers by using a neural network architecture. Although the mainstream models of neural network are mainly applied in question answering not in question retrieval, the theories are the same. @cite_4 encoded questions and answers in semantic space and model their interactions in a convolutional neural tensor network architecture. The model is a general architecture with no need for lexical or syntactic analysis. @cite_29 utilized a similarity matrix which contains both lexical and sequential information to effectively model the complicated matching relations between questions and answers. | {
"cite_N": [
"@cite_29",
"@cite_4",
"@cite_17"
],
"mid": [
"2264274310",
"2291880741",
"2192410469"
],
"abstract": [
"Community-based Question Answering (CQA) has become popular in knowledge sharing sites since it allows users to get answers to complex, detailed, and personal questions directly from other users. Large archives of historical questions and associated answers have been accumulated. Retrieving relevant historical answers that best match a question is an essential component of a CQA service. Most state of the art approaches are based on bag-of-words models, which have been proven successful in a range of text matching tasks, but are insufficient for capturing the important word sequence information in short text matching. In this paper, a new architecture is proposed to more effectively model the complicated matching relations between questions and answers. It utilises a similarity matrix which contains both lexical and sequential information. Afterwards the information is put into a deep architecture to find potentially suitable answers. The experimental study shows its potential in improving matching accuracy of question and answer.",
"Retrieving similar questions is very important in community-based question answering. A major challenge is the lexical gap in sentence matching. In this paper, we propose a convolutional neural tensor network architecture to encode the sentences in semantic space and model their interactions with a tensor layer. Our model integrates sentence modeling and semantic matching into a single model, which can not only capture the useful information with convolutional and pooling layers, but also learn the matching metrics between the question and its answer. Besides, our model is a general architecture, with no need for the other knowledge such as lexical or syntactic analysis. The experimental results shows that our method outperforms the other methods on two matching tasks.",
"Learning the semantic representation using neural network architecture.The neural network is trained via pre-training and fine-tuning phase.The learned semantic level feature is incorporated into a LTR framework. In community question answering (cQA), users pose queries (or questions) on portals like Yahoo! Answers which can then be answered by other users who are often knowledgeable on the subject. cQA is increasingly popular on the Web, due to its convenience and effectiveness in connecting users with queries and those with answers. In this article, we study the problem of finding previous queries (e.g., posed by other users) which may be similar to new queries, and adapting their answers as the answers to the new queries. A key challenge here is to the bridge the lexical gap between new queries and old answers. For example, \"company\" in the queries may correspond to \"firm\" in the answers. To address this challenge, past research has proposed techniques similar to machine translation that \"translate\" old answers to ones using the words in the new queries. However, a key limitation of these works is that they assume queries and answers are parallel texts, which is hardly true in reality. As a result, the translated or rephrased answers may not look intuitive.In this article, we propose a novel approach to learn the semantic representation of queries and answers by using a neural network architecture. The learned semantic level features are finally incorporated into a learning to rank framework. We have evaluated our approach using a large-scale data set. Results show that the approach can significantly outperform existing approaches."
]
} |
1611.08209 | 2950450215 | We consider the problem of fault-tolerant parallel search on an infinite line by @math robots. Starting from the origin, the robots are required to find a target at an unknown location. The robots can move with maximum speed @math and can communicate in wireless mode among themselves. However, among the @math robots, there are @math robots that exhibit byzantine faults . A faulty robot can fail to report the target even after reaching it, or it can make malicious claims about having found the target when in fact it has not. Given the presence of such faulty robots, the search for the target can only be concluded when the non-faulty robots have sufficient verification that the target has been found. We aim to design algorithms that minimize the value of @math , the time to find a target at a distance @math from the origin by @math robots among which @math are faulty. We give several different algorithms whose running time depends on the ratio @math , the density of faulty robots, and also prove lower bounds. Our algorithms are optimal for some densities of faulty robots. | A search problem is usually seen as localization of a hidden target using searchers capable to move in the environment. It is an optimization question, usually attempting to minimize the time needed to complete the search. The question has been studied in numerous variations involving static or moving targets, one or many searchers, known or unknown environment, synchronous or asynchronous settings, different speed agents and many others (cf. @cite_10 ). | {
"cite_N": [
"@cite_10"
],
"mid": [
"2106518318"
],
"abstract": [
"Graph searching encompasses a wide variety of combinatorial problems related to the problem of capturing a fugitive residing in a graph using the minimum number of searchers. In this annotated bibliography, we give an elementary classification of problems and results related to graph searching and provide a source of bibliographical references on this field."
]
} |
1611.08209 | 2950450215 | We consider the problem of fault-tolerant parallel search on an infinite line by @math robots. Starting from the origin, the robots are required to find a target at an unknown location. The robots can move with maximum speed @math and can communicate in wireless mode among themselves. However, among the @math robots, there are @math robots that exhibit byzantine faults . A faulty robot can fail to report the target even after reaching it, or it can make malicious claims about having found the target when in fact it has not. Given the presence of such faulty robots, the search for the target can only be concluded when the non-faulty robots have sufficient verification that the target has been found. We aim to design algorithms that minimize the value of @math , the time to find a target at a distance @math from the origin by @math robots among which @math are faulty. We give several different algorithms whose running time depends on the ratio @math , the density of faulty robots, and also prove lower bounds. Our algorithms are optimal for some densities of faulty robots. | In several studies, when the environment is not known in advance, search implies exploration, often involving mapping and localizing searchers within the environment @cite_17 @cite_5 @cite_30 @cite_15 @cite_20 @cite_18 . However, even for the case of a known, simple environment like a line, there were several interesting studies attempting to optimize the search time. They started with the independent works of Bellman @cite_29 and Beck @cite_6 , in which the authors attempted to minimize the competitive ratio in a stochastic setting. More exactly, they proved that time @math is needed to guarantee finding the target situated at a ( unknown) distance @math from the origin. Several other works on linear search followed (e.g. see @cite_9 @cite_6 @cite_23 @cite_16 @cite_24 @cite_26 @cite_29 ). More recently the search by a single searcher was studied for different models, e.g., when the turn cost was considered @cite_19 , when the bounds on the distance to the target are known in advance @cite_14 , and when the target was moving or for more general linear cost functions @cite_22 . | {
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_26",
"@cite_14",
"@cite_22",
"@cite_29",
"@cite_9",
"@cite_6",
"@cite_24",
"@cite_19",
"@cite_23",
"@cite_5",
"@cite_15",
"@cite_16",
"@cite_20",
"@cite_17"
],
"mid": [
"2126572568",
"1992876208",
"",
"1548168317",
"",
"",
"1501957312",
"1990654078",
"2020391710",
"1995434882",
"2068224788",
"2158277215",
"2075002099",
"1978075484",
"2159175077",
""
],
"abstract": [
"The authors consider the problem faced by a newborn that must explore and learn an unknown room with obstacles in it. They seek algorithms that achieve a bounded ratio of the worst-case distance traversed in order to see all visible points of the environment (thus creating a map), divided by the optimum distance needed to verify the map. The situation is complicated by the fact that the latter offline problem (optimally verifying a map) is NP-hard and thus must be solved approximately. Although the authors show that there is no such competitive algorithm for general obstacle courses, they give a competitive algorithm for the case of a polygonal room with a bounded number of obstacles in it. >",
"We study several versions of the shortest-path problem when the map is not known in advanced, but is specified dynamically. We are seeking dynamic decision rules that optimize the worst-case ratio of the distance covered to the length of the (statically) optimal path. We describe optimal decision rules for two cases: Layered graphs of bounded width, and two-dimensional scenes with unit square obstacles. The optimal rules turn out to be intuitive, common-sense heuristics. For slightly more general graphs and scenes, we show that no bounded ratio is possible. We also show that the computational problem of devising a strategy that achieves a given worst-case ratio to the optimum path in a graph is a universal two-person game, and thus PSPACE-complete, whereas optimizing the expected ratio is #P-hard.",
"",
"We revisit the problem of searching for a target at an unknown location on a line when given upper and lower bounds on the distance D that separates the initial position of the searcher from the target. Prior to this work, only asymptotic bounds were known for the optimal competitive ratio achievable by any search strategy in the worst case. We present the first tight bounds on the exact optimal competitive ratio achievable, parametrized in terms of the given range for D, along with an optimal search strategy that achieves this competitive ratio. We prove that this optimal strategy is unique and that it cannot be computed exactly in general. We characterize the conditions under which an optimal strategy can be computed exactly and, when it cannot, we explain how numerical methods can be used efficiently. In addition, we answer several related open questions and we discuss how to generalize these results to m rays, for any m ≥ 2.",
"",
"",
"Search Theory is one of the original disciplines within the field of Operations Research. It deals with the problem faced by a Searcher who wishes to minimize the time required to find a hidden object, or “target. ” The Searcher chooses a path in the “search space” and finds the target when he is sufficiently close to it. Traditionally, the target is assumed to have no motives of its own regarding when it is found; it is simply stationary and hidden according to a known distribution (e. g. , oil), or its motion is determined stochastically by known rules (e. g. , a fox in a forest). The problems dealt with in this book assume, on the contrary, that the “target” is an independent player of equal status to the Searcher, who cares about when he is found. We consider two possible motives of the target, and divide the book accordingly. Book I considers the zero-sum game that results when the target (here called the Hider) does not want to be found. Such problems have been called Search Games (with the “ze- sum” qualifier understood). Book II considers the opposite motive of the target, namely, that he wants to be found. In this case the Searcher and the Hider can be thought of as a team of agents (simply called Player I and Player II) with identical aims, and the coordination problem they jointly face is called the Rendezvous Search Problem.",
"A man in an automobile searches for another man who is located at some point of a certain road. He starts at a given point and knows in advance the probability that the second man is at any given point of the road. Since the man being sought might be in either direction from the starting point, the searcher will, in general, have to turn around many times before finding his target. How does he search so as to minimize the expected distance travelled? When can this minimum expectation actually be achieved? This paper answers the second of these questions.",
"The linear search problem has been discussed previously by one of the present authors. In this paper, the probability distribution of the point sought in the real line is not known to the searcher. Since there is noa priori choice of distribution which recommends itself above all others, we treat the situation as a game and obtain minimax type solutions. Different minimaxima apply depending on the factors which one wishes to minimize (resp. maximize). Certain criteria are developed which help the reader judge whether the results obtained can be considered “good advice” in the solution of real problems analogous to this one.",
"We consider the problem of searching for an object on a line at an unknown distance OPT from the original position of the searcher, in the presence of a cost of d for each time the searcher changes direction. This is a generalization of the well-studied linear-search problem. We describe a strategy that is guaranteed to find the object at a cost of at most 9 ċ OPT + 2d, which has the optimal competitive ratio 9 with respect to OPT plus the minimum corresponding additive term. Our argument for upper and lower bound uses an infinite linear program, which we solve by experimental solution of an infinite series of approximating finite linear programs, estimating the limits, and solving the resulting recurrences for an explicit proof of optimality. We feel that this technique is interesting in its own right and should help solve other searching problems. In particular, we consider the star search or cowpath problem with turn cost, where the hidden object is placed on one of m rays emanating from the original position of the searcher. For this problem we give a tight bound of (1 + 2mm (m - 1)m-1)OPT + m((m (m - 1))m-1 - 1)d. We also discuss tradeoffs between the corresponding coefficients and we consider randomized strategies on the line.",
"The linear search problem concerns a search made in the real line for a point selected according to a given probability distribution. The search begins at zero and is made by continuous motion with constant speed along the line, first in one direction and then the other. The problem is to search in such a manner that the expected time required for finding the point according to the chosen plan of search is a minimum. This plan of search is usually conceived of as having a first step, a second,etc., and in that case, this author has previously shown a necessary and sufficient condition on the probability distribution for the existence of a search plan which minimizes the expected searching time. In this paper, we define a notion of search in which there is no first step, but the steps are instead numbered from negative to positive infinity. These new rules change the problem, and under them, there is always a minimizing search procedure. In those cases which satisfy the earlier criterion, the solutions obtained are essentially the same as those obtained previously.",
"We study exploration problems where a robot has to construct a complete map of an unknown environment using a path that is as short as possible.",
"We present an on-line strategy that enables a mobile robot with vision to explore an unknown simple polygon. We prove that the resulting tour is less than 26.5 times as long as the shortest watchman tour that could be computed off-line. Our analysis is doubly founded on a novel geometric structure called angle hull. Let D be a connected region inside a simple polygon, P. We define the angle hull of D, @math , to be the set of all points in P that can see two points of D at a right angle. We show that the perimeter of @math cannot exceed in length the perimeter of D by more than a factor of 2. This upper bound is tight.",
"I wish to find something which is located on a certain road. I start at a point on the road, but I do not know in which direction the object sought is to be found. Somehow, I must incorporate in my way of searching the possibility that it is either to the right or to the left. Thus, I must search first to the right, and then to the left, and then to the right again until it is found. What is a good way of conducting this search, and what is a bad way?",
"",
""
]
} |
1611.08209 | 2950450215 | We consider the problem of fault-tolerant parallel search on an infinite line by @math robots. Starting from the origin, the robots are required to find a target at an unknown location. The robots can move with maximum speed @math and can communicate in wireless mode among themselves. However, among the @math robots, there are @math robots that exhibit byzantine faults . A faulty robot can fail to report the target even after reaching it, or it can make malicious claims about having found the target when in fact it has not. Given the presence of such faulty robots, the search for the target can only be concluded when the non-faulty robots have sufficient verification that the target has been found. We aim to design algorithms that minimize the value of @math , the time to find a target at a distance @math from the origin by @math robots among which @math are faulty. We give several different algorithms whose running time depends on the ratio @math , the density of faulty robots, and also prove lower bounds. Our algorithms are optimal for some densities of faulty robots. | Most recently variants of linear search were studied for collections of collaborating searchers (robots). @cite_11 considered linear group search, when the process is completed when the target is reached by the last robot visiting it. The robots collaborate attempting to minimize the group search time. However, @cite_11 shows that having many robots does not help and the optimal search time is still bounded from below by @math . Group search using a pair of robots having distinct maximal speeds was studied in @cite_13 , in which techniques producing optimal search time were designed. | {
"cite_N": [
"@cite_13",
"@cite_11"
],
"mid": [
"2548659391",
"32433637"
],
"abstract": [
"Two mobile robots are initially placed at the same point on an infinite line. Each robot may move on the line in either direction not exceeding its maximal speed. The robots need to find a stationary target placed at an unknown location on the line. The search is completed when both robots arrive at the target point. The target is discovered at the moment when either robot arrives at its position. The robot knowing the placement of the target may communicate it to the other robot. We look for the algorithm with the shortest possible search time (i.e. the worst-case time at which both robots meet at the target) measured as a function of the target distance from the origin (i.e. the time required to travel directly from the starting point to the target at unit velocity).",
"In this paper we consider the group search problem, or evacu- ation problem, in which k mobile entities ( ( M E )s) located on the line perform search for a specific destination. The ( M E )s are initially placed at the same origin on the line L and the target is located at an unknown distance d, either to the left or to the right from the origin. All ( M E )s must simultaneously occupy the destination, and the goal is to minimize the time necessary for this to happen. The problem with k = 1 is known as the cow-path problem, and the time required for this problem is known to be 9d − o(d) in the worst case (when the cow moves at unit speed); it is also known that this is the case for k ≥ 1 unit-speed ( M E )s. In this paper we present a clear argument for this claim by showing a rather counter-intuitive result. Namely, independent of the number of ( M E )s, group search cannot be performed faster than in time 9d − o(d). We also examine the case of k = 2 ( M E )s with different speeds, showing a surprising result that the bound of 9d can be achieved when one ( M E ) has unit speed, and the other ( M E ) moves with speed at least 1 3."
]
} |
1611.08209 | 2950450215 | We consider the problem of fault-tolerant parallel search on an infinite line by @math robots. Starting from the origin, the robots are required to find a target at an unknown location. The robots can move with maximum speed @math and can communicate in wireless mode among themselves. However, among the @math robots, there are @math robots that exhibit byzantine faults . A faulty robot can fail to report the target even after reaching it, or it can make malicious claims about having found the target when in fact it has not. Given the presence of such faulty robots, the search for the target can only be concluded when the non-faulty robots have sufficient verification that the target has been found. We aim to design algorithms that minimize the value of @math , the time to find a target at a distance @math from the origin by @math robots among which @math are faulty. We give several different algorithms whose running time depends on the ratio @math , the density of faulty robots, and also prove lower bounds. Our algorithms are optimal for some densities of faulty robots. | Fault tolerance was studied in distributed computing in various settings in the past (e.g., see @cite_27 @cite_2 @cite_3 ). However, the subject of unreliability was mainly for static components of the environment (e.g. network nodes or links), which was sometimes modelled by dynamically evolving environments (cf. @cite_28 @cite_7 ). The malfunctions arising to mobile robots were investigated for various problems of gathering or pattern forming @cite_1 @cite_25 @cite_0 @cite_4 or patrolling @cite_21 . Recently @cite_12 investigated crash faults of robots performing linear search, where the time of finding the target by the first reliable robot was optimized. However, dealing with Byzantine agents is in general more tricky, requiring to identify and to refute the most malicious adversarial behavior (e.g., see @cite_0 ). | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_28",
"@cite_21",
"@cite_1",
"@cite_3",
"@cite_0",
"@cite_27",
"@cite_2",
"@cite_25",
"@cite_12"
],
"mid": [
"1569673105",
"2120741723",
"2952844013",
"2395192821",
"1988782771",
"",
"2128634904",
"1551152239",
"2120510885",
"2041571902",
"2477849570"
],
"abstract": [
"This paper considers a system of asynchronous autonomous mobile robots that can move freely in a two-dimensional plane with no agreement on a common coordinate system. Starting from any initial configuration, the robots are required to eventually gather at a single point, not fixed in advance (gathering problem). Prior work has shown that gathering oblivious (i.e., stateless) robots cannot be achieved deterministically without additional assumptions. In particular, if robots can detect multiplicity (i.e., count robots that share the same location) gathering is possible for three or more robots. Similarly, gathering of any number of robots is possible if they share a common direction, as given by compasses, with no errors. Our work is motivated by the pragmatic standpoint that (1) compasses are error-prone devices in reality, and (2) multiplicity detection, while being easy to achieve, allows for gathering in situations with more than two robots. Consequently, this paper focusses on gathering two asynchronous mobile robots equipped with inaccurate compasses. In particular, we provide a self-stabilizing algorithm to gather, in a finite time, two oblivious robots equipped with compasses that can differ by as much as π 4.",
"In this paper we investigate distributed computation in dynamic networks in which the network topology changes from round to round. We consider a worst-case model in which the communication links for each round are chosen by an adversary, and nodes do not know who their neighbors for the current round are before they broadcast their messages. The model captures mobile networks and wireless networks, in which mobility and interference render communication unpredictable. In contrast to much of the existing work on dynamic networks, we do not assume that the network eventually stops changing; we require correctness and termination even in networks that change continually. We introduce a stability property called T -interval connectivity (for T >= 1), which stipulates that for every T consecutive rounds there exists a stable connected spanning subgraph. For T = 1 this means that the graph is connected in every round, but changes arbitrarily between rounds. We show that in 1-interval connected graphs it is possible for nodes to determine the size of the network and compute any com- putable function of their initial inputs in O(n2) rounds using messages of size O(log n + d), where d is the size of the input to a single node. Further, if the graph is T-interval connected for T > 1, the computation can be sped up by a factor of T, and any function can be computed in O(n + n2 T) rounds using messages of size O(log n + d). We also give two lower bounds on the token dissemination problem, which requires the nodes to disseminate k pieces of information to all the nodes in the network. The T-interval connected dynamic graph model is a novel model, which we believe opens new avenues for research in the theory of distributed computing in wireless, mobile and dynamic networks.",
"The past few years have seen intensive research efforts carried out in some apparently unrelated areas of dynamic systems -- delay-tolerant networks, opportunistic-mobility networks, social networks -- obtaining closely related insights. Indeed, the concepts discovered in these investigations can be viewed as parts of the same conceptual universe; and the formal models proposed so far to express some specific concepts are components of a larger formal description of this universe. The main contribution of this paper is to integrate the vast collection of concepts, formalisms, and results found in the literature into a unified framework, which we call TVG (for time-varying graphs). Using this framework, it is possible to express directly in the same formalism not only the concepts common to all those different areas, but also those specific to each. Based on this definitional work, employing both existing results and original observations, we present a hierarchical classification of TVGs; each class corresponds to a significant property examined in the distributed computing literature. We then examine how TVGs can be used to study the evolution of network properties, and propose different techniques, depending on whether the indicators for these properties are a-temporal (as in the majority of existing studies) or temporal. Finally, we briefly discuss the introduction of randomness in TVGs.",
"A team of k mobile robots is deployed on a weighted graph whose edge weights represent distances. The robots move perpetually along the domain, represented by all points belonging to the graph edges, without exceeding their maximum speed. The robots need to patrol the graph by regularly visiting all points of the domain. In this paper, we consider a team of robots (patrolmen), at most f of which may be unreliable, i.e., they fail to comply with their patrolling duties. What algorithm should be followed so as to minimize the maximum time between successive visits of every edge point by a reliable patrolman? The corresponding measure of efficiency of patrolling called idleness has been widely accepted in the robotics literature. We extend it to the case of untrusted patrolmen; we denote by ( I _k^f (G) ) the maximum time that a point of the domain may remain unvisited by reliable patrolmen. The objective is to find patrolling strategies minimizing ( I _k^f (G) ). We investigate this problem for various classes of graphs. We design optimal algorithms for line segments, which turn out to be surprisingly different from strategies for related patrolling problems proposed in the literature. We then use these results to study general graphs. For Eulerian graphs G, we give an optimal patrolling strategy with idleness ( I ^f_k(G) = (f+1) |E| k ), where |E| is the sum of the lengths of the edges of G. Further, we show the hardness of the problem of computing the idle time for three robots, at most one of which is faulty, by reduction from 3-edge-coloring of cubic graphs—a known NP-hard problem. A byproduct of our proof is the investigation of classes of graphs minimizing idle time (with respect to the total length of edges); an example of such a class is known in the literature under the name of Kotzig graphs.",
"This paper studies fault-tolerant algorithms for the problem of gathering @math autonomous mobile robots. A gathering algorithm, executed independently by each robot, must ensure that all robots are gathered at one point within finite time. In a failure-prone system, a gathering algorithm is required to successfully gather the nonfaulty robots, independently of the behavior of the faulty ones. Both crash and Byzantine faults are considered. It is first observed that most existing algorithms fail to operate correctly in a setting allowing crash failures. Subsequently, an algorithm tolerant against one crash-faulty robot in a system of three or more robots is presented. It is then observed that all known algorithms fail to operate correctly in a system prone to Byzantine faults, even in the presence of a single fault. Moreover, it is shown that in an asynchronous environment it is impossible to perform a successful gathering in a @math -robot system, even if at most one of them might fail in a Byzantine manner. Thus, the problem is studied in a fully synchronous system. An algorithm is provided in this model for gathering @math robots with at most a single faulty robot, and a more general gathering algorithm is given in an @math -robot system with up to @math faults, where @math .",
"",
"A team consisting of an unknown number of mobile agents, starting from different nodes of an unknown network, have to meet at the same node. Agents move in synchronous rounds. Each agent has a different label. Up to f of the agents are Byzantine. We consider two levels of Byzantine behavior. A strongly Byzantine agent can choose an arbitrary port when it moves and it can convey arbitrary information to other agents, while a weakly Byzantine agent can do the same, except changing its label. What is the minimum number of good agents that guarantees deterministic gathering of all of them, with terminationq We solve exactly this Byzantine gathering problem in arbitrary networks for weakly Byzantine agents and give approximate solutions for strongly Byzantine agents, both when the size of the network is known and when it is unknown. It turns out that both the strength versus the weakness of Byzantine behavior and the knowledge of network size significantly impact the results. For weakly Byzantine agents, we show that any number of good agents permits solving the problem for networks of known size. If the size is unknown, then this minimum number is fp2. More precisely, we show a deterministic polynomial algorithm that gathers all good agents in an arbitrary network, provided that there are at least fp2 of them. We also provide a matching lower bound: we prove that if the number of good agents is at most fp1, then they are not able to gather deterministically with termination in some networks. For strongly Byzantine agents, we give a lower bound of fp1, even when the graph is known: we show that f good agents cannot gather deterministically in the presence of f Byzantine agents even in a ring of known size. On the positive side, we give deterministic gathering algorithms for at least 2fp1 good agents when the size of the network is known and for at least 4fp2 good agents when it is unknown.",
"Considerable attention in recent theoretical computer science is devoted to parallel computing. Here, we would like to present a special part of this large topic, namely, the part devoted to an abstract study of the dissemination of information in interconnection networks. The importance of this research area lies in the fact that the ability of a network to effectively disseminate information is an important qualitative measure for the suitabilty of the network for parallel computing. This follows simply from the observation that the communication among processes working in parallel is one of the main parts of the whole parallel computation. So, the effectivity of information exchange among processors essentially influences the effectivity of the whole computation process.",
"Reliable computer systems must handle malfunctioning components that give conflicting information to different parts of the system. This situation can be expressed abstractly in terms of a group of generals of the Byzantine army camped with their troops around an enemy city. Communicating only by messenger, the generals must agree upon a common battle plan. However, one or more of them may be traitors who will try to confuse the others. The problem is to find an algorithm to ensure that the loyal generals will reach agreement. It is shown that, using only oral messages, this problem is solvable if and only if more than two-thirds of the generals are loyal; so a single traitor can confound two loyal generals. With unforgeable written messages, the problem is solvable for any number of generals and possible traitors. Applications of the solutions to reliable computer systems are then discussed.",
"This paper considers the convergence problem in autonomous mobile robot systems. A natural algorithm for the problem requires the robots to move towards their center of gravity. This paper proves the correctness of the gravitational algorithm in the fully asynchronous model. It also analyzes its convergence rate and establishes its convergence in the presence of crash faults.",
"We consider the problem of searching on a line using n mobile robots, of which at most f are faulty, and the remaining are reliable. The robots start at the same location and move in parallel along the line with the same speed. There is a target placed on the line at a location unknown to the robots. Reliable robots can find the target when they reach its location, but faulty robots cannot detect the target. Our goal is to design a parallel algorithm minimizing the competitive ratio, represented by the worst case ratio between the time of arrival of the first reliable robot at the target, and the distance from the source to the target. If n ≥ 2f+2, there is a simple algorithm with competitive ratio 1. For f Our search algorithm is easily seen to be optimal for the case n=f+1. We also show that as @math tends to ∞ the competitive ratio of our algorithm for the case $n = 2f+1 approaches 3 and this is optimal. More precisely, we show that asymptotically, the competitive ratio of our proportional schedule algorithm A(2f+1,f) is at most 3 + 4ln n n , while any search algorithm has a lower bound 3 + 2ln n n on its competitive ratio."
]
} |
1611.07727 | 2949738342 | In this work, we introduce the challenging problem of joint multi-person pose estimation and tracking of an unknown number of persons in unconstrained videos. Existing methods for multi-person pose estimation in images cannot be applied directly to this problem, since it also requires to solve the problem of person association over time in addition to the pose estimation for each person. We therefore propose a novel method that jointly models multi-person pose estimation and tracking in a single formulation. To this end, we represent body joint detections in a video by a spatio-temporal graph and solve an integer linear program to partition the graph into sub-graphs that correspond to plausible body pose trajectories for each person. The proposed approach implicitly handles occlusion and truncation of persons. Since the problem has not been addressed quantitatively in the literature, we introduce a challenging "Multi-Person PoseTrack" dataset, and also propose a completely unconstrained evaluation protocol that does not make any assumptions about the scale, size, location or the number of persons. Finally, we evaluate the proposed approach and several baseline methods on our new dataset. | Previous datasets used to benchmark pose estimation algorithms in-the-wild are summarized in Tab. . While there exists a number of datasets to evaluate single person pose estimation methods in videos, such as , J-HMDB @cite_7 and Penn-Action @cite_44 , none of the video datasets provides annotations to benchmark multi-person pose estimation and tracking at the same time. To allow for a quantitative evaluation of this problem, we therefore also introduce a new Multi-Person PoseTrack'' dataset which provides pose annotations for multiple persons in each video to measure pose estimation accuracy, and also provides a unique ID for each of the annotated persons to benchmark multi-person pose tracking. The proposed dataset introduces new challenges to the field of human pose estimation and tracking since it contains a large amount of appearance and pose variations, body part occlusion and truncation, large scale variations, fast camera and person movements, motion blur, and a sufficiently large number of persons per video. | {
"cite_N": [
"@cite_44",
"@cite_7"
],
"mid": [
"2137591992",
"2034014085"
],
"abstract": [
"This paper presents a novel approach for analyzing human actions in non-scripted, unconstrained video settings based on volumetric, x-y-t, patch classifiers, termed actemes. Unlike previous action-related work, the discovery of patch classifiers is posed as a strongly-supervised process. Specifically, key point labels (e.g., position) across space time are used in a data-driven training process to discover patches that are highly clustered in the space time key point configuration space. To support this process, a new human action dataset consisting of challenging consumer videos is introduced, where notably the action label, the 2D position of a set of key points and their visibilities are provided for each video frame. On a novel input video, each acteme is used in a sliding volume scheme to yield a set of sparse, non-overlapping detections. These detections provide the intermediate substrate for segmenting out the action. For action classification, the proposed representation shows significant improvement over state-of-the-art low-level features, while providing spatiotemporal localization as additional output, which sheds further light into detailed action understanding.",
"Although action recognition in videos is widely studied, current methods often fail on real-world datasets. Many recent approaches improve accuracy and robustness to cope with challenging video sequences, but it is often unclear what affects the results most. This paper attempts to provide insights based on a systematic performance evaluation using thoroughly-annotated data of human actions. We annotate human Joints for the HMDB dataset (J-HMDB). This annotation can be used to derive ground truth optical flow and segmentation. We evaluate current methods using this dataset and systematically replace the output of various algorithms with ground truth. This enables us to discover what is important - for example, should we work on improving flow algorithms, estimating human bounding boxes, or enabling pose estimation? In summary, we find that high-level pose features greatly outperform low mid level features, in particular, pose over time is critical, but current pose estimation algorithms are not yet reliable enough to provide this information. We also find that the accuracy of a top-performing action recognition framework can be greatly increased by refining the underlying low mid level features, this suggests it is important to improve optical flow and human detection algorithms. Our analysis and J-HMDB dataset should facilitate a deeper understanding of action recognition algorithms."
]
} |
1611.07781 | 2548291110 | In the scope of gestural action recognition , the size of the feature vector representing movements is in general quite large especially when full body movements are considered. Furthermore, this feature vector evolves during the movement performance so that a complete movement is fully represented by a matrix M of size DxT, whose element (M_ i,j ) represents the value of feature i at timestamps j. Many studies have addressed dimensionality reduction considering only the size of the feature vector lying in ( R ^D ) to reduce both the variability of gestural sequences expressed in the reduced space, and the computational complexity of their processing. In return, very few of these methods have explicitly addressed the dimensionality reduction along the time axis. Yet this is a major issue when considering the use of elastic distances which are characterized by a quadratic complexity along the time axis. We present in this paper an evaluation of straightforward approaches aiming at reducing the dimensionality of the matrix M for each movement, leading to consider both the dimensionality reduction of the feature vector as well as its reduction along the time axis. The dimensionality reduction of the feature vector is achieved by selecting remarkable joints in the skeleton performing the movement, basically the extremities of the articulatory chains composing the skeleton. The temporal dimensionality reduction is achieved using either a regular or adaptive down-sampling that seeks to minimize the reconstruction error of the movements. Elastic and Euclidean kernels are then compared through support vector machine learning. Two data sets that are widely referenced in the domain of human gesture recognition, and quite distinctive in terms of quality of motion capture, are used for the experimental assessment of the proposed approaches. On these data sets we experimentally show that it is feasible, and possibly desirable, to significantly reduce simultaneously the size of the feature vector and the number of skeleton frames to represent body movements while maintaining a very good recognition rate. The method proves to give satisfactory results at a level currently reached by state-of-the-art methods on these data sets. We experimentally show that the computational complexity reduction that is obtained makes this approach eligible for real-time applications. | Numerous method are available to carry out such dimensionality reduction, the most popular being linear approaches such as Principal Component Analysis (PCA, @cite_34 , @cite_21 ), Linear Discriminant Analysis (LDA, @cite_29 ), or linear projections preserving locally neighborhoods (Locality Preserving Projection) @cite_19 . Among non-linear approaches, Locally Linear Embeddings (LLE, @cite_30 ), Metric Multidimensional Scaling (MDS, @cite_27 ) and variants like Laplacian Eigenmap @cite_0 , Isomap @cite_9 have been implemented to embed postures in low dimensional spaces in which a more efficient time warp (DTW, see section ) algorithm can be used to classify movements. An extension of this method, called ST-Isomap, considers temporal relationships in local neighborhoods that can be propagated globally via a shortest-path mechanism @cite_37 . Models based on Gaussian processes with latent variables are also largely used, for instance a hierarchical version has been recently exploited for gesture recognition @cite_8 . | {
"cite_N": [
"@cite_30",
"@cite_37",
"@cite_8",
"@cite_29",
"@cite_21",
"@cite_9",
"@cite_0",
"@cite_19",
"@cite_27",
"@cite_34"
],
"mid": [
"2053186076",
"2030991373",
"2068915126",
"2129476886",
"2124635854",
"2001141328",
"2097308346",
"2154872931",
"",
"51824616"
],
"abstract": [
"Many areas of science depend on exploratory data analysis and visualization. The need to analyze large amounts of multivariate data raises the fundamental problem of dimensionality reduction: how to discover compact representations of high-dimensional data. Here, we introduce locally linear embedding (LLE), an unsupervised learning algorithm that computes low-dimensional, neighborhood-preserving embeddings of high-dimensional inputs. Unlike clustering methods for local dimensionality reduction, LLE maps its inputs into a single global coordinate system of lower dimensionality, and its optimizations do not involve local minima. By exploiting the local symmetries of linear reconstructions, LLE is able to learn the global structure of nonlinear manifolds, such as those generated by images of faces or documents of text. How do we judge similarity? Our mental representations of the world are formed by processing large numbers of sensory in",
"We present an extension of Isomap nonlinear dimension reduction (, 2000) for data with both spatial and temporal relationships. Our method, ST-Isomap, augments the existing Isomap framework to consider temporal relationships in local neighborhoods that can be propagated globally via a shortest-path mechanism. Two instantiations of ST-Isomap are presented for sequentially continuous and segmented data. Results from applying ST-Isomap to real-world data collected from human motion performance and humanoid robot teleoperation are also presented.",
"In this paper, we propose a hierarchical discriminative approach for human action recognition. It consists of feature extraction with mutual motion pattern analysis and discriminative action modeling in the hierarchical manifold space. Hierarchical Gaussian Process Latent Variable Model (HGPLVM) is employed to learn the hierarchical manifold space in which motion patterns are extracted. A cascade CRF is also presented to estimate the motion patterns in the corresponding manifold subspace, and the trained SVM classifier predicts the action label for the current observation. Using motion capture data, we test our method and evaluate how body parts make effect on human action recognition. The results on our test set of synthetic images are also presented to demonstrate the robustness.",
"Provides a systematic account of the subject area, concentrating on the most recent advances in the field. While the focus is on practical considerations, both theoretical and practical issues are explored. Among the advances covered are: regularized discriminant analysis and bootstrap-based assessment of the performance of a sample-based discriminant rule and extensions of discriminant analysis motivated by problems in statistical image analysis. Includes over 1,200 references in the bibliography.",
"Abstract This article deals with the problem of classification of human activities from video. Our approach uses motion features that are computed very efficiently, and subsequently projected into a lower dimensional space where matching is performed. Each action is represented as a manifold in this lower dimensional space and matching is done by comparing these manifolds. To demonstrate the effectiveness of this approach, it was used on a large data set of similar actions, each performed by many different actors. Classification results were very accurate and show that this approach is robust to challenges such as variations in performers' physical attributes, color of clothing, and style of motion. An important result of this article is that the recovery of the three-dimensional properties of a moving person, or even the two-dimensional tracking of the person's limbs need not precede action recognition.",
"Scientists working with large volumes of high-dimensional data, such as global climate patterns, stellar spectra, or human gene distributions, regularly confront the problem of dimensionality reduction: finding meaningful low-dimensional structures hidden in their high-dimensional observations. The human brain confronts the same problem in everyday perception, extracting from its high-dimensional sensory inputs—30,000 auditory nerve fibers or 106 optic nerve fibers—a manageably small number of perceptually relevant features. Here we describe an approach to solving dimensionality reduction problems that uses easily measured local metric information to learn the underlying global geometry of a data set. Unlike classical techniques such as principal component analysis (PCA) and multidimensional scaling (MDS), our approach is capable of discovering the nonlinear degrees of freedom that underlie complex natural observations, such as human handwriting or images of a face under different viewing conditions. In contrast to previous algorithms for nonlinear dimensionality reduction, ours efficiently computes a globally optimal solution, and, for an important class of data manifolds, is guaranteed to converge asymptotically to the true structure.",
"One of the central problems in machine learning and pattern recognition is to develop appropriate representations for complex data. We consider the problem of constructing a representation for data lying on a low-dimensional manifold embedded in a high-dimensional space. Drawing on the correspondence between the graph Laplacian, the Laplace Beltrami operator on the manifold, and the connections to the heat equation, we propose a geometrically motivated algorithm for representing the high-dimensional data. The algorithm provides a computationally efficient approach to nonlinear dimensionality reduction that has locality-preserving properties and a natural connection to clustering. Some potential applications and illustrative examples are discussed.",
"Many problems in information processing involve some form of dimensionality reduction. In this paper, we introduce Locality Preserving Projections (LPP). These are linear projective maps that arise by solving a variational problem that optimally preserves the neighborhood structure of the data set. LPP should be seen as an alternative to Principal Component Analysis (PCA) – a classical linear technique that projects the data along the directions of maximal variance. When the high dimensional data lies on a low dimensional manifold embedded in the ambient space, the Locality Preserving Projections are obtained by finding the optimal linear approximations to the eigenfunctions of the Laplace Beltrami operator on the manifold. As a result, LPP shares many of the data representation properties of nonlinear techniques such as Laplacian Eigenmaps or Locally Linear Embedding. Yet LPP is linear and more crucially is defined everywhere in ambient space rather than just on the training data points. This is borne out by illustrative examples on some high dimensional data sets.",
"",
""
]
} |
1611.07781 | 2548291110 | In the scope of gestural action recognition , the size of the feature vector representing movements is in general quite large especially when full body movements are considered. Furthermore, this feature vector evolves during the movement performance so that a complete movement is fully represented by a matrix M of size DxT, whose element (M_ i,j ) represents the value of feature i at timestamps j. Many studies have addressed dimensionality reduction considering only the size of the feature vector lying in ( R ^D ) to reduce both the variability of gestural sequences expressed in the reduced space, and the computational complexity of their processing. In return, very few of these methods have explicitly addressed the dimensionality reduction along the time axis. Yet this is a major issue when considering the use of elastic distances which are characterized by a quadratic complexity along the time axis. We present in this paper an evaluation of straightforward approaches aiming at reducing the dimensionality of the matrix M for each movement, leading to consider both the dimensionality reduction of the feature vector as well as its reduction along the time axis. The dimensionality reduction of the feature vector is achieved by selecting remarkable joints in the skeleton performing the movement, basically the extremities of the articulatory chains composing the skeleton. The temporal dimensionality reduction is achieved using either a regular or adaptive down-sampling that seeks to minimize the reconstruction error of the movements. Elastic and Euclidean kernels are then compared through support vector machine learning. Two data sets that are widely referenced in the domain of human gesture recognition, and quite distinctive in terms of quality of motion capture, are used for the experimental assessment of the proposed approaches. On these data sets we experimentally show that it is feasible, and possibly desirable, to significantly reduce simultaneously the size of the feature vector and the number of skeleton frames to represent body movements while maintaining a very good recognition rate. The method proves to give satisfactory results at a level currently reached by state-of-the-art methods on these data sets. We experimentally show that the computational complexity reduction that is obtained makes this approach eligible for real-time applications. | Gesture recognition Recognition methods essentially aim at modeling the dynamics of gestures. Some approaches, based on linear dynamic models @cite_24 , have used autoregressive (AR) and autoregressive moving-average (ARMA) models to characterize the kinematics of movements, while other approaches, based on nonlinear dynamic models @cite_14 , have developed movement analysis and recognition scheme based on dynamical models controlled by Gaussian processes. @cite_23 propose a synthesis of the major gesture recognition approaches relying on Hidden Markov Models (HMM). Histograms of oriented 4D normals have also been proposed in @cite_13 for the recognition of gestural actions from sequences of depth images. @cite_7 have exploited conditional random fields to model joint dependencies and thus increase the discrimination of HMM-like models. Recurrent neural network models have also been used @cite_35 ; among them, conditional restricted Boltzman's machines @cite_10 have been studied recently in the context of motion captured data modeling and classification. | {
"cite_N": [
"@cite_35",
"@cite_14",
"@cite_7",
"@cite_10",
"@cite_24",
"@cite_23",
"@cite_13"
],
"mid": [
"1408639475",
"2129620337",
"2145835757",
"2105577021",
"2133600694",
"2168392347",
"2085735683"
],
"abstract": [
"In this work we resolve the long-outstanding problem of how to effectively train recurrent neural networks (RNNs) on complex and difficult sequence modeling problems which may contain long-term data dependencies. Utilizing recent advances in the Hessian-free optimization approach (Martens, 2010), together with a novel damping scheme, we successfully train RNNs on two sets of challenging problems. First, a collection of pathological synthetic datasets which are known to be impossible for standard optimization approaches (due to their extremely long-term dependencies), and second, on three natural and highly complex real-world sequence datasets where we find that our method significantly outperforms the previous state-of-the-art method for training neural sequence models: the Long Short-term Memory approach of Hochreiter and Schmidhuber (1997). Additionally, we offer a new interpretation of the generalized Gauss-Newton matrix of Schraudolph (2002) which is used within the HF approach of Martens.",
"We address the problem of performing decision tasks and, in particular, classification and recognition in the space of dynamical models in order to compare time series of data. Motivated by the application of recognition of human motion in image sequences, we consider a class of models that include linear dynamics, both stable and marginally stable (periodic), both minimum and nonminimum phases, driven by non-Gaussian processes. This requires extending existing learning and system identification algorithms to handle periodic modes and nonminimum-phase behavior while taking into account higher order statistics of the data. Once a model is identified, we define a kernel-based cord distance between models, which includes their dynamics, their initial conditions, and input distribution. This is made possible by a novel kernel defined between two arbitrary (non-Gaussian) distributions, which is computed by efficiently solving an optimal transport problem. We validate our choice of models, inference algorithm, and distance on the tasks of human motion synthesis (sample paths of the learned models) and recognition (nearest-neighbor classification in the computed distance). However, our work can be applied more broadly where one needs to compare historical data while taking into account periodic trends, nonminimum-phase behavior, and non-Gaussian input distributions.",
"We introduce a discriminative hidden-state approach for the recognition of human gestures. Gesture sequences often have a complex underlying structure, and models that can incorporate hidden structures have proven to be advantageous for recognition tasks. Most existing approaches to gesture recognition with hidden states employ a Hidden Markov Model or suitable variant (e.g., a factored or coupled state model) to model gesture streams; a significant limitation of these models is the requirement of conditional independence of observations. In addition, hidden states in a generative model are selected to maximize the likelihood of generating all the examples of a given gesture class, which is not necessarily optimal for discriminating the gesture class against other gestures. Previous discriminative approaches to gesture sequence recognition have shown promising results, but have not incorporated hidden states nor addressed the problem of predicting the label of an entire sequence. In this paper, we derive a discriminative sequence model with a hidden state structure, and demonstrate its utility both in a detection and in a multi-way classification formulation. We evaluate our method on the task of recognizing human arm and head gestures, and compare the performance of our method to both generative hidden state and discriminative fully-observable models.",
"Recent developments have demonstrated the capacity of restricted Boltzmann machines (RBM) to be powerful generative models, able to extract useful features from input data or construct deep artificial neural networks. In such settings, the RBM only yields a preprocessing or an initialization for some other model, instead of acting as a complete supervised model in its own right. In this paper, we argue that RBMs can provide a self-contained framework for developing competitive classifiers. We study the Classification RBM (ClassRBM), a variant on the RBM adapted to the classification setting. We study different strategies for training the ClassRBM and show that competitive classification performances can be reached when appropriately combining discriminative and generative training objectives. Since training according to the generative objective requires the computation of a generally intractable gradient, we also compare different approaches to estimating this gradient and address the issue of obtaining such a gradient for problems with very high dimensional inputs. Finally, we describe how to adapt the ClassRBM to two special cases of classification problems, namely semi-supervised and multitask learning.",
"Human gait and activity analysis from video is presently attracting a lot of attention in the computer vision community. In this paper we analyze the role of two of the most important cues in human motion-shape and kinematics. We present an experimental framework whereby it is possible to evaluate the relative importance of these two cues in computer vision based recognition algorithms. In the process, we propose a new gait recognition algorithm by computing the distance between two sequences of shapes that lie on a spherical manifold. In our experiments, shape is represented using Kendall's definition of shape. Kinematics is represented using a Linear Dynamical system We place particular emphasis on human gait. Our conclusions show that shape plays a role which is more significant than kinematics in current automated gait based human identification algorithms. As a natural extension we study the role of shape and kinematics in activity recognition. Our experiments indicate that we require models that contain both shape and kinematics in order to perform accurate activity classification. These conclusions also allow us to explain the relative performance of many existing methods in computer-based human activity modeling.",
"Gesture recognition pertains to recognizing meaningful expressions of motion by a human, involving the hands, arms, face, head, and or body. It is of utmost importance in designing an intelligent and efficient human-computer interface. The applications of gesture recognition are manifold, ranging from sign language through medical rehabilitation to virtual reality. In this paper, we provide a survey on gesture recognition with particular emphasis on hand gestures and facial expressions. Applications involving hidden Markov models, particle filtering and condensation, finite-state machines, optical flow, skin color, and connectionist models are discussed in detail. Existing challenges and future research possibilities are also highlighted",
"We present a new descriptor for activity recognition from videos acquired by a depth sensor. Previous descriptors mostly compute shape and motion features independently, thus, they often fail to capture the complex joint shape-motion cues at pixel-level. In contrast, we describe the depth sequence using a histogram capturing the distribution of the surface normal orientation in the 4D space of time, depth, and spatial coordinates. To build the histogram, we create 4D projectors, which quantize the 4D space and represent the possible directions for the 4D normal. We initialize the projectors using the vertices of a regular polychoron. Consequently, we refine the projectors using a discriminative density measure, such that additional projectors are induced in the directions where the 4D normals are more dense and discriminative. Through extensive experiments, we demonstrate that our descriptor better captures the joint shape-motion cues in the depth sequence, and thus outperforms the state-of-the-art on all relevant benchmarks."
]
} |
1611.08091 | 2558539460 | Deep models have achieved impressive performance for face hallucination tasks. However, we observe that directly feeding the hallucinated facial images into recog- nition models can even degrade the recognition performance despite the much better visualization quality. In this paper, we address this problem by jointly learning a deep model for two tasks, i.e. face hallucination and recognition. In particular, we design an end-to-end deep convolution network with hallucination sub-network cascaded by recognition sub-network. The recognition sub- network are responsible for producing discriminative feature representations using the hallucinated images as inputs generated by hallucination sub-network. During training, we feed LR facial images into the network and optimize the parameters by minimizing two loss items, i.e. 1) face hallucination loss measured by the pixel wise difference between the ground truth HR images and network-generated images; and 2) verification loss which is measured by the classification error and intra-class distance. We extensively evaluate our method on LFW and YTF datasets. The experimental results show that our method can achieve recognition accuracy 97.95 on 4x down-sampled LFW testing set, outperforming the accuracy 96.35 of conventional face recognition model. And on the more challenging YTF dataset, we achieve recognition accuracy 90.65 , a margin over the recognition accuracy 89.45 obtained by conventional face recognition model on the 4x down-sampled version. | Conventional hallucination methods @cite_28 @cite_11 are often designed for controlled settings and cannot handle varying conditions. Deep models are also applied to face hallucination tasks @cite_6 @cite_27 . @cite_6 proposed a Bi-channel Convolutional Neural Network, which extracts robust face representations from raw input by using deep convolutional network, then adaptively integrates 2 channels of information to predict the HR image. | {
"cite_N": [
"@cite_28",
"@cite_27",
"@cite_6",
"@cite_11"
],
"mid": [
"",
"2507235960",
"2201706299",
"2141631520"
],
"abstract": [
"",
"We present a novel framework for hallucinating faces of unconstrained poses and with very low resolution (face size as small as 5pxIOD). In contrast to existing studies that mostly ignore or assume pre-aligned face spatial configuration (e.g. facial landmarks localization or dense correspondence field), we alternatingly optimize two complementary tasks, namely face hallucination and dense correspondence field estimation, in a unified framework. In addition, we propose a new gated deep bi-network that contains two functionality-specialized branches to recover different levels of texture details. Extensive experiments demonstrate that such formulation allows exceptional hallucination quality on in-the-wild low-res faces with significant pose and illumination variations.",
"Face hallucination method is proposed to generate high-resolution images from low-resolution ones for better visualization. However, conventional hallucination methods are often designed for controlled settings and cannot handle varying conditions of pose, resolution degree, and blur. In this paper, we present a new method of face hallucination, which can consistently improve the resolution of face images even with large appearance variations. Our method is based on a novel network architecture called Bi-channel Convolutional Neural Network (Bi-channel CNN). It extracts robust face representations from raw input by using deep convolu-tional network, then adaptively integrates two channels of information (the raw input image and face representations) to predict the high-resolution image. Experimental results show our system outperforms the prior state-of-the-art methods.",
"In video surveillance, the faces of interest are often of small size. Image resolution is an important factor affecting face recognition by human and computer. In this paper, we propose a new face hallucination method using eigentransformation. Different from most of the proposed methods based on probabilistic models, this method views hallucination as a transformation between different image styles. We use Principal Component Analysis (PCA) to fit the input face image as a linear combination of the low-resolution face images in the training set. The high-resolution image is rendered by replacing the low-resolution training images with high-resolution ones, while retaining the same combination coefficients. Experiments show that the hallucinated face images are not only very helpful for recognition by humans, but also make the automatic recognition procedure easier, since they emphasize the face difference by adding more high-frequency details."
]
} |
1611.08091 | 2558539460 | Deep models have achieved impressive performance for face hallucination tasks. However, we observe that directly feeding the hallucinated facial images into recog- nition models can even degrade the recognition performance despite the much better visualization quality. In this paper, we address this problem by jointly learning a deep model for two tasks, i.e. face hallucination and recognition. In particular, we design an end-to-end deep convolution network with hallucination sub-network cascaded by recognition sub-network. The recognition sub- network are responsible for producing discriminative feature representations using the hallucinated images as inputs generated by hallucination sub-network. During training, we feed LR facial images into the network and optimize the parameters by minimizing two loss items, i.e. 1) face hallucination loss measured by the pixel wise difference between the ground truth HR images and network-generated images; and 2) verification loss which is measured by the classification error and intra-class distance. We extensively evaluate our method on LFW and YTF datasets. The experimental results show that our method can achieve recognition accuracy 97.95 on 4x down-sampled LFW testing set, outperforming the accuracy 96.35 of conventional face recognition model. And on the more challenging YTF dataset, we achieve recognition accuracy 90.65 , a margin over the recognition accuracy 89.45 obtained by conventional face recognition model on the 4x down-sampled version. | Low-resolution face recognition (LR FR) aims to recognize faces from small size or poor quality images with varying pose, illumination, expression, etc. @cite_31 reported a degradation of the recognition performance when face regions became smaller than @math . @cite_16 proposed a Partially Coupled Super-Resolution Networks (PCSRN), as the pre-training part of recognition model. | {
"cite_N": [
"@cite_31",
"@cite_16"
],
"mid": [
"2054515210",
"2295634579"
],
"abstract": [
"This paper addresses the very low resolution (VLR) problem in face recognition in which the resolution of the face image to be recognized is lower than 16 × 16. With the increasing demand of surveillance camera-based applications, the VLR problem happens in many face application systems. Existing face recognition algorithms are not able to give satisfactory performance on the VLR face image. While face super-resolution (SR) methods can be employed to enhance the resolution of the images, the existing learning-based face SR methods do not perform well on such a VLR face image. To overcome this problem, this paper proposes a novel approach to learn the relationship between the high-resolution image space and the VLR image space for face SR. Based on this new approach, two constraints, namely, new data and discriminative constraints, are designed for good visuality and face recognition applications under the VLR problem, respectively. Experimental results show that the proposed SR algorithm based on relationship learning outperforms the existing algorithms in public face databases.",
"Visual recognition research often assumes a sufficient resolution of the region of interest (ROI). That is usually violated in practice, inspiring us to explore the Very Low Resolution Recognition (VLRR) problem. Typically, the ROI in a VLRR problem can be smaller than @math pixels, and is challenging to be recognized even by human experts. We attempt to solve the VLRR problem using deep learning methods. Taking advantage of techniques primarily in super resolution, domain adaptation and robust regression, we formulate a dedicated deep learning method and demonstrate how these techniques are incorporated step by step. Any extra complexity, when introduced, is fully justified by both analysis and simulation results. The resulting achieves feature enhancement and recognition simultaneously. It allows for both the flexibility to combat the LR-HR domain mismatch, and the robustness to outliers. Finally, the effectiveness of the proposed models is evaluated on three different VLRR tasks, including face identification, digit recognition and font recognition, all of which obtain very impressive performances."
]
} |
1611.08096 | 2951896856 | Community based question answering services have arisen as a popular knowledge sharing pattern for netizens. With abundant interactions among users, individuals are capable of obtaining satisfactory information. However, it is not effective for users to attain answers within minutes. Users have to check the progress over time until the satisfying answers submitted. We address this problem as a user personalized satisfaction prediction task. Existing methods usually exploit manual feature selection. It is not desirable as it requires careful design and is labor intensive. In this paper, we settle this issue by developing a new multiple instance deep learning framework. Specifically, in our settings, each question follows a weakly supervised learning multiple instance learning assumption, where its obtained answers can be regarded as instance sets and we define the question resolved with at least one satisfactory answer. We thus design an efficient framework exploiting multiple instance learning property with deep learning to model the question answer pairs. Extensive experiments on large scale datasets from Stack Exchange demonstrate the feasibility of our proposed framework in predicting askers personalized satisfaction. Our framework can be extended to numerous applications such as UI satisfaction Prediction, multi armed bandit problem, expert finding and so on. | Previous models suffer from weak interaction between two entities in the vector space. To address this problem, Socher @cite_10 first introduced the neural tensor network to allow the entities and relations to interact multiplicatively. They successively applied the neural tensor network to solve the problem in typical Natural Language Processing field. @cite_10 focused on predicting additional true relations between entities. @cite_29 studied the problem of learning new facts with semantic words. @cite_3 introduced a recursive neural tensor network to remedy sentiment detection task. Neural tensor network out-performed other linear combination approaches significantly and raised much attention among researchers. In CQA field, researchers also adopt the idea of neural tensor network. Xia @cite_14 modeled document novelty with neural tensor network for search result diversification task, they automatically learned a nonlinear novelty function based on preliminary representations of a document and other candidate documents. Qiu @cite_15 integrated Q-A pairs semantic matching with convolutional and pooling layers, and exploited neural tensor network to learn the matching metrics. In our paper, we integrate neural tensor network to link the relevance of the user's attitude towards to the question accompanied with answers. | {
"cite_N": [
"@cite_14",
"@cite_29",
"@cite_3",
"@cite_15",
"@cite_10"
],
"mid": [
"2337233909",
"1771625187",
"2251939518",
"2291880741",
"2127426251"
],
"abstract": [
"Search result diversification has attracted considerable attention as a means to tackle the ambiguous or multi-faceted information needs of users. One of the key problems in search result diversification is novelty, that is, how to measure the novelty of a candidate document with respect to other documents. In the heuristic approaches, the predefined document similarity functions are directly utilized for defining the novelty. In the learning approaches, the novelty is characterized based on a set of handcrafted features. Both the similarity functions and the features are difficult to manually design in real world due to the complexity of modeling the document novelty. In this paper, we propose to model the novelty of a document with a neural tensor network. Instead of manually defining the similarity functions or features, the new method automatically learns a nonlinear novelty function based on the preliminary representation of the candidate document and other documents. New diverse learning to rank models can be derived under the relational learning to rank framework. To determine the model parameters, loss functions are constructed and optimized with stochastic gradient descent. Extensive experiments on three public TREC datasets show that the new derived algorithms can significantly outperform the baselines, including the state-of-the-art relational learning to rank models.",
"Knowledge bases provide applications with the benefit of easily accessible, systematic relational knowledge but often suffer in practice from their incompleteness and lack of knowledge of new entities and relations. Much work has focused on building or extending them by finding patterns in large unannotated text corpora. In contrast, here we mainly aim to complete a knowledge base by predicting additional true relationships between entities, based on generalizations that can be discerned in the given knowledgebase. We introduce a neural tensor network (NTN) model which predicts new relationship entries that can be added to the database. This model can be improved by initializing entity representations with word vectors learned in an unsupervised fashion from text, and when doing this, existing relations can even be queried for entities that were not present in the database. Our model generalizes and outperforms existing models for this problem, and can classify unseen relationships in WordNet with an accuracy of 75.8 .",
"Semantic word spaces have been very useful but cannot express the meaning of longer phrases in a principled way. Further progress towards understanding compositionality in tasks such as sentiment detection requires richer supervised training and evaluation resources and more powerful models of composition. To remedy this, we introduce a Sentiment Treebank. It includes fine grained sentiment labels for 215,154 phrases in the parse trees of 11,855 sentences and presents new challenges for sentiment compositionality. To address them, we introduce the Recursive Neural Tensor Network. When trained on the new treebank, this model outperforms all previous methods on several metrics. It pushes the state of the art in single sentence positive negative classification from 80 up to 85.4 . The accuracy of predicting fine-grained sentiment labels for all phrases reaches 80.7 , an improvement of 9.7 over bag of features baselines. Lastly, it is the only model that can accurately capture the effects of negation and its scope at various tree levels for both positive and negative phrases.",
"Retrieving similar questions is very important in community-based question answering. A major challenge is the lexical gap in sentence matching. In this paper, we propose a convolutional neural tensor network architecture to encode the sentences in semantic space and model their interactions with a tensor layer. Our model integrates sentence modeling and semantic matching into a single model, which can not only capture the useful information with convolutional and pooling layers, but also learn the matching metrics between the question and its answer. Besides, our model is a general architecture, with no need for the other knowledge such as lexical or syntactic analysis. The experimental results shows that our method outperforms the other methods on two matching tasks.",
"Knowledge bases are an important resource for question answering and other tasks but often suffer from incompleteness and lack of ability to reason over their discrete entities and relationships. In this paper we introduce an expressive neural tensor network suitable for reasoning over relationships between two entities. Previous work represented entities as either discrete atomic units or with a single entity vector representation. We show that performance can be improved when entities are represented as an average of their constituting word vectors. This allows sharing of statistical strength between, for instance, facts involving the \"Sumatran tiger\" and \"Bengal tiger.\" Lastly, we demonstrate that all models improve when these word vectors are initialized with vectors learned from unsupervised large corpora. We assess the model by considering the problem of predicting additional true relations between entities given a subset of the knowledge base. Our model outperforms previous models and can classify unseen relationships in WordNet and FreeBase with an accuracy of 86.2 and 90.0 , respectively."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | @cite_5 proposed to detect faults based on inconsistent renaming in code clones. They tested their approach on Linux and FreeBSD where they were able to find 49 and 31 faults, respectively. | {
"cite_N": [
"@cite_5"
],
"mid": [
"2104301886"
],
"abstract": [
"Recent studies have shown that large software suites contain significant amounts of replicated code. It is assumed that some of this replication is due to copy-and-paste activity and that a significant proportion of bugs in operating systems are due to copy-paste errors. Existing static code analyzers are either not scalable to large software suites or do not perform robustly where replicated code is modified with insertions and deletions. Furthermore, the existing tools do not detect copy-paste related bugs. In this paper, we propose a tool, CP-Miner, that uses data mining techniques to efficiently identify copy-pasted code in large software suites and detects copy-paste bugs. Specifically, it takes less than 20 minutes for CP-Miner to identify 190,000 copy-pasted segments in Linux and 150,000 in FreeBSD. Moreover, CP-Miner has detected many new bugs in popular operating systems, 49 in Linux and 31 in FreeBSD, most of which have since been confirmed by the corresponding developers and have been rectified in the following releases. In addition, we have found some interesting characteristics of copy-paste in operating system code. Specifically, we analyze the distribution of copy-pasted code by size (number lines of code), granularity (basic blocks and functions), and modification within copy-pasted code. We also analyze copy-paste across different modules and various software versions."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | Jiang, Su and Chiu @cite_9 developed an approach for fault discovery using context-based inconsistencies between clones. They validated their approach on the Linux kernel and Eclipse. They also were able to detect a number of faults: 57 in the Linux kernel and 38 in Eclipse. | {
"cite_N": [
"@cite_9"
],
"mid": [
"2008085811"
],
"abstract": [
"Studies show that programs contain much similar code, commonly known as clones. One of the main reasons for introducing clones is programmers' tendency to copy and paste code to quickly duplicate functionality. We commonly believe that clones can make programs difficult to maintain and introduce subtle bugs. Although much research has proposed techniques for detecting and removing clones to improve software maintainability, little has considered how to detect latent bugs introduced by clones. In this paper, we introduce a general notion of context-based inconsistencies among clones and develop an efficient algorithm to detect such inconsistencies for locating bugs. We have implemented our algorithm and evaluated it on large open source projects including the latest versions of the Linux kernel and Eclipse. We have discovered many previously unknown bugs and programming style issues in both projects (with 57 for the Linux kernel and 38 for Eclipse). We have also categorized the bugs and style issues and noticed that they exhibit diverse characteristics and are difficult to detect with any single existing bug detection technique. We believe that our approach complements well these existing techniques."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | @cite_10 introduced the notion of : The genealogy of code clones describes how groups of code clones change over multiple versions of a program.'' This kind of approach was used in several studies to investigate if type-3 clones are associated to faults. | {
"cite_N": [
"@cite_10"
],
"mid": [
"2074529754"
],
"abstract": [
"It has been broadly assumed that code clones are inherently bad and that eliminating clones by refactoring would solve the problems of code clones. To investigate the validity of this assumption, we developed a formal denition of clone evolution and built a clone genealogy tool that automatically extracts the history of code clones from a source code repository. Using our tool we extracted clone genealogy information for two Java open source projects and analyzed their evolution. Our study contradicts some conventional wisdom about clones. In particular, refactoring may not always improve software with respect to clones for two reasons. First, many code clones exist in the system for only a short time; extensive refactoring of such short-lived clones may not be worthwhile if they are likely diverge from one another very soon. Second, many clones, especially long-lived clones that have changed consistently with other elements in the same group, are not easily refactorable due to programming language limitations. These insights show that refactoring will not help in dealing with some types of clones and open up opportunities for complementary clone maintenance tools that target these other classes of clones."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | Krinke @cite_13 analysed five open-source systems and their clone genealogies for inconsistencies between clones. He found that half of the changes in the systems' history were inconsistent. Yet, inconsistent clones did evolve mostly independently which is an indication that they did not represent faults. | {
"cite_N": [
"@cite_13"
],
"mid": [
"2145891223"
],
"abstract": [
"Code cloning is regarded as a threat to software maintenance, because it is generally assumed that a change to a code clone usually has to be applied to the other clones of the clone group as well. However, there exists little empirical data that supports this assumption. This paper presents a study on the changes applied to code clones in open source software systems based on the changes between versions of the system. It is analyzed if changes to code clones are consistent to all code clones of a clone group or not. The results show that usually half of the changes to code clone groups are inconsistent changes. Moreover, the study observes that when there are inconsistent changes to a code clone group in a near version, it is rarely the case that there are additional changes in later versions such that the code clone group then has only consistent changes."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | Bakota, Ferenc and Gyimothy @cite_27 also followed the evolution of clones of the history of an open-source system: Firefox. They found two faults that had been documented and fixed as well as five further potential faults. | {
"cite_N": [
"@cite_27"
],
"mid": [
"2152075773"
],
"abstract": [
"Although source code cloning (copy&paste programming) represents a significant threat to the maintainability of a software system, problems usually start to arise only when the system evolves. Most of the related research papers tackle the question of finding code clones in one particular version of the software only, leaving the dynamic behavior of the clones out of consideration. Eliminating these clones in large software systems often seems absolutely hopeless, as there might exist several thousands of them. Alternatively, tracking the evolution of individual clones can be used to identify those occurrences that could really cause problems in the future versions. In this paper we present an approach for mapping clones from one particular version of the software to another one, based on a similarity measure. This mapping is used to define conditions under which clones become suspicious (or \"smelly\") compared to their other occurrences. Accordingly, these conditions introduce the notion of dynamic clone smells. The usefulness of these smells is validated on the Mozilla Firefox internet browser, where the approach was able to find specific bugs that resulted from neglecting earlier copy&paste activities."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | @cite_21 investigated clone genealogies in four open-source systems. They found that below 16 results such that these late propagation are likely to be caused by bug fixes. | {
"cite_N": [
"@cite_21"
],
"mid": [
"2013638584"
],
"abstract": [
"Code cloning has been very often indicated as a bad software development practice. However, many studies appearing in the literature indicate that this is not always the case. In fact, either changes occurring in cloned code are consistently propagated, or cloning is used as a sort of templating strategy, where cloned source code fragments evolve independently. This paper (a) proposes an automatic approach to classify the evolution of source code clone fragments, and (b) reports a fine-grained analysis of clone evolution in four different Java and C software systems, aimed at investigating to what extent clones are consistently propagated or they evolve independently. Also, the paper investigates the relationship between the presence of clone evolution patterns and other characteristics such as clone radius, clone size and the kind of change the clones underwent, i.e., corrective maintenance or enhancement."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | Rahman, Bird and Devanbu @cite_0 analysed clone genealogies in four open-source projects. They found that the majority of reported defects in the projects were not associated with clones. They also found that clones are less defect-prone and faults in clones need a similar fix effort as in other code. | {
"cite_N": [
"@cite_0"
],
"mid": [
"1973650376"
],
"abstract": [
"Clones are generally considered bad programming practice in software engineering folklore. They are identified as a bad smell ( 1999) and a major contributor to project maintenance difficulties. Clones inherently cause code bloat, thus increasing project size and maintenance costs. In this work, we try to validate the conventional wisdom empirically to see whether cloning makes code more defect prone. This paper analyses the relationship between cloning and defect proneness. For the four medium to large open source projects that we studied, we find that, first, the great majority of bugs are not significantly associated with clones. Second, we find that clones may be less defect prone than non-cloned code. Third, we find little evidence that clones with more copies are actually more error prone. Fourth, we find little evidence to support the claim that clone groups that span more than one file or directory are more defect prone than collocated clones. Finally, we find that developers do not need to put a disproportionately higher effort to fix clone dense bugs. Our findings do not support the claim that clones are really a \"bad smell\" ( 1999). Perhaps we can clone, and breathe easily, at the same time."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | Barbour, Khomh and Zou @cite_28 especially investigated faulty clone genealogies and different patterns of inconsistent clones. They found that certain types of late propagations between clones are most risky to be faulty. | {
"cite_N": [
"@cite_28"
],
"mid": [
"1508590353"
],
"abstract": [
"SUMMARY Two similar code segments, or clones, form a clone pair within a software system. The changes to the clones over time create a clone evolution history. In this work, we study late propagation, a specific pattern of clone evolution. In late propagation, one clone in a clone pair is modified, causing the clone pair to diverge. The code segments are then reconciled in a later commit. Existing work has established late propagation as a clone evolution pattern and suggested that the pattern is related to a high number of faults. In this study, we examine the characteristics of late propagation in three long-lived software systems using the Simian ( Simon Harris, Victoria, Australia, http: www.harukizaemon.com simian), CCFinder, and NiCad (Software Technology Laboratory, Queen's University, Kingston, ON, Canada) clone detection tools. We define eight types of late propagation and compare them to other forms of clone evolution. Our results not only verify that late propagation is more harmful to software systems but also establish that some specific types of late propagations are more harmful than others. Specifically, two types are most risky: (1) when a clone experiences diverging changes and then a reconciling change without any modification to the other clone in a clone pair; and (2) when two clones undergo a diverging modification followed by a reconciling change that modifies both the clones in a clone pair. We also observe that the reconciliation in the former case is more prone to faults than in the latter case. We determine that the size of the clones experiencing late propagation has an effect on the fault proneness of specific types of late propagation genealogies. Lastly, we cannot report a correlation between the delay of the propagation of changes and its faults, as the fault proneness of each delay period is system dependent. Copyright © 2013 John Wiley & Sons, Ltd."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | Mondal, Roy and Schneider @cite_25 compared the fault proneness of different types of clones in a set of open-source projects. They found that type-3 clones are the most fault-prone. | {
"cite_N": [
"@cite_25"
],
"mid": [
"2171368158"
],
"abstract": [
"Code clones are defined to be the exactly or nearly similar code fragments in a software system's code-base. The existing clone related studies reveal that code clones are likely to introduce bugs and inconsistencies in the code-base. However, although there are different types of clones, it is still unknown which types of clones have a higher likeliness of introducing bugs to the software systems and so, should be considered more important for managing with techniques such as refactoring or tracking. With this focus, we performed a study that compared the bug-proneness of the major clone-types: Type 1, Type 2, and Type 3. According to our experimental results on thousands of revisions of seven diverse subject systems, Type 3 clones exhibit the highest bug-proneness among the three clone-types. The bug-proneness of Type 1 clones is the lowest. Also, Type 3 clones have the highest likeliness of being co-changed consistently while experiencing bug-fixing changes. Moreover, the Type 3 clones that experience bug-fixes have a higher possibility of evolving following a Similarity Preserving Change Pattern (SPCP) compared to the bug-fix clones of the other two clone-types. From the experimental results it is clear that Type 3 clones should be given a higher priority than the other two clone-types when making clone management decisions. We believe that our study provides useful implications for ranking clones for refactoring and tracking."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | @cite_12 is the only study known to us that analysed closed-source systems so far. They investigated four industrial systems from two companies and an academic open-source system for type-3 clones. The researchers showed all type-3 clones, which they considered to be true positives, to the developers of the systems and let them annotate them whether the inconsistency was intentional and it constitutes a fault. They found that 28 G "ode and Koschke @cite_18 investigated the clone genealogies in three open-source systems for changes to clones and unintentional inconsistent changes. They found that 57 inspected the remaining changes manually. They found that only 15 . | {
"cite_N": [
"@cite_18",
"@cite_12"
],
"mid": [
"2114056383",
"2119887272"
],
"abstract": [
"Code Clones - duplicated source fragments - are said to increase maintenance effort and to facilitate problems caused by inconsistent changes to identical parts. While this is certainly true for some clones and certainly not true for others, it is unclear how many clones are real threats to the system's quality and need to be taken care of. Our analysis of clone evolution in mature software projects shows that most clones are rarely changed and the number of unintentional inconsistent changes to clones is small. We thus have to carefully select the clones to be managed to avoid unnecessary effort managing clones with no risk potential.",
"Code cloning is not only assumed to inflate maintenance costs but also considered defect-prone as inconsistent changes to code duplicates can lead to unexpected behavior. Consequently, the identification of duplicated code, clone detection, has been a very active area of research in recent years. Up to now, however, no substantial investigation of the consequences of code cloning on program correctness has been carried out. To remedy this shortcoming, this paper presents the results of a large-scale case study that was undertaken to find out if inconsistent changes to cloned code can indicate faults. For the analyzed commercial and open source systems we not only found that inconsistent changes to clones are very frequent but also identified a significant number of faults induced by such changes. The clone detection tool used in the case study implements a novel algorithm for the detection of inconsistent clones. It is available as open source to enable other researchers to use it as basis for further investigations."
]
} |
1611.08005 | 2253512695 | Background: Code cloning - copying and reusing pieces of source code - is a common phenomenon in software development in practice. There have been several empirical studies on the effects of cloning, but there are contradictory results regarding the connection of cloning and faults. Objective: Our aim is to clarify the relationship between code clones and faults. In particular, we focus on inconsistent (or type-3) clones in this work. Method: We conducted a case study with TWT GmbH where we detected the code clones in three Java systems, set them into relation to information from issue tracking and version control and interviewed three key developers. Results: Of the type-3 clones, 17 contain faults. Developers modified most of the type-3 clones simultaneously and thereby fixed half of the faults in type-3 clones consistently. Type-2 clones with faults all evolved to fixed type-3 clones. Clone length is only weakly correlated with faultiness. Conclusion: There are indications that the developers in two cases have been aware of clones. It might be a reason for the weak relationship between type-3 clones and faults. Hence, it seems important to keep developers aware of clones, potentially with new tool support. Future studies need to investigate if the rate of faults in type-3 clones justifies using them as cues in defect detection. | @cite_1 investigated clone evolution in three open-source systems. They manually inspected all inconsistent changes themselves and judged whether the change should have been applied to all clone instances, i.e. introducing a fault. They found that of these inconsistent changes 1 faults. | {
"cite_N": [
"@cite_1"
],
"mid": [
"2067377566"
],
"abstract": [
"To study the impact of code clones on software quality, researchers typically carry out their studies based on fine-grained analysis of inconsistent changes at the revision level. As a result, they capture much of the chaotic and experimental nature inherent in any on-going software development process. Analyzing highly fluctuating and short-lived clones is likely to exaggerate the ill effects of inconsistent changes on the quality of the released software product, as perceived by the end user. To gain a broader perspective, we perform an empirical study on the effect of inconsistent changes on software quality at the release level. Based on a case study on three open source software systems, we observe that only 1.02 -4.00 of all clone genealogies introduce software defects at the release level, as opposed to the substantially higher percentages reported by previous studies at the revision level. Our findings suggest that clones do not have a significant impact on the post-release quality of the studied systems, and that the developers are able to effectively manage the evolution of cloned code."
]
} |
1611.07865 | 2950078543 | Neural Style Transfer has shown very exciting results enabling new forms of image manipulation. Here we extend the existing method to introduce control over spatial location, colour information and across spatial scale. We demonstrate how this enhances the method by allowing high-resolution controlled stylisation and helps to alleviate common failure cases such as applying ground textures to sky regions. Furthermore, by decomposing style into these perceptual factors we enable the combination of style information from multiple sources to generate new, perceptually appealing styles from existing ones. We also describe how these methods can be used to more efficiently produce large size, high-quality stylisation. Finally we show how the introduced control measures can be applied in recent methods for Fast Neural Style Transfer. | There is a large body of work on image stylisation techniques. The first example-based technique was Image Analogies @cite_25 , which built on patch-based texture synthesis techniques @cite_10 @cite_27 . This method introduced stylisation based on an example painting, as well as ways to preserve colour, and to control stylisation of different regions separately. The method used a coarse-to-fine texture synthesis procedure for speed @cite_16 . Since then, improvements to the optimisation method and new applications @cite_28 @cite_4 have been proposed. Patch-based methods have also been used with CNN features @cite_1 @cite_12 , leading to improved texture representations and stylisation results. Scale control has been developed for patch-based texture synthesis @cite_7 and many other techniques have been developed for transferring colour style @cite_9 . There are also many procedural stylisation techniques that provide extensive user control in the non-photorealistic rendering literature, e.g., @cite_19 @cite_6 @cite_15 . These procedural methods provide separate controls for adjusting spatial variation in styles, colour transformation, and brush stroke style, but cannot work from training data. | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_28",
"@cite_9",
"@cite_1",
"@cite_6",
"@cite_19",
"@cite_27",
"@cite_15",
"@cite_16",
"@cite_10",
"@cite_25",
"@cite_12"
],
"mid": [
"2470355596",
"2025630032",
"2150515446",
"2113540472",
"2952139859",
"2243570189",
"2100281410",
"",
"2170080555",
"",
"2127006916",
"",
""
],
"abstract": [
"We present an approach to example-based stylization of 3D renderings that better preserves the rich expressiveness of hand-created artwork. Unlike previous techniques, which are mainly guided by colors and normals, our approach is based on light propagation in the scene. This novel type of guidance can distinguish among context-dependent illumination effects, for which artists typically use different stylization techniques, and delivers a look closer to realistic artwork. In addition, we demonstrate that the current state of the art in guided texture synthesis produces artifacts that can significantly decrease the fidelity of the synthesized imagery, and propose an improved algorithm that alleviates them. Finally, we demonstrate our method's effectiveness on a variety of scenes and styles, in applications like interactive shading study or autocompletion.",
"Example-based texture synthesis algorithms have gained widespread popularity for their ability to take a single input image and create a perceptually similar non-periodic texture. However, previous methods rely on single input exemplars that can capture only a limited band of spatial scales. For example, synthesizing a continent-like appearance at a variety of zoom levels would require an impractically high input resolution. In this paper, we develop a multiscale texture synthesis algorithm. We propose a novel example-based representation, which we call an exemplar graph, that simply requires a few low-resolution input exemplars at different scales. Moreover, by allowing loops in the graph, we can create infinite zooms and infinitely detailed textures that are impossible with current example-based methods. We also introduce a technique that ameliorates inconsistencies in the user's input, and show that the application of this method yields improved interscale coherence and higher visual quality. We demonstrate optimizations for both CPU and GPU implementations of our method, and use them to produce animations with zooming and panning at multiple scales, as well as static gigapixel-sized images with features spanning many spatial scales.",
"This paper describes CMS (constrained minimization synthesis), a fast, robust texture synthesis algorithm that creates output textures while satisfying constraints. We show that constrained texture synthesis can be posed in a principled way as an energy minimization problem that requires balancing two measures of quality: constraint satisfaction and texture seamlessness. We then present an efficient algorithm for finding good solutions to this problem using an adaptation of graphcut energy minimization. CMS is particularly well suited to detail synthesis, the process of adding high-resolution detail to low-resolution images. It also supports the full image analogies framework, while providing superior image quality and performance. CMS is easily extended to handle multiple constraints on a single output, thus enabling novel applications that combine both user-specified and image-based control",
"The objective of colour mapping or colour transfer methods is to recolour a given image or video by deriving a mapping between that image and another image serving as a reference. These methods have received considerable attention in recent years, both in academic literature and industrial applications. Methods for recolouring images have often appeared under the labels of colour correction, colour transfer or colour balancing, to name a few, but their goal is always the same: mapping the colours of one image to another. In this paper, we present a comprehensive overview of these methods and offer a classification of current solutions depending not only on their algorithmic formulation but also their range of applications. We also provide a new dataset and a novel evaluation technique called 'evaluation by colour mapping roundtrip'. We discuss the relative merit of each class of techniques through examples and show how colour mapping solutions can have been applied to a diverse range of problems.",
"This paper studies a combination of generative Markov random field (MRF) models and discriminatively trained deep convolutional neural networks (dCNNs) for synthesizing 2D images. The generative MRF acts on higher-levels of a dCNN feature pyramid, controling the image layout at an abstract level. We apply the method to both photographic and non-photo-realistic (artwork) synthesis tasks. The MRF regularizer prevents over-excitation artifacts and reduces implausible feature mixtures common to previous dCNN inversion approaches, permitting synthezing photographic content with increased visual plausibility. Unlike standard MRF-based texture synthesis, the combined system can both match and adapt local features with considerable variability, yielding results far out of reach of classic generative MRF methods.",
"We present a system that lets a designer directly annotate a 3D model with strokes, imparting a personal aesthetic to the non-photorealistic rendering of the object. The artist chooses a \"brush\" style, then draws strokes over the model from one or more viewpoints. When the system renders the scene from any new viewpoint, it adapts the number and placement of the strokes appropriately to maintain the original look.",
"Current digital painting tools are primarily targeted at professionals and are often overwhelmingly complex for use by novices. At the same time, simpler tools may not invoke the user creatively, or are limited to plain styles that lack visual sophistication. There are many people who are not art professionals, yet would like to partake in digital creative expression. Challenges and rewards for novices differ greatly from those for professionals. In this paper, we leverage existing works in Creativity and Creativity Support Tools (CST) to formulate design goals specifically for digital art creation tools for novices. We implemented these goals within a digital painting system, called Painting with Bob. We evaluate the efficacy of the design and our prototype with a user study, and we find that users are highly satisfied with the user experience, as well as the paintings created with our system.",
"",
"This paper presents an interactive system for creating painterly animation from video sequences. Previous approaches to painterly animation typically emphasize either purely automatic stroke synthesis or purely manual stroke key framing. Our system supports a spectrum of interaction between these two approaches which allows the user more direct control over stroke synthesis. We introduce an approach for controlling the results of painterly animation: keyframed Control Strokes can affect automatic stroke's placement, orientation, movement, and color. Furthermore, we introduce a new automatic synthesis algorithm that traces strokes through a video sequence in a greedy manner, but, instead of a vector field, uses an objective function to guide placement. This allows the method to capture fine details, respect region boundaries, and achieve greater temporal coherence than previous methods. All editing is performed with a WYSIWYG interface where the user can directly refine the animation. We demonstrate a variety of examples using both automatic and user-guided results, with a variety of styles and source videos.",
"",
"We present a universal statistical model for texture images in the context of an overcomplete complex wavelet transform. The model is parameterized by a set of statistics computed on pairs of coefficients corresponding to basis functions at adjacent spatial locations, orientations, and scales. We develop an efficient algorithm for synthesizing random images subject to these constraints, by iteratively projecting onto the set of images satisfying each constraint, and we use this to test the perceptual validity of the model. In particular, we demonstrate the necessity of subgroups of the parameter set by showing examples of texture synthesis that fail when those parameters are removed from the set. We also demonstrate the power of our model by successfully synthesizing examples drawn from a diverse collection of artificial and natural textures.",
"",
""
]
} |
1611.07865 | 2950078543 | Neural Style Transfer has shown very exciting results enabling new forms of image manipulation. Here we extend the existing method to introduce control over spatial location, colour information and across spatial scale. We demonstrate how this enhances the method by allowing high-resolution controlled stylisation and helps to alleviate common failure cases such as applying ground textures to sky regions. Furthermore, by decomposing style into these perceptual factors we enable the combination of style information from multiple sources to generate new, perceptually appealing styles from existing ones. We also describe how these methods can be used to more efficiently produce large size, high-quality stylisation. Finally we show how the introduced control measures can be applied in recent methods for Fast Neural Style Transfer. | However, the representation of image style within the parametric neural texture model @cite_30 allows far less intuitive control over the stylisation outcome than patch-based methods. The texture parameters can be used to influence the stylisation but their interplay is extremely complex due to the complexity of the deep representations they are defined on. Therefore it is difficult to predict their perceptual effect on the stylisation result. Our main goal in this work is to introduce intuitive ways to control Neural Style Transfer to combine the advantages of that method with the more fine-grained user control of earlier stylisation methods. Note that concurrent work @cite_14 independently developed a similar approach for spatial control as presented here. | {
"cite_N": [
"@cite_30",
"@cite_14"
],
"mid": [
"2161208721",
"2585235684"
],
"abstract": [
"Here we introduce a new model of natural textures based on the feature spaces of convolutional neural networks optimised for object recognition. Samples from the model are of high perceptual quality demonstrating the generative power of neural networks trained in a purely discriminative fashion. Within the model, textures are represented by the correlations between feature maps in several layers of the network. We show that across layers the texture representations increasingly capture the statistical properties of natural images while making object information more and more explicit. The model provides a new tool to generate stimuli for neuroscience and might offer insights into the deep representations learned by convolutional neural networks.",
"Recently, methods have been proposed that perform texture synthesis and style transfer by using convolutional neural networks (e.g. [2015,2016]). These methods are exciting because they can in some cases create results with state-of-the-art quality. However, in this paper, we show these methods also have limitations in texture quality, stability, requisite parameter tuning, and lack of user controls. This paper presents a multiscale synthesis pipeline based on convolutional neural networks that ameliorates these issues. We first give a mathematical explanation of the source of instabilities in many previous approaches. We then improve these instabilities by using histogram losses to synthesize textures that better statistically match the exemplar. We also show how to integrate localized style losses in our multiscale framework. These losses can improve the quality of large features, improve the separation of content and style, and offer artistic controls such as paint by numbers. We demonstrate that our approach offers improved quality, convergence in fewer iterations, and more stability over the optimization."
]
} |
1611.08037 | 2950583455 | We consider an orienteering problem (OP) where an agent needs to visit a series (possibly a subset) of depots, from which the maximal accumulated profits are desired within given limited time budget. Different from most existing works where the profits are assumed to be static, in this work we investigate a variant that has arbitrary time-dependent profits. Specifically, the profits to be collected change over time and they follow different (e.g., independent) time-varying functions. The problem is of inherent nonlinearity and difficult to solve by existing methods. To tackle the challenge, we present a simple and effective framework that incorporates time-variations into the fundamental planning process. Specifically, we propose a deterministic spatio-temporal representation where both spatial description and temporal logic are unified into one routing topology. By employing existing basic sorting and searching algorithms, the routing solutions can be computed in an extremely efficient way. The proposed method is easy to implement and extensive numerical results show that our approach is time efficient and generates near-optimal solutions. | Although in general OPs have been well researched, the time-dependent OPs (TOPs) have received relatively less attention comparing with other variants @cite_8 @cite_15 . Even so, most of existing TOPs discuss the time-varying properties that are associated with the real travel time between pairwise nodes, and assume that travel time between two nodes depends on the departure time at the first (or an earlier) node @cite_24 @cite_7 @cite_5 @cite_4 . Very rarely we could find the works that discuss about time-varying scores of OPs. One work that share certain similarity with this proposed problem is @cite_9 , where multiple vehicles need to serve a number of clients and the profit of each client follows a decreasing function of time. The work analyzed a lower bound and upper bound based on a classic MIP formulation. | {
"cite_N": [
"@cite_4",
"@cite_7",
"@cite_8",
"@cite_9",
"@cite_24",
"@cite_5",
"@cite_15"
],
"mid": [
"2059762964",
"2039097929",
"1679712078",
"",
"",
"2073588680",
"2346699423"
],
"abstract": [
"This paper introduces a fast solution procedure to solve 100-node instances of the time-dependent orienteering problem (TD-OP) within a few seconds of computation time. Orienteering problems occur in logistic situations were an optimal combination of locations needs to be selected and the routing between the selected locations needs to be optimized. In the time-dependent variant, the travel time between two locations depends on the departure time at the first location. Next to a mathematical formulation of the TD-OP, the main contribution of this paper is the design of a fast and effective algorithm to tackle this problem. This algorithm combines the principles of an ant colony system (ACS) with a time-dependent local search procedure equipped with a local evaluation metric. Additionally, realistic benchmark instances with varying size and properties are constructed. The average score gap with the known optimal solution on these test instances is only 1.4 with an average computation time of 0.5seconds. An extensive sensitivity analysis shows that the performance of the algorithm is insensitive to small changes in its parameter settings.",
"The time-dependent orienteering problem is dual to the time-dependent traveling salesman problem. It consists of visiting a maximum number of sites within a given deadline. The traveling time between two sites is in general dependent on the starting time.For any ɛ > 0, we provide a (2 + ɛ)-approximation algorithm for the time-dependent orienteering problem which runs in polynomial time if the ratio between the maximum and minimum traveling time between any two sites is constant. No prior upper approximation bounds were known for this time-dependent problem.",
"The Time Dependent Orienteering Problem with Time Windows (TDOPTW) consists of a set of locations with associated time windows and scores Visiting a location allows to collect its score as a reward Traveling time between locations varies depending on the leave time The objective is to obtain a route that maximizes the obtained score within a limited amount of time In this paper we target the use of public transportation in a city, where users may move on foot or by public transportation The approach can also be applied to the logistic sector, for example to the multimodal freight transportation We apply an hybrid approach to tackle the problem Experimental results for the city of San Sebastian show we are able to obtain valid routes in real-time.",
"",
"",
"Vehicle Routing Problems have been extensively analyzed to reduce transportation costs. More particularly, the Vehicle Routing Problem with Time Windows (VRPTW) imposes the period of time of customer availability as a constraint, a common characteristic in real world situations. Using minimization of the total distance as the main objective to be fulfilled, this work implements an efficient algorithm which associates non-monotonic Simulated Annealing to Hill-Climbing and Random Restart. The algorithm is compared to the best results published in the literature for the 56 Solomon instances and it is shown how statistical methods can be used to boost the performance of the method.",
"The Orienteering Problem (OP) has received a lot of attention in the past few decades. The OP is a routing problem in which the goal is to determine a subset of nodes to visit, and in which order, so that the total collected score is maximized and a given time budget is not exceeded. A number of typical variants has been studied, such as the Team OP, the (Team) OP with Time Windows and the Time Dependent OP. Recently, a number of new variants of the OP was introduced, such as the Stochastic OP, the Generalized OP, the Arc OP, the Multi-agent OP, the Clustered OP and others. This paper focuses on a comprehensive and thorough survey of recent variants of the OP, including the proposed solution approaches. Moreover, the OP has been used as a model in many different practical applications. The most recent applications of the OP, such as the Tourist Trip Design Problem and the mobile-crowdsourcing problem are discussed. Finally, we also present some promising topics for future research."
]
} |
1611.07808 | 2552783079 | A unit disk graph is the intersection graph of a set of unit diameter disks in the plane. In this paper we consider liar's domination problem on unit disk graphs, a variant of dominating set problem. We call this problem as Euclidean liar's domination problem . In the Euclidean liar's domination problem, a set @math of @math points (disk centers) are given in the Euclidean plane. For @math , @math is a subset of @math such that for any @math , the Euclidean distance between @math and @math is less than or equal to 1, i.e., the corresponding unit diameter disks intersect. The objective of the Euclidean liar's domination problem is to find a subset @math of minimum size having the following properties : (i) @math for @math , and (ii) @math for @math . This article aims to prove the Euclidean liar's domination problem is NP-complete. | The liar's domination problem is introduced by Slater in 2009 and showed that the problem is NP-hard for general graphs @cite_2 . Later, Roden and Slater showed that the problem is NP-hard even for bipartite graphs @cite_6 . Panda and Paul @cite_3 proved that the problem is NP-hard for split graphs and chordal graphs. The authors also proposed a linear time algorithm for computing a minimum cardinality liar's dominating set in a tree. | {
"cite_N": [
"@cite_3",
"@cite_6",
"@cite_2"
],
"mid": [
"2052665499",
"2089132532",
""
],
"abstract": [
"A set L@?V(G) of a graph G=(V,E) is a liar's dominating set if (1) for all v@?V(G), |N\"G[v]@?L|>=2 and (2) for every pair u,v@?V(G) of distinct vertices, |(N\"G[u]@?N\"G[v])@?L|>=3. A graph G=(V,E) admits a liar's dominating set if each of its connected component has at least three vertices. Given a graph G=(V,E) and an integer K, the liar's domination decision problem (LR-DOMDP) is to decide whether G has a liar's dominating set of cardinality at most K. Slater [P.J. Slater, Liar's Domination, Networks, 54(2) (2009) 70-74] proved that the LR-DOMDP is NP-complete for general graphs. Subsequently, Roden and Slater [M.L. Roden and P.J. Slater, Liar's domination in graphs, Discrete Math., 309(19) (2009) 5884-5890] showed a more general family of problems to each be NP-complete for bipartite graphs. Besides this result, no other algorithmic result for the liar's dominating set problem is available in the literature. In this paper, we first strengthen the complexity result of the LR-DOMDP by showing that this problem remains NP-complete for split graphs and hence for chordal graphs. Finally, we propose a linear time algorithm for computing a minimum cardinality liar's dominating set in a tree.",
"Assume that each vertex of a graph G is the possible location for an “intruder” such as a thief, or a saboteur, a fire in a facility or some possible processor fault in a computer network. A device at a vertex v is assumed to be able to detect the intruder at any vertex in its closed neighborhood N[v] and to identify at which vertex in N[v] the intruder is located. One must then have a dominating set S ⊆ V(G), a set with ∪v∈SN[v] = V(G), to be able to identify any intruder's location. If any one device can fail to detect the intruder, then one needs a double-dominating set. This article introduces the study of liar's dominating sets, sets that can identify an intruder's location even when any one device in the neighborhood of the intruder vertex can lie, that is, any one device in the neighborhood of the intruder vertex can misidentify any vertex in its closed neighborhood as the intruder location. Liar's dominating sets lie between double-dominating sets and triple-dominating sets because every triple-dominating set is a liar's dominating set, and every liar's dominating set must double dominate. © 2009 Wiley Periodicals, Inc. NETWORKS, 2009",
""
]
} |
1611.07379 | 2550024422 | Selling reserved instances (or virtual machines) is a basic service in cloud computing. In this paper, we consider a more flexible pricing model for instance reservation, in which a customer can propose the time length and number of resources of her request, while in today's industry, customers can only choose from several predefined reservation packages. Under this model, we design randomized mechanisms for customers coming online to optimize social welfare and providers' revenue. We first consider a simple case, where the requests from the customers do not vary too much in terms of both length and value density. We design a randomized mechanism that achieves a competitive ratio @math for both and , which is a improvement as there is usually no revenue guarantee in previous works such as azar2015ec,wang2015selling . This ratio can be improved up to @math when we impose a realistic constraint on the maximum number of resources used by each request. On the hardness side, we show an upper bound @math on competitive ratio for any randomized mechanism. We then extend our mechanism to the general case and achieve a competitive ratio @math for both social welfare and revenue, where @math is the ratio of the maximum request length to the minimum request length and @math is the ratio of the maximum request value density to the minimum request value density. This result outperforms the previous upper bound @math for deterministic mechanisms wang2015selling . We also prove an upper bound @math for any randomized mechanism. All the mechanisms we provide are in a greedy style. They are truthful and easy to be integrated into practical cloud systems. | In @cite_15 , the authors study the same model as us. Truthful deterministic mechanisms are designed under the condition that the number of resources used by each reservation is constrained. For general case, an upper bound @math for deterministic algorithms is provided. As aforementioned, their mechanisms do not provide any guarantee to provider's revenue. | {
"cite_N": [
"@cite_15"
],
"mid": [
"2261897571"
],
"abstract": [
"In this paper, we study the problem of designing new mechanisms for selling reserved instances (also referred to as virtual machines) in cloud computing. Unlike the practice in today's clouds in which users only have a few predefined options to reserve instances (i.e., either 1-year reservation or 3-year reservation), we allow users to reserve resources for any length and from any time point in the future. Our goal is to maximize the social welfare. We propose two mechanisms, one for the case where all the jobs are tight (their lengths are exactly their reservation time intervals), and the other for the more general case where jobs are delayable and have some flexibility on their reservations. Both of the mechanisms are prompt in the sense that the acceptance and the payment for a job is determined at the very moment of its arrival. We use competitive analysis to evaluate the performance of our mechanisms, and show that both of the mechanisms have a competitive ratio of O(ln(kT)) under some mild assumption, where k (res. T) is the maximum ratio between per-instance-hour valuation (res. length) of any two jobs. We then prove that no algorithm can achieve a competitive ratio better than ln(2kT) under the same assumption. Therefore, our mechanisms are optimal within a constant factor."
]
} |
1611.07379 | 2550024422 | Selling reserved instances (or virtual machines) is a basic service in cloud computing. In this paper, we consider a more flexible pricing model for instance reservation, in which a customer can propose the time length and number of resources of her request, while in today's industry, customers can only choose from several predefined reservation packages. Under this model, we design randomized mechanisms for customers coming online to optimize social welfare and providers' revenue. We first consider a simple case, where the requests from the customers do not vary too much in terms of both length and value density. We design a randomized mechanism that achieves a competitive ratio @math for both and , which is a improvement as there is usually no revenue guarantee in previous works such as azar2015ec,wang2015selling . This ratio can be improved up to @math when we impose a realistic constraint on the maximum number of resources used by each request. On the hardness side, we show an upper bound @math on competitive ratio for any randomized mechanism. We then extend our mechanism to the general case and achieve a competitive ratio @math for both social welfare and revenue, where @math is the ratio of the maximum request length to the minimum request length and @math is the ratio of the maximum request value density to the minimum request value density. This result outperforms the previous upper bound @math for deterministic mechanisms wang2015selling . We also prove an upper bound @math for any randomized mechanism. All the mechanisms we provide are in a greedy style. They are truthful and easy to be integrated into practical cloud systems. | There are many works that concern online resource allocation for cloud computing @cite_18 @cite_24 @cite_1 @cite_13 @cite_16 @cite_21 . Among these works, the pay-as-you-go model is considered in @cite_13 , and authors in @cite_24 propose a framework for cloud resource allocation when agents' value functions are continuous and concave. In @cite_16 , a more general scheduling problem is studied. The authors also take both social welfare and revenue into consideration. However, the their competitive ratios are related to the total amount of resources. In @cite_21 , the authors propose online strategies to reserve instances without any a priori knowledge of future demands. Those strategies are optimal when considering cost management. | {
"cite_N": [
"@cite_18",
"@cite_21",
"@cite_1",
"@cite_24",
"@cite_16",
"@cite_13"
],
"mid": [
"2087111544",
"2051996845",
"2125053293",
"",
"",
"2310680702"
],
"abstract": [
"Current cloud computing providers allocate their virtual machine (VM) instances via fixed price-based or auction-like mechanisms. However, these mechanisms have one limitation, they are all offline mechanisms, therefore they need to collect information and be invoked periodically. In this paper, we address this limitation by designing an online mechanism for dynamic provisioning and allocation of VM instances in clouds. Our proposed mechanism, MOVMPA, is invoked as soon as a user places a request or some VM instances already allocated become available again. When invoked, the mechanism selects users who would be allocated VM instances for the period they requested for, and ensures that those users will continue using those VMs for the entire period requested. We prove that the mechanism is incentive compatible and also investigate its performance through extensive simulation experiments.",
"Infrastructure-as-a-service (IaaS) clouds offer diverse instance purchasing options. A user can either run instances on demand and pay only for what it uses, or it can prepay to reserve instances for a long period, during which a usage discount is entitled. An important problem facing a user is how these two instance options can be dynamically combined to serve time-varying demands at minimum cost. Existing strategies in the literature, however, require either exact knowledge or the distribution of demands in the long-term future, which significantly limits their use in practice. Unlike existing works, we propose two practical online algorithms , one deterministic and another randomized, that dynamically combine the two instance options online without any knowledge of the future. We show that the proposed deterministic (resp., randomized) algorithm incurs no more than @math (resp., @math ) times the minimum cost obtained by an optimal offline algorithm that knows the exact future a priori , where @math is the entitled discount after reservation. Our online algorithms achieve the best possible competitive ratios in both the deterministic and randomized cases, and can be easily extended to cases when short-term predictions are reliable. Simulations driven by a large volume of real-world traces show that significant cost savings can be achieved with prevalent IaaS prices.",
"Auction mechanisms have recently attracted substantial attention as an efficient approach to pricing and resource allocation in cloud computing. This work, to the authors' knowledge, represents the first online combinatorial auction designed in the cloud computing paradigm, which is general and expressive enough to both (a) optimize system efficiency across the temporal domain instead of at an isolated time point, and (b) model dynamic provisioning of heterogeneous Virtual Machine (VM) types in practice. The final result is an online auction framework that is truthful, computationally efficient, and guarantees a competitive ratio e+ 1 over e-1 3.30 in social welfare in typical scenarios. The framework consists of three main steps: (1) a tailored primal-dual algorithm that decomposes the long-term optimization into a series of independent one-shot optimization problems, with an additive loss of 1 over e-1 in competitive ratio, (2) a randomized auction sub-framework that applies primal-dual optimization for translating a centralized co-operative social welfare approximation algorithm into an auction mechanism, retaining a similar approximation ratio while adding truthfulness, and (3) a primal-dual update plus dual fitting algorithm for approximating the one-shot optimization with a ratio λ close to e. The efficacy of the online auction framework is validated through theoretical analysis and trace-driven simulation studies. We are also in the hope that the framework, as well as its three independent modules, can be instructive in auction design for other related problems.",
"",
"",
"Cloud providers provision their various resources such as CPUs, memory, and storage in the form of virtual machine (VM) instances which are then allocated to the users. The users are charged based on a pay-as-you-go model, and their payments should be determined by considering both their incentives and the incentives of the cloud providers. Auction markets can capture such incentives, where users name their own prices for their requested VMs. We design an auction-based online mechanism for VM provisioning, allocation, and pricing in clouds that considers several types of resources. Our proposed online mechanism makes no assumptions about future demand of VMs, which is the case in real cloud settings. The proposed online mechanism is invoked as soon as a user places a request or some of the allocated resources are released and become available. The mechanism allocates VM instances to selected users for the period they are requested for, and ensures that the users will continue using their VM instances for the entire requested period. In addition, the mechanism determines the payment the users have to pay for using the allocated resources. We prove that the mechanism is incentive-compatible, that is, it gives incentives to the users to reveal their actual requests. We investigate the performance of our proposed mechanism through extensive experiments."
]
} |
1611.07379 | 2550024422 | Selling reserved instances (or virtual machines) is a basic service in cloud computing. In this paper, we consider a more flexible pricing model for instance reservation, in which a customer can propose the time length and number of resources of her request, while in today's industry, customers can only choose from several predefined reservation packages. Under this model, we design randomized mechanisms for customers coming online to optimize social welfare and providers' revenue. We first consider a simple case, where the requests from the customers do not vary too much in terms of both length and value density. We design a randomized mechanism that achieves a competitive ratio @math for both and , which is a improvement as there is usually no revenue guarantee in previous works such as azar2015ec,wang2015selling . This ratio can be improved up to @math when we impose a realistic constraint on the maximum number of resources used by each request. On the hardness side, we show an upper bound @math on competitive ratio for any randomized mechanism. We then extend our mechanism to the general case and achieve a competitive ratio @math for both social welfare and revenue, where @math is the ratio of the maximum request length to the minimum request length and @math is the ratio of the maximum request value density to the minimum request value density. This result outperforms the previous upper bound @math for deterministic mechanisms wang2015selling . We also prove an upper bound @math for any randomized mechanism. All the mechanisms we provide are in a greedy style. They are truthful and easy to be integrated into practical cloud systems. | In @cite_8 @cite_19 @cite_5 , mechanisms design for scheduling problem with commitments is studied. Those mechanisms proposed either complete a reservation or reject it when there is enough time for this reservation to be completed before deadline. In @cite_19 , the authors study offline settings and design a near-optimal mechanism with commitments. In @cite_8 , a heuristic truthful mechanism for online scheduling is proposed, but no formal bound of competitive ratio is given. In @cite_5 , the authors follow @cite_8 and design truthful online mechanisms with a constant competitive when reservations are enough. | {
"cite_N": [
"@cite_19",
"@cite_5",
"@cite_8"
],
"mid": [
"2122985136",
"2092461546",
""
],
"abstract": [
"We consider a market-based resource allocation model for batch jobs in cloud computing clusters. In our model, we incorporate the importance of the due date of a job rather than the number of servers allocated to it at any given time. Each batch job is characterized by the work volume of total computing units (e.g., CPU hours) along with a bound on maximum degree of parallelism. Users specify, along with these job characteristics, their desired due date and a value for finishing the job by its deadline. Given this specification, the primary goal is to determine the scheduling of cloud computing instances under capacity constraints in order to maximize the social welfare (i.e., sum of values gained by allocated users). Our main result is a new ( C (C-k) ⋅ s (s-1))-approximation algorithm for this objective, where C denotes cloud capacity, k is the maximal bound on parallelized execution (in practical settings, k l C) and s is the slackness on the job completion time i.e., the minimal ratio between a specified deadline and the earliest finish time of a job. Our algorithm is based on utilizing dual fitting arguments over a strengthened linear program to the problem. Based on the new approximation algorithm, we construct truthful allocation and pricing mechanisms, in which reporting the job true value and properties (deadline, work volume and the parallelism bound) is a dominant strategy for all users. To that end, we provide a general framework for transforming allocation algorithms into truthful mechanisms in domains of single-value and multi-properties. We then show that the basic mechanism can be extended under proper Bayesian assumptions to the objective of maximizing revenues, which is important for public clouds. We empirically evaluate the benefits of our approach through simulations on data-center job traces, and show that the revenues obtained under our mechanism are comparable with an ideal fixed-price mechanism, which sets an on-demand price using oracle knowledge of users' valuations. Finally, we discuss how our model can be extended to accommodate uncertainties in job work volumes, which is a practical challenge in cloud settings.",
"We study online mechanisms for preemptive scheduling with deadlines, with the goal of maximizing the total value of completed jobs. This problem is fundamental to deadline-aware cloud scheduling, but there are strong lower bounds even for the algorithmic problem without incentive constraints. However, these lower bounds can be circumvented under the natural assumption of deadline slackness, i.e., that there is a guaranteed lower bound s > 1 on the ratio between a job's size and the time window in which it can be executed. In this paper, we construct a truthful scheduling mechanism with a constant competitive ratio, given slackness s > 1. Furthermore, we show that if s is large enough then we can construct a mechanism that also satisfies a commitment property: it can be determined whether or not a job will finish, and the requisite payment if so, well in advance of each job's deadline. This is notable because, in practice, users with strict deadlines may find it unacceptable to discover only very close to their deadline that their job has been rejected.",
""
]
} |
1611.07379 | 2550024422 | Selling reserved instances (or virtual machines) is a basic service in cloud computing. In this paper, we consider a more flexible pricing model for instance reservation, in which a customer can propose the time length and number of resources of her request, while in today's industry, customers can only choose from several predefined reservation packages. Under this model, we design randomized mechanisms for customers coming online to optimize social welfare and providers' revenue. We first consider a simple case, where the requests from the customers do not vary too much in terms of both length and value density. We design a randomized mechanism that achieves a competitive ratio @math for both and , which is a improvement as there is usually no revenue guarantee in previous works such as azar2015ec,wang2015selling . This ratio can be improved up to @math when we impose a realistic constraint on the maximum number of resources used by each request. On the hardness side, we show an upper bound @math on competitive ratio for any randomized mechanism. We then extend our mechanism to the general case and achieve a competitive ratio @math for both social welfare and revenue, where @math is the ratio of the maximum request length to the minimum request length and @math is the ratio of the maximum request value density to the minimum request value density. This result outperforms the previous upper bound @math for deterministic mechanisms wang2015selling . We also prove an upper bound @math for any randomized mechanism. All the mechanisms we provide are in a greedy style. They are truthful and easy to be integrated into practical cloud systems. | In this work, the number of resources requested by a reservation is fixed and determined by the customer. However, in some scenarios, the customer only provides the total size (number of resources times length) of her reservation, and it is the cloud scheduler that determines how many resources are allocated to process this reservation. Those reservations are called malleable reservations. There are some works in this fields, such as @cite_2 @cite_22 . | {
"cite_N": [
"@cite_22",
"@cite_2"
],
"mid": [
"2054626117",
"2048210274"
],
"abstract": [
"In this paper, we study online algorithms that schedule malleable jobs, i.e., jobs that can be parallelized on any subset of the available @math m identical machines. We study a model that accounts for the tradeoff between multiprocessor speedup and overhead time, namely, if job @math j has processing requirement @math pj and is assigned to run on @math kj machines, then @math j's execution time becomes @math pj kj+(kj-1)c, where @math c is a constant parameter to the problem. For @math m=2, we present an online algorithm OCS that has a strong competitive ratio of 3 2, matching a previously established lower bound. We also present an online algorithm ASYM2 that is asymptotically @math ((4-∈) (3-∈))-competitive when @math m=2, where @math 0<∈≤2 is a parameter to the algorithm, improving upon an existing asymptotically (3 2)-competitive algorithm. Finally, we present an online algorithm OTO that is strongly @math 2-competitive when @math m=3, improving upon the previous best upper bound of @math 9 4.",
"We consider the online scheduling of malleable jobs on parallel systems, such as clusters, symmetric multiprocessing computers, and multi-core processor computers. Malleable jobs is a model of parallel processing in which jobs adapt to the number of processors assigned to them. This model permits the scheduler and resource manager to make more efficient use of the available resources. Each malleable job is characterized by arrival time, deadline, and value. If the job completes by its deadline, the user earns the payoff indicated by the value; otherwise, she earns a payoff of zero. The scheduling objective is to maximize the sum of the values of the jobs that complete by their associated deadlines. Complicating the matter is that users in the real world are rational and they will attempt to manipulate the scheduler by misreporting their jobs' parameters if it benefits them to do so. To mitigate this behavior, we design an incentive compatible online scheduling mechanism. Incentive compatibility assures us that the users will obtain the maximum payoff only if they truthfully report their jobs' parameters to the scheduler. Finally, we simulate and study the mechanism to show the effects of misreports on the cheaters and on the system."
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | The work at hand enhances previous results published in by investigating how quality models help in understanding data gained from static analysis. We focus on applying static quality analysis to leverage SMEs mitigating their risks of defect-related costs. Deviating from our aim, the research community devotes its attention primarily to software process improvement (SPI) in SMEs. @cite_26 and @cite_28 @cite_6 report on SPI introduced in many SMEs from the late 90's till 2006 to assess software processes, change the organisations and increase their software productivity. However, they do not investigate the impact of SPI on software product quality attributes at all. Even the process improvements are often only measured by means of informal and non-objective processes based on SME employees' perceptions. @cite_28 conclude that SEI CMM(i) and ISO SPICE are difficult to be applied to SMEs. | {
"cite_N": [
"@cite_28",
"@cite_26",
"@cite_6"
],
"mid": [
"1988388703",
"2113981284",
""
],
"abstract": [
"Small and medium enterprises are a very important cog in the gears of the world economy. The software industry in most countries is composed of an industrial scheme that is made up mainly of small and medium software enterprises--SMEs. To strengthen these types of organizations, efficient Software Engineering practices are needed--practices which have been adapted to their size and type of business. Over the last two decades, the Software Engineering community has expressed special interest in software process improvement (SPI) in an effort to increase software product quality, as well as the productivity of software development. However, there is a widespread tendency to make a point of stressing that the success of SPI is only possible for large companies. In this article, a systematic review of published case studies on the SPI efforts carried out in SMEs is presented. Its objective is to analyse the existing approaches towards SPI which focus on SMEs and which report a case study carried out in industry. A further objective is that of discussing the significant issues related to this area of knowledge, and to provide an up-to-date state of the art, from which innovative research activities can be thought of and planned.",
"Software process improvement is a demanding and complex undertaking. To support the constitution and implementation of software process improvement schemes the Software Engineering Institute (SEI) proposes a framework, the so-called IDEAL model. This model is based on experiences from large organizations. The aim of the research described here was to investigate the suitability of the model for small software enterprises. It has therefore been deployed and adjusted for successful use in a small Danish software company. The course of the project and the application of the model are presented and the case is reflected on the background of current knowledge about managing software process improvement as organizational change.",
""
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | report on application of clone detection to a large telecommunication software system. They restricted the technique to the comparison of whole methods. Opposed to that, we allowed arbitrary code fragments to be compared with each other but analysed smaller systems. Nevertheless, we confirm that clone detection can improve software maintainability. @cite_19 performed clone detection in different software artefacts like source code, requirements specifications and models at five enterprises. He presents a large case study investigating the impact of cloning on program correctness, an analytical cost model that quantifies the impact of code cloning on maintenance activities and a comprehensive method for clone control and tool support for practical use. | {
"cite_N": [
"@cite_19"
],
"mid": [
"2765706740"
],
"abstract": [
"Software cloning is a well-known software quality defect. Nevertheless, it abounds in practice -- across artifacts, organizations and domains. This has two principal reasons: first, the significance of cloning is not well understood. Second, no comprehensive method exists that guides practitioners to establish successful clone control. This thesis contributes to both areas. First, we present empirical results on the significance of cloning. Our case studies on source code, requirements specifications and Matlab Simulink models demonstrate that cloning needs to be controlled to avoid its negative impact on correctness and costs. Second, we present a comprehensive method for clone control and demonstrate its effectiveness through an industrial case study, where it successfully reduced the amount of cloning in the participating system. Finally, we identify the limitations of clone detection and control by a controlled experiment."
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | Based on an industrial robot control system ( @math 2500 kLoC), @cite_4 encountered drawbacks of generic ASA, i.e. low effectiveness, too many violations or false positives, and lack of result verifiability. They propose ASA to be tailored by system-specific rule selection to reduce false positives and ease violation handling. This includes the analysis of annotations in comments, C macros, coding constraints specific to files or methods (i.e. call structure, control or data flow patterns), and constraints on thread scheduling or shared data usage. Accordingly, @cite_23 report an overrating of many maintainability metrics or predictors---often represented by bug pattern rules---and suggest more sophisticated evaluation of their system-level impacts. For the positive impact on software quality characteristics, they conjecture: Individual rule selection is better than taking generic tool presets. They did not empirically validate this conjecture. But we qualitatively confirm it through our results on RQ 2.1 which also motivated rule selection as a cornerstone of our usage guidelines. | {
"cite_N": [
"@cite_4",
"@cite_23"
],
"mid": [
"2047625679",
"2134058295"
],
"abstract": [
"In this paper, we are exploring the approach to utilize system-specific static analyses of code with the goal to improve software quality for specific software systems. Specialized analyses, tailored for a particular system, make it possible to take advantage of system domain knowledge that is not available to more generic analyses. Furthermore, analyses can be selected and or developed in order to best meet the challenges and specific issues of the system at hand. As a result, such analyses can be used as a complement to more generic code analysis tools because they are likely to have a better impact on (business) concerns such as improving certain software quality attributes and reducing certain classes of failures. We present a case study of a large, industrial embedded system, giving examples of what kinds of analyses could be realized and demonstrate the feasibility of implementing such analyses. We synthesize lessons learned based on our case study and provide recommendations on how to realize system-specific analyses and how to get them adopted by industry.",
"Context: Many metrics are used in software engineering research as surrogates for maintainability of software systems. Aim: Our aim was to investigate whether such metrics are consistent among themselves and the extent to which they predict maintenance effort at the entire system level. Method: The Maintainability Index, a set of structural measures, two code smells (Feature Envy and God Class) and size were applied to a set of four functionally equivalent systems. The metrics were compared with each other and with the outcome of a study in which six developers were hired to perform three maintenance tasks on the same systems. Results: The metrics were not mutually consistent. Only system size and low cohesion were strongly associated with increased maintenance effort. Conclusion: Apart from size, surrogate maintainability measures may not reflect future maintenance effort. Surrogates need to be evaluated in the contexts for which they will be used. While traditional metrics are used to identify problematic areas in the code, the improvements of the worst areas may, inadvertently, lead to more problems for the entire system. Our results suggest that local improvements should be accompanied by an evaluation at the system level."
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | @cite_17 reported on a survey with SMEs and their usage of quality standards. They found that quality is often not a prime objective in smaller companies and especially the micro-companies do not even know the standards. | {
"cite_N": [
"@cite_17"
],
"mid": [
"2157209130"
],
"abstract": [
"Under the studies of general core activities including software inspection, review and testing to achieve quality objectives in small-medium size enterprises (SMEs), the paper presents a contemporary view of such companies against quality measures. The results from a local empirical investigation of quality standards in the Turkish software industry are reported. Around 150 software companies have been approached from which 17 detailed feedback inform that in order to ensure software quality, standards including internationally recognized International Standards Organization (ISO) and Capability Maturity Model Integration (CMMI) are given credit. However the substantial workload and resources required to obtain them are also reported as serious; downscaled frameworks of such large models proposed in the literature are not well known by the SMEs either. The paper also discusses \"work around\" that bypasses such standards to ease delivery of products while keeping certificates as labels just to acquire new jobs for the business."
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | @cite_12 proposed QMOOD, a quality model based on object-oriented metrics. These metrics are static measures capturing the quality of an object-oriented design. In a validation study, they found a statistically significant correlation between their quality index and expert opinions. They applied a much larger quality model outside an SME context. | {
"cite_N": [
"@cite_12"
],
"mid": [
"2167363007"
],
"abstract": [
"The paper describes an improved hierarchical model for the assessment of high-level design quality attributes in object-oriented designs. In this model, structural and behavioral design properties of classes, objects, and their relationships are evaluated using a suite of object-oriented design metrics. This model relates design properties such as encapsulation, modularity, coupling, and cohesion to high-level quality attributes such as reusability, flexibility, and complexity using empirical and anecdotal information. The relationship or links from design properties to quality attributes are weighted in accordance with their influence and importance. The model is validated by using empirical and expert opinion to compare with the model results on several large commercial object-oriented systems. A key attribute of the model is that it can be easily modified to include different relationships and weights, thus providing a practical quality assessment tool adaptable to a variety of demands."
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | @cite_14 developed a quality model similar to what we used in our study. Their model includes static maintainability measures. They discussed measurements for real systems but no further validation, especially not in SMEs. @cite_13 applied a maintainability model to analyse open source systems and compared the quality rating with defect resolution times. They found a statistically significant correlation. Again, we used a more detailed quality model in the SME context. | {
"cite_N": [
"@cite_14",
"@cite_13"
],
"mid": [
"1976902149",
"2008155546"
],
"abstract": [
"The amount of effort needed to maintain a software system is related to the technical quality of the source code of that system. The ISO 9126 model for software product quality recognizes maintainability as one of the 6 main characteristics of software product quality, with adaptability, changeability, stability, and testability as subcharacteristics of maintainability. Remarkably, ISO 9126 does not provide a consensual set of measures for estimating maintainability on the basis of a system's source code. On the other hand, the maintainability index has been proposed to calculate a single number that expresses the maintainability of a system. In this paper, we discuss several problems with the MI, and we identify a number of requirements to be fulfilled by a maintainability model to be usable in practice. We sketch a new maintainability model that alleviates most of these problems, and we discuss our experiences with using such as system for IT management consultancy activities.",
"We performed an empirical study of the relation between technical quality of software products and the issue resolution performance of their maintainers. In particular, we tested the hypothesis that ratings for source code maintainability, as employed by the Software Improvement Group (SIG) quality model, are correlated with ratings for issue resolution speed. We tested the hypothesis for issues of type defect and of type enhancement. This study revealed that all but one of the metrics of the SIG quality model show a significant positive correlation with the resolution speed of defects, enhancements, or both."
]
} |
1611.07560 | 2076609859 | Today, small- and medium-sized enterprises (SMEs) in the software industry face major challenges. Their resource constraints require high efficiency in development. Furthermore, quality assurance (QA) measures need to be taken to mitigate the risk of additional, expensive effort for bug fixes or compensations. Automated static analysis (ASA) can reduce this risk because it promises low application effort. SMEs seem to take little advantage of this opportunity. Instead, they still mainly rely on the dynamic analysis approach of software testing. In this article, we report on our experiences from a technology transfer project. Our aim was to evaluate the results static analysis can provide for SMEs as well as the problems that occur when introducing and using static analysis in SMEs. We analysed five software projects from five collaborating SMEs using three different ASA techniques: code clone detection, bug pattern detection and architecture conformance analysis. Following the analysis, we applied a quality model to aggregate and evaluate the results. Our study shows that the effort required to introduce ASA techniques in SMEs is small (mostly below one person-hour each). Furthermore, we encountered only few technical problems. By means of the analyses, we could detect multiple defects in production code. The participating companies perceived the analysis results to be a helpful addition to their current QA and will include the analyses in their QA process. With the help of the Quamoco quality model, we could efficiently aggregate and rate static analysis results. However, we also encountered a partial mismatch with the opinions of the SMEs. We conclude that ASA and quality models can be a valuable and affordable addition to the QA process of SMEs. | @cite_15 proposed a method for continuous code quality management using static analysis based on a quality model. They did not prescribe a specific model but they suggested to tailor existing ones. These authors used such a model to get different views on and understand static analysis results. Furthermore, they reported on positive experience inside Siemens, a very large organisation, but neglected the specific situation of SMEs. | {
"cite_N": [
"@cite_15"
],
"mid": [
"2077139938"
],
"abstract": [
"The quality of source code is a key factor for any software product and its continuous monitoring is an indispensable task for a software development project. We have developed a method for systematic assessing and improving the code quality of ongoing projects by using the results of various static code analysis tools. With different approaches for monitoring the quality (a trend-based one and a benchmarking-based one) and an according tool support we are able to manage the large amount of data that is generated by these static analyses. First experiences when applying the method with software projects in practice have shown the feasibility of our method."
]
} |
1611.07245 | 2625169091 | Dense and accurate 3-D mapping from a monocular sequence is a key technology for several applications and still an open research area. This letter leverages recent results on single-view convolutional network (CNN)-based depth estimation and fuses them with multiview depth estimation. Both approaches present complementary strengths. Multiview depth is highly accurate but only in high-texture areas and high-parallax cases. Single-view depth captures the local structure of midlevel regions, including texture-less areas, but the estimated depth lacks global coherence. The single and multiview fusion we propose is challenging in several aspects. First, both depths are related by a deformation that depends on the image content. Second, the selection of multiview points of high accuracy might be difficult for low-parallax configurations. We present contributions for both problems. Our results in the public datasets of NYUv2 and TUM shows that our algorithm outperforms the individual single and multiview approaches. A video showing the key aspects of mapping in our single and multiview depth proposal is available at https: youtu.be ipc5HukTb4k. | In the multi-view depth estimation, @cite_7 @cite_23 @cite_26 are the first works that achieved dense and real-time reconstructions from monocular sequences. Some of the most relevant aspects are the direct minimization of the photometric error --instead of the traditional geometric error of sparse reconstructions-- and the regularization of the multi-view estimation by adding the total variation (TV) norm to the cost function. | {
"cite_N": [
"@cite_26",
"@cite_23",
"@cite_7"
],
"mid": [
"1514909517",
"1996510988",
"2108134361"
],
"abstract": [
"We present a novel variational approach to estimate dense depth maps from multiple images in real-time. By using robust penalizers for both data term and regularizer, our method preserves discontinuities in the depth map. We demonstrate that the integration of multiple images substantially increases the robustness of estimated depth maps to noise in the input images. The integration of our method into recently published algorithms for camera tracking allows dense geometry reconstruction in real-time using a single handheld camera. We demonstrate the performance of our algorithm with real-world data.",
"We present a system that is capable of interactively reconstructing a scene from a single live camera. We use a dense volumetric representation of the surface, which means there are no constraints concerning the 3D-scene topology. Reconstruction is based on range image fusion using a total variation formulation, where the surface is represented implicitly by a signed distance function. The final 3D-model is obtained by minimizing a global convex energy and extracting the zero level-set of the solution. The whole reconstruction process is designed to be online. Users can constantly inspect the current reconstruction and adapt camera movement to get the desired amount of detail for specific parts of the reconstructed scene.",
"DTAM is a system for real-time camera tracking and reconstruction which relies not on feature extraction but dense, every pixel methods. As a single hand-held RGB camera flies over a static scene, we estimate detailed textured depth maps at selected keyframes to produce a surface patchwork with millions of vertices. We use the hundreds of images available in a video stream to improve the quality of a simple photometric data term, and minimise a global spatially regularised energy functional in a novel non-convex optimisation framework. Interleaved, we track the camera's 6DOF motion precisely by frame-rate whole image alignment against the entire dense model. Our algorithms are highly parallelisable throughout and DTAM achieves real-time performance using current commodity GPU hardware. We demonstrate that a dense model permits superior tracking performance under rapid motion compared to a state of the art method using features; and also show the additional usefulness of the dense model for real-time scene interaction in a physics-enhanced augmented reality application."
]
} |
1611.07245 | 2625169091 | Dense and accurate 3-D mapping from a monocular sequence is a key technology for several applications and still an open research area. This letter leverages recent results on single-view convolutional network (CNN)-based depth estimation and fuses them with multiview depth estimation. Both approaches present complementary strengths. Multiview depth is highly accurate but only in high-texture areas and high-parallax cases. Single-view depth captures the local structure of midlevel regions, including texture-less areas, but the estimated depth lacks global coherence. The single and multiview fusion we propose is challenging in several aspects. First, both depths are related by a deformation that depends on the image content. Second, the selection of multiview points of high accuracy might be difficult for low-parallax configurations. We present contributions for both problems. Our results in the public datasets of NYUv2 and TUM shows that our algorithm outperforms the individual single and multiview approaches. A video showing the key aspects of mapping in our single and multiview depth proposal is available at https: youtu.be ipc5HukTb4k. | TV regularization has low accuracy for large textureless areas, as shown recently in @cite_15 @cite_3 @cite_25 among others. In order to overcome this @cite_15 proposes a piecewise-planar regularization; the plane parameters coming from multi-view superpixel triangulation @cite_18 or layout estimation @cite_0 . @cite_3 proposes higher-order regularization terms that enforce piecewise affine constraints even in separated pixels. @cite_25 selects the best regularization function among a set using sparse laser data. Building on @cite_15 , @cite_19 adds the sparse data-driven 3D primitives of @cite_9 as a regularization prior. Compared to these works, our fusion is the first one where the information added to the multi-view depth is fully dense, data-driven and single-view; and hence it does not rely on additional sensors, parallax or Manhattan and piecewise-planar assumptions. It only relies on the network capabilities for the current domain, assuming that the test data follows the same distribution that the data used for training. | {
"cite_N": [
"@cite_18",
"@cite_9",
"@cite_3",
"@cite_0",
"@cite_19",
"@cite_15",
"@cite_25"
],
"mid": [
"2061637641",
"2146814781",
"1575629906",
"2534523274",
"969383841",
"2296476661",
"1602724287"
],
"abstract": [
"",
"What primitives should we use to infer the rich 3D world behind an image? We argue that these primitives should be both visually discriminative and geometrically informative and we present a technique for discovering such primitives. We demonstrate the utility of our primitives by using them to infer 3D surface normals given a single image. Our technique substantially outperforms the state-of-the-art and shows improved cross-dataset performance.",
"This paper is about dense depthmap estimation using a monocular camera in workspaces with extensive textureless surfaces. Current state of the art techniques have been shown to work in real time with an admirable performance in desktop-size environments. Unfortunately, as we show in this paper, when applied to larger indoor environments, performance often degrades. A common cause is the presence of large affine texture-less areas like by walls, floors, ceilings and drab objects such as chairs and tables. These produce noisy and worse still, grossly erroneous initial seeds for the depthmap that greatly impede successful optimisation. We solve this problem via the introduction of a new non-local higher-order regularisation term that enforces piecewise affine constraints between image pixels that are far apart in the image. This property leverages the observation that the depth at the edges of bland regions are often well estimated whereas their inner pixels are deeply problematic. A welcome by-product of our proposed technique is an estimate of the surface normals at each pixel. We will show that in terms of implementation, our algorithm is a natural extension of the often used variational approaches. We evaluate the proposed technique using real datasets for which we have ground truth models.",
"In this paper, we consider the problem of recovering the spatial layout of indoor scenes from monocular images. The presence of clutter is a major problem for existing single-view 3D reconstruction algorithms, most of which rely on finding the ground-wall boundary. In most rooms, this boundary is partially or entirely occluded. We gain robustness to clutter by modeling the global room space with a parameteric 3D “box” and by iteratively localizing clutter and refitting the box. To fit the box, we introduce a structured learning algorithm that chooses the set of parameters to minimize error, based on global perspective cues. On a dataset of 308 images, we demonstrate the ability of our algorithm to recover spatial layout in cluttered rooms and show several examples of estimated free space.",
"This paper presents a dense monocular mapping algorithm that improves the accuracy of the state-of-the-art variational and multiview stereo methods by incorporating scene priors into its formulation. Most of the improvement of our proposal is in low-textured image regions and for low-parallax camera motions; two typical failure cases of multiview mapping. The specific priors we model are the planarity of homogeneous color regions, the repeating geometric primitives of the scene--that can be learned from data--and the Manhattan structure of indoor rooms. We evaluate the performance of our method in our own sequences and in the publicly available NYU dataset, emphasizing its strengths and weaknesses in different cases.",
"",
"In this paper we address the problem of dense depth map estimation from sparse noisy range data to reconstruct large heterogeneous outdoor scenes. We propose a surface inpainting solution through energy minimisation with an adaptive selection of surface regularisers among a set of well known convex and non-convex regularisers. In fact, the selection of norm is pivotal with respect to the intrinsic surface characteristics. Our goal is to show how dense interpolation of sparse range data can be leveraged of more exotic and non-convex regularisers such as the log and logTGV [1] which can better capture the scene geometry. In contrast to state of the art solutions, we do not restrict ourselves to this set of norms, instead we search for the most apt norm for each semantically segmented part of the scene. Our energy model selection use Bayesian optimisation to learn the best choice of free parameters. This results in an adaptive model selection and the generalisation of well studied regularisation norms. We conclude with a detailed experimental analysis of our approach using a basis of four norms over a set of challenging outdoor scenes."
]
} |
1611.07245 | 2625169091 | Dense and accurate 3-D mapping from a monocular sequence is a key technology for several applications and still an open research area. This letter leverages recent results on single-view convolutional network (CNN)-based depth estimation and fuses them with multiview depth estimation. Both approaches present complementary strengths. Multiview depth is highly accurate but only in high-texture areas and high-parallax cases. Single-view depth captures the local structure of midlevel regions, including texture-less areas, but the estimated depth lacks global coherence. The single and multiview fusion we propose is challenging in several aspects. First, both depths are related by a deformation that depends on the image content. Second, the selection of multiview points of high accuracy might be difficult for low-parallax configurations. We present contributions for both problems. Our results in the public datasets of NYUv2 and TUM shows that our algorithm outperforms the individual single and multiview approaches. A video showing the key aspects of mapping in our single and multiview depth proposal is available at https: youtu.be ipc5HukTb4k. | Due to the difficulty of estimating an accurate and fully dense map from monocular views there are several approaches that estimate only the depth for the highest-gradient pixels (e.g., @cite_21 ). While this approach produces maps of higher density than the more traditional feature-based ones (e.g., @cite_5 ), they are still incomplete models of the scene and hence their applicability might be more limited. | {
"cite_N": [
"@cite_5",
"@cite_21"
],
"mid": [
"1612997784",
"612478963"
],
"abstract": [
"This paper presents ORB-SLAM, a feature-based monocular simultaneous localization and mapping (SLAM) system that operates in real time, in small and large indoor and outdoor environments. The system is robust to severe motion clutter, allows wide baseline loop closing and relocalization, and includes full automatic initialization. Building on excellent algorithms of recent years, we designed from scratch a novel system that uses the same features for all SLAM tasks: tracking, mapping, relocalization, and loop closing. A survival of the fittest strategy that selects the points and keyframes of the reconstruction leads to excellent robustness and generates a compact and trackable map that only grows if the scene content changes, allowing lifelong operation. We present an exhaustive evaluation in 27 sequences from the most popular datasets. ORB-SLAM achieves unprecedented performance with respect to other state-of-the-art monocular SLAM approaches. For the benefit of the community, we make the source code public.",
"We propose a direct (feature-less) monocular SLAM algorithm which, in contrast to current state-of-the-art regarding direct methods, allows to build large-scale, consistent maps of the environment. Along with highly accurate pose estimation based on direct image alignment, the 3D environment is reconstructed in real-time as pose-graph of keyframes with associated semi-dense depth maps. These are obtained by filtering over a large number of pixelwise small-baseline stereo comparisons. The explicitly scale-drift aware formulation allows the approach to operate on challenging sequences including large variations in scene scale. Major enablers are two key novelties: (1) a novel direct tracking method which operates on ( sim (3) ), thereby explicitly detecting scale-drift, and (2) an elegant probabilistic solution to include the effect of noisy depth values into tracking. The resulting direct monocular SLAM system runs in real-time on a CPU."
]
} |
1611.07245 | 2625169091 | Dense and accurate 3-D mapping from a monocular sequence is a key technology for several applications and still an open research area. This letter leverages recent results on single-view convolutional network (CNN)-based depth estimation and fuses them with multiview depth estimation. Both approaches present complementary strengths. Multiview depth is highly accurate but only in high-texture areas and high-parallax cases. Single-view depth captures the local structure of midlevel regions, including texture-less areas, but the estimated depth lacks global coherence. The single and multiview fusion we propose is challenging in several aspects. First, both depths are related by a deformation that depends on the image content. Second, the selection of multiview points of high accuracy might be difficult for low-parallax configurations. We present contributions for both problems. Our results in the public datasets of NYUv2 and TUM shows that our algorithm outperforms the individual single and multiview approaches. A video showing the key aspects of mapping in our single and multiview depth proposal is available at https: youtu.be ipc5HukTb4k. | Depth can be estimated from a single view using different image cues, for example focus (e.g., @cite_24 ) or perspective (e.g., @cite_10 ). Learning-based approaches, as the one we use, basically discover RGB patterns that are relevant for accurate depth regression. | {
"cite_N": [
"@cite_24",
"@cite_10"
],
"mid": [
"2143496191",
"2040939933"
],
"abstract": [
"The concept of depth from focus involves calculating distances to points in an observed scene by modeling the effect that the camera's focal parameters have on images acquired with a small depth of field. This technique is passive and requires only a single camera. The most difficult segment of calculating depth from focus is deconvolving the defocus operator from the scene and modeling it. Most current methods for determining the defocus operator employ inverse filtering. The authors reveal some fundamental problems with inverse filtering: inaccuracies in finding the frequency domain representation, windowing effects, and border effects. A general, matrix-based method using regularization is presented, which eliminates these problems. The new method is confirmed experimentally, with the results showing an RMS error of 1.3 .",
"We present an approach for 3D reconstruction of objects from a single image. Obviously, constraints on the 3D structure are needed to perform this task. Our approach is based on user-provided coplanarity, perpendicularity and parallelism constraints. These are used to calibrate the image and perform 3D reconstruction. The method is described in detail and results are provided."
]
} |
1611.07245 | 2625169091 | Dense and accurate 3-D mapping from a monocular sequence is a key technology for several applications and still an open research area. This letter leverages recent results on single-view convolutional network (CNN)-based depth estimation and fuses them with multiview depth estimation. Both approaches present complementary strengths. Multiview depth is highly accurate but only in high-texture areas and high-parallax cases. Single-view depth captures the local structure of midlevel regions, including texture-less areas, but the estimated depth lacks global coherence. The single and multiview fusion we propose is challenging in several aspects. First, both depths are related by a deformation that depends on the image content. Second, the selection of multiview points of high accuracy might be difficult for low-parallax configurations. We present contributions for both problems. Our results in the public datasets of NYUv2 and TUM shows that our algorithm outperforms the individual single and multiview approaches. A video showing the key aspects of mapping in our single and multiview depth proposal is available at https: youtu.be ipc5HukTb4k. | The pioneering work of Saxena et al. @cite_20 trained a MRF to model depth from a set of global and local image features. Before that, @cite_16 presented an early approach to depth prediction from monocular and stereo cues. Eigen et al. @cite_13 presented a two deep convolutional neural network (CNN) stacked, one to predict global depth an the second one that refines it locally. Build upon this method, @cite_27 recently presented a three scale convolutional network to estimate depth, surface normals and semantic labeling. Liu et al. @cite_6 use a unified continuous CRF-and-CNN framework to estimate depth. The CNN is used to learn the unary and pairwise potentials that the CRF uses for depth prediction. | {
"cite_N": [
"@cite_6",
"@cite_27",
"@cite_16",
"@cite_13",
"@cite_20"
],
"mid": [
"2125416623",
"1905829557",
"127013308",
"2171740948",
"2132947399"
],
"abstract": [
"We consider the problem of depth estimation from a single monocular image in this work. It is a challenging task as no reliable depth cues are available, e.g., stereo correspondences, motions etc. Previous efforts have been focusing on exploiting geometric priors or additional sources of information, with all using hand-crafted features. Recently, there is mounting evidence that features from deep convolutional neural networks (CNN) are setting new records for various vision applications. On the other hand, considering the continuous characteristic of the depth values, depth estimations can be naturally formulated into a continuous conditional random field (CRF) learning problem. Therefore, we in this paper present a deep convolutional neural field model for estimating depths from a single image, aiming to jointly explore the capacity of deep CNN and continuous CRF. Specifically, we propose a deep structured learning scheme which learns the unary and pairwise potentials of continuous CRF in a unified deep CNN framework. The proposed method can be used for depth estimations of general scenes with no geometric priors nor any extra information injected. In our case, the integral of the partition function can be analytically calculated, thus we can exactly solve the log-likelihood optimization. Moreover, solving the MAP problem for predicting depths of a new image is highly efficient as closed-form solutions exist. We experimentally demonstrate that the proposed method outperforms state-of-the-art depth estimation methods on both indoor and outdoor scene datasets.",
"In this paper we address three different computer vision tasks using a single basic architecture: depth prediction, surface normal estimation, and semantic labeling. We use a multiscale convolutional network that is able to adapt easily to each task using only small modifications, regressing from the input image to the output map directly. Our method progressively refines predictions using a sequence of scales, and captures many image details without any superpixels or low-level segmentation. We achieve state-of-the-art performance on benchmarks for all three tasks.",
"Depth estimation in computer vision and robotics is most commonly done via stereo vision (stereopsis), in which images from two cameras are used to triangulate and estimate distances. However, there are also numerous monocular visual cues--such as texture variations and gradients, defocus, color haze, etc. --that have heretofore been little exploited in such systems. Some of these cues apply even in regions without texture, where stereo would work poorly. In this paper, we apply a Markov Random Field (MRF) learning algorithm to capture some of these monocular cues, and incorporate them into a stereo system. We show that by adding monocular cues to stereo (triangulation) ones, we obtain significantly more accurate depth estimates than is possible using either monocular or stereo cues alone. This holds true for a large variety of environments, including both indoor environments and unstructured outdoor environments containing trees forests, buildings, etc. Our approach is general, and applies to incorporating monocular cues together with any off-the-shelf stereo system.",
"Predicting depth is an essential component in understanding the 3D geometry of a scene. While for stereo images local correspondence suffices for estimation, finding depth relations from a single image is less straightforward, requiring integration of both global and local information from various cues. Moreover, the task is inherently ambiguous, with a large source of uncertainty coming from the overall scale. In this paper, we present a new method that addresses this task by employing two deep network stacks: one that makes a coarse global prediction based on the entire image, and another that refines this prediction locally. We also apply a scale-invariant error to help measure depth relations rather than scale. By leveraging the raw datasets as large sources of training data, our method achieves state-of-the-art results on both NYU Depth and KITTI, and matches detailed depth boundaries without the need for superpixelation.",
"We consider the problem of estimating detailed 3D structure from a single still image of an unstructured environment. Our goal is to create 3D models that are both quantitatively accurate as well as visually pleasing. For each small homogeneous patch in the image, we use a Markov random field (MRF) to infer a set of \"plane parametersrdquo that capture both the 3D location and 3D orientation of the patch. The MRF, trained via supervised learning, models both image depth cues as well as the relationships between different parts of the image. Other than assuming that the environment is made up of a number of small planes, our model makes no explicit assumptions about the structure of the scene; this enables the algorithm to capture much more detailed 3D structure than does prior art and also give a much richer experience in the 3D flythroughs created using image-based rendering, even for scenes with significant nonvertical structure. Using this approach, we have created qualitatively correct 3D models for 64.9 percent of 588 images downloaded from the Internet. We have also extended our model to produce large-scale 3D models from a few images."
]
} |
1611.07245 | 2625169091 | Dense and accurate 3-D mapping from a monocular sequence is a key technology for several applications and still an open research area. This letter leverages recent results on single-view convolutional network (CNN)-based depth estimation and fuses them with multiview depth estimation. Both approaches present complementary strengths. Multiview depth is highly accurate but only in high-texture areas and high-parallax cases. Single-view depth captures the local structure of midlevel regions, including texture-less areas, but the estimated depth lacks global coherence. The single and multiview fusion we propose is challenging in several aspects. First, both depths are related by a deformation that depends on the image content. Second, the selection of multiview points of high accuracy might be difficult for low-parallax configurations. We present contributions for both problems. Our results in the public datasets of NYUv2 and TUM shows that our algorithm outperforms the individual single and multiview approaches. A video showing the key aspects of mapping in our single and multiview depth proposal is available at https: youtu.be ipc5HukTb4k. | Based on @cite_27 , @cite_22 incorporates mid-level features in its prediction using . It shows competitive results and a small batch-size training strategy that makes their network faster to train. @cite_17 introduces a different method to predict depth from single-view using deep neural networks, showing that training the network with a much richer output improves the accuracy. @cite_4 formulates the depth prediction as a classification problem and the net output is a pixel-wise distribution over a discrete depth range. Finally, @cite_2 presents an unsupervised network for depth prediction using stereo images. | {
"cite_N": [
"@cite_4",
"@cite_22",
"@cite_27",
"@cite_2",
"@cite_17"
],
"mid": [
"2354576866",
"2471239744",
"1905829557",
"2520707372",
"2962952719"
],
"abstract": [
"Depth estimation from single monocular images is a key component of scene understanding and has benefited largely from deep convolutional neural networks (CNN) recently. In this article, we take advantage of the recent deep residual networks and propose a simple yet effective approach to this problem. We formulate depth estimation as a pixel-wise classification task. Specifically, we first discretize the continuous depth values into multiple bins and label the bins according to their depth range. Then we train fully convolutional deep residual networks to predict the depth label of each pixel. Performing discrete depth label classification instead of continuous depth value regression allows us to predict a confidence in the form of probability distribution. We further apply fully-connected conditional random fields (CRF) as a post processing step to enforce local smoothness interactions, which improves the results. We evaluate our approach on both indoor and outdoor datasets and achieve state-of-the-art performance.",
"Inferring the underlying depth map from a single image is an ill-posed and inherently ambiguous problem. State-of-the-art deep learning methods can now estimate accurate depth maps, but when projected into 3D, still lack local detail and are often highly distorted. We propose a multi-scale convolution neural network to learn from single RGB images fine-scaled depth maps that result in realistic 3D reconstructions. To encourage spatial coherency, we introduce spatial coordinate feature maps and a local relative depth constraint. In our network, the three scales are closely integrated with skip fusion layers, making it highly efficient to train with large-scale data. Experiments on the NYU Depth v2 dataset shows that our depth predictions are not only competitive with state-of-the-art but also leading to 3D reconstructions that are accurate and rich with detail.",
"In this paper we address three different computer vision tasks using a single basic architecture: depth prediction, surface normal estimation, and semantic labeling. We use a multiscale convolutional network that is able to adapt easily to each task using only small modifications, regressing from the input image to the output map directly. Our method progressively refines predictions using a sequence of scales, and captures many image details without any superpixels or low-level segmentation. We achieve state-of-the-art performance on benchmarks for all three tasks.",
"Learning based methods have shown very promising results for the task of depth estimation in single images. However, most existing approaches treat depth prediction as a supervised regression problem and as a result, require vast quantities of corresponding ground truth depth data for training. Just recording quality depth data in a range of environments is a challenging problem. In this paper, we innovate beyond existing approaches, replacing the use of explicit depth data during training with easier-to-obtain binocular stereo footage. We propose a novel training objective that enables our convolutional neural network to learn to perform single image depth estimation, despite the absence of ground truth depth data. Ex-ploiting epipolar geometry constraints, we generate disparity images by training our network with an image reconstruction loss. We show that solving for image reconstruction alone results in poor quality depth images. To overcome this problem, we propose a novel training loss that enforces consistency between the disparities produced relative to both the left and right images, leading to improved performance and robustness compared to existing approaches. Our method produces state of the art results for monocular depth estimation on the KITTI driving dataset, even outperforming supervised methods that have been trained with ground truth depth.",
"A single color image can contain many cues informative towards different aspects of local geometric structure. We approach the problem of monocular depth estimation by using a neural network to produce a mid-level representation that summarizes these cues. This network is trained to characterize local scene geometry by predicting, at every image location, depth derivatives of different orders, orientations and scales. However, instead of a single estimate for each derivative, the network outputs probability distributions that allow it to express confidence about some coefficients, and ambiguity about others. Scene depth is then estimated by harmonizing this overcomplete set of network predictions, using a globalization procedure that finds a single consistent depth map that best matches all the local derivative distributions. We demonstrate the efficacy of this approach through evaluation on the NYU v2 depth data set."
]
} |
1611.07715 | 2950738231 | Deep convolutional neutral networks have achieved great success on image recognition tasks. Yet, it is non-trivial to transfer the state-of-the-art image recognition networks to videos as per-frame evaluation is too slow and unaffordable. We present deep feature flow, a fast and accurate framework for video recognition. It runs the expensive convolutional sub-network only on sparse key frames and propagates their deep feature maps to other frames via a flow field. It achieves significant speedup as flow computation is relatively fast. The end-to-end training of the whole architecture significantly boosts the recognition accuracy. Deep feature flow is flexible and general. It is validated on two recent large scale video datasets. It makes a large step towards practical video recognition. | Deep learning has been successful on image recognition tasks. The network architectures have evolved and become powerful on image classification @cite_1 @cite_14 @cite_8 @cite_49 @cite_44 @cite_9 . For object detection, the region-based methods @cite_24 @cite_28 @cite_32 @cite_4 @cite_48 have become the dominant paradigm. For semantic segmentation, fully convolutional networks (FCNs) @cite_36 @cite_2 @cite_26 have dominated the field. However, it is computationally unaffordable to directly apply such image recognition networks on all the frames for video recognition. Our work provides an effective and efficient solution. | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_26",
"@cite_8",
"@cite_28",
"@cite_48",
"@cite_9",
"@cite_36",
"@cite_1",
"@cite_32",
"@cite_44",
"@cite_24",
"@cite_49",
"@cite_2"
],
"mid": [
"",
"",
"",
"",
"2179352600",
"2950800384",
"2949650786",
"",
"1677182931",
"",
"",
"2102605133",
"",
"1923697677"
],
"abstract": [
"",
"",
"",
"",
"Existing deep convolutional neural networks (CNNs) require a fixed-size (e.g. 224×224) input image. This requirement is “artificial” and may hurt the recognition accuracy for the images or sub-images of an arbitrary size scale. In this work, we equip the networks with a more principled pooling strategy, “spatial pyramid pooling”, to eliminate the above requirement. The new network structure, called SPP-net, can generate a fixed-length representation regardless of image size scale. By removing the fixed-size limitation, we can improve all CNN-based image classification methods in general. Our SPP-net achieves state-of-the-art accuracy on the datasets of ImageNet 2012, Pascal VOC 2007, and Caltech101.",
"We present region-based, fully convolutional networks for accurate and efficient object detection. In contrast to previous region-based detectors such as Fast Faster R-CNN that apply a costly per-region subnetwork hundreds of times, our region-based detector is fully convolutional with almost all computation shared on the entire image. To achieve this goal, we propose position-sensitive score maps to address a dilemma between translation-invariance in image classification and translation-variance in object detection. Our method can thus naturally adopt fully convolutional image classifier backbones, such as the latest Residual Networks (ResNets), for object detection. We show competitive results on the PASCAL VOC datasets (e.g., 83.6 mAP on the 2007 set) with the 101-layer ResNet. Meanwhile, our result is achieved at a test-time speed of 170ms per image, 2.5-20x faster than the Faster R-CNN counterpart. Code is made publicly available at: this https URL",
"Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57 error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28 relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.",
"",
"Rectified activation units (rectifiers) are essential for state-of-the-art neural networks. In this work, we study rectifier neural networks for image classification from two aspects. First, we propose a Parametric Rectified Linear Unit (PReLU) that generalizes the traditional rectified unit. PReLU improves model fitting with nearly zero extra computational cost and little overfitting risk. Second, we derive a robust initialization method that particularly considers the rectifier nonlinearities. This method enables us to train extremely deep rectified models directly from scratch and to investigate deeper or wider network architectures. Based on the learnable activation and advanced initialization, we achieve 4.94 top-5 test error on the ImageNet 2012 classification dataset. This is a 26 relative improvement over the ILSVRC 2014 winner (GoogLeNet, 6.66 [33]). To our knowledge, our result is the first to surpass the reported human-level performance (5.1 , [26]) on this dataset.",
"",
"",
"Object detection performance, as measured on the canonical PASCAL VOC dataset, has plateaued in the last few years. The best-performing methods are complex ensemble systems that typically combine multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30 relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3 . Our approach combines two key insights: (1) one can apply high-capacity convolutional neural networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data is scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, yields a significant performance boost. Since we combine region proposals with CNNs, we call our method R-CNN: Regions with CNN features. We also present experiments that provide insight into what the network learns, revealing a rich hierarchy of image features. Source code for the complete system is available at http: www.cs.berkeley.edu rbg rcnn.",
"",
"Deep Convolutional Neural Networks (DCNNs) have recently shown state of the art performance in high level vision tasks, such as image classification and object detection. This work brings together methods from DCNNs and probabilistic graphical models for addressing the task of pixel-level classification (also called \"semantic image segmentation\"). We show that responses at the final layer of DCNNs are not sufficiently localized for accurate object segmentation. This is due to the very invariance properties that make DCNNs good for high level tasks. We overcome this poor localization property of deep networks by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF). Qualitatively, our \"DeepLab\" system is able to localize segment boundaries at a level of accuracy which is beyond previous methods. Quantitatively, our method sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 71.6 IOU accuracy in the test set. We show how these results can be obtained efficiently: Careful network re-purposing and a novel application of the 'hole' algorithm from the wavelet community allow dense computation of neural net responses at 8 frames per second on a modern GPU."
]
} |
1611.07593 | 2554246103 | Zero-shot recognition (ZSR) aims to recognize target-domain data instances of unseen classes based on the models learned from associated pairs of seen-class source and target domain data. One of the key challenges in ZSR is the relative scarcity of source-domain features (e.g. one feature vector per class), which do not fully account for wide variability in target-domain instances. In this paper we propose a novel framework of learning data-dependent feature transforms for scoring similarity between an arbitrary pair of source and target data instances to account for the wide variability in target domain. Our proposed approach is based on optimizing over a parameterized family of local feature displacements that maximize the source-target adaptive similarity functions. Accordingly we propose formulating zero-shot learning (ZSL) using latent structural SVMs to learn our similarity functions from training data. As demonstration we design a specific algorithm under the proposed framework involving bilinear similarity functions and regularized least squares as penalties for feature displacement. We test our approach on several benchmark datasets for ZSR and show significant improvement over the state-of-the-art. For instance, on aP&Y dataset we can achieve 80.89 in terms of recognition accuracy, outperforming the state-of-the-art by 11.15 . | Transductive Setting: Recently researchers have begun to incorporate test-time unseen-class data in target domain into ZSL ZSR as unlabeled data analogous to the transductive setting. This has led to approaches that attempt to account for domain shift @cite_39 @cite_54 @cite_14 @cite_24 @cite_40 . In this setting, during test time, we are given a list of all unlabelled target instances in addition to unseen-class source-domain descriptions. Potentially these methods can be used in conjunction with any similarity learning procedure trained on seen-class data, as demonstrated in @cite_40 , to score similarity between unseen classes and target domain data instances. | {
"cite_N": [
"@cite_14",
"@cite_54",
"@cite_39",
"@cite_24",
"@cite_40"
],
"mid": [
"2141350700",
"43954826",
"2209594346",
"2514265125",
"2518962550"
],
"abstract": [
"Most existing zero-shot learning approaches exploit transfer learning via an intermediate semantic representation shared between an annotated auxiliary dataset and a target dataset with different classes and no annotation. A projection from a low-level feature space to the semantic representation space is learned from the auxiliary dataset and applied without adaptation to the target dataset. In this paper we identify two inherent limitations with these approaches. First, due to having disjoint and potentially unrelated classes, the projection functions learned from the auxiliary dataset domain are biased when applied directly to the target dataset domain. We call this problem the projection domain shift problem and propose a novel framework, transductive multi-view embedding , to solve it. The second limitation is the prototype sparsity problem which refers to the fact that for each target class, only a single prototype is available for zero-shot learning given a semantic representation. To overcome this problem, a novel heterogeneous multi-view hypergraph label propagation method is formulated for zero-shot learning in the transductive embedding space. It effectively exploits the complementary information offered by different semantic representations and takes advantage of the manifold structures of multiple representation spaces in a coherent manner. We demonstrate through extensive experiments that the proposed approach (1) rectifies the projection shift between the auxiliary and target domains, (2) exploits the complementarity of multiple semantic representations, (3) significantly outperforms existing methods for both zero-shot and N-shot recognition on three image and video benchmark datasets, and (4) enables novel cross-view annotation tasks.",
"Most existing zero-shot learning approaches exploit transfer learning via an intermediate-level semantic representation such as visual attributes or semantic word vectors. Such a semantic representation is shared between an annotated auxiliary dataset and a target dataset with no annotation. A projection from a low-level feature space to the semantic space is learned from the auxiliary dataset and is applied without adaptation to the target dataset. In this paper we identify an inherent limitation with this approach. That is, due to having disjoint and potentially unrelated classes, the projection functions learned from the auxiliary dataset domain are biased when applied directly to the target dataset domain. We call this problem the projection domain shift problem and propose a novel framework, transductive multi-view embedding, to solve it. It is ‘transductive’ in that unlabelled target data points are explored for projection adaptation, and ‘multi-view’ in that both low-level feature (view) and multiple semantic representations (views) are embedded to rectify the projection shift. We demonstrate through extensive experiments that our framework (1) rectifies the projection shift between the auxiliary and target domains, (2) exploits the complementarity of multiple semantic representations, (3) achieves state-of-the-art recognition results on image and video benchmark datasets, and (4) enables novel cross-view annotation tasks.",
"Zero-shot learning (ZSL) can be considered as a special case of transfer learning where the source and target domains have different tasks label spaces and the target domain is unlabelled, providing little guidance for the knowledge transfer. A ZSL method typically assumes that the two domains share a common semantic representation space, where a visual feature vector extracted from an image video can be projected embedded using a projection function. Existing approaches learn the projection function from the source domain and apply it without adaptation to the target domain. They are thus based on naive knowledge transfer and the learned projections are prone to the domain shift problem. In this paper a novel ZSL method is proposed based on unsupervised domain adaptation. Specifically, we formulate a novel regularised sparse coding framework which uses the target domain class labels' projections in the semantic space to regularise the learned target domain projection thus effectively overcoming the projection domain shift problem. Extensive experiments on four object and action recognition benchmark datasets show that the proposed ZSL method significantly outperforms the state-of-the-arts.",
"Zero-shot Recognition (ZSR) is to learn recognition models for novel classes without labeled data. It is a challenging task and has drawn considerable attention in recent years. The basic idea is to transfer knowledge from seen classes via the shared attributes. This paper focus on the transductive ZSR, i.e., we have unlabeled data for novel classes. Instead of learning models for seen and novel classes separately as in existing works, we put forward a novel joint learning approach which learns the shared model space (SMS) for models such that the knowledge can be effectively transferred between classes using the attributes. An effective algorithm is proposed for optimization. We conduct comprehensive experiments on three benchmark datasets for ZSR. The results demonstrates that the proposed SMS can significantly outperform the state-of-the-art related approaches which validates its efficacy for the ZSR task.",
"We develop a novel method for zero shot learning (ZSL) based on test-time adaptation of similarity functions learned using training data. Existing methods exclusively employ source-domain side information for recognizing unseen classes during test time. We show that for batch-mode applications, accuracy can be significantly improved by adapting these predictors to the observed test-time target-domain ensemble. We develop a novel structured prediction method for maximum a posteriori (MAP) estimation, where parameters account for test-time domain shift from what is predicted primarily using source domain information. We propose a Gaussian parameterization for the MAP problem and derive an efficient structure prediction algorithm. Empirically we test our method on four popular benchmark image datasets for ZSL, and show significant improvement over the state-of-the-art, on average, by 11.50 and 30.12 in terms of accuracy for recognition and mean average precision (mAP) for retrieval, respectively."
]
} |
1611.07289 | 2953267074 | In this paper we present a novel method for the correction of motion artifacts that are present in fet al Magnetic Resonance Imaging (MRI) scans of the whole uterus. Contrary to current slice-to-volume registration (SVR) methods, requiring an inflexible anatomical enclosure of a single investigated organ, the proposed patch-to-volume reconstruction (PVR) approach is able to reconstruct a large field of view of non-rigidly deforming structures. It relaxes rigid motion assumptions by introducing a specific amount of redundant information that is exploited with parallelized patch-wise optimization, super-resolution, and automatic outlier rejection. We further describe and provide an efficient parallel implementation of PVR allowing its execution within reasonable time on commercially available graphics processing units (GPU), enabling its use in the clinical practice. We evaluate PVR's computational overhead compared to standard methods and observe improved reconstruction accuracy in presence of affine motion artifacts of approximately 30 compared to conventional SVR in synthetic experiments. Furthermore, we have evaluated our method qualitatively and quantitatively on real fet al MRI data subject to maternal breathing and sudden fet al movements. We evaluate peak-signal-to-noise ratio (PSNR), structural similarity index (SSIM), and cross correlation (CC) with respect to the originally acquired data and provide a method for visual inspection of reconstruction uncertainty. With these experiments we demonstrate successful application of PVR motion compensation to the whole uterus, the human fetus, and the human placenta. | Most motion compensation approaches for fet al MRI are based on SVR techniques that aim to obtain a motion-free and high resolution volume of a fet al target region. Registration of individual 2D slices with a higher resolution 3D volume @cite_11 is the core approach of these algorithms. SVR methods assume that all acquired image stacks are centered at a specific fet al organ (, brain, thorax) and cover three orthogonal image directions. Fig. shows the core elements of SVR and the contribution of previous frameworks from the literature. | {
"cite_N": [
"@cite_11"
],
"mid": [
"2109149988"
],
"abstract": [
"In this study, we registered live-time interventional magnetic resonance imaging (iMRI) slices with a previously obtained high-resolution MRI volume that in turn can be registered with a variety of functional images, e.g., PET, SPECT, for tumor targeting. We created and evaluated a slice-to-volume (SV) registration algorithm with special features for its potential use in iMRI-guided radio-frequency (RF) thermal ablation of prostate cancer. The algorithm features included a multiresolution approach, two similarity measures, and automatic restarting to avoid local minima. Imaging experiments were performed on volunteers using a conventional 1.5-T MR scanner and a clinical 0.2-T C-arm iMRI system under realistic conditions. Both high-resolution MR volumes and actual iMRI image slices were acquired from the same volunteers. Actual and simulated iMRI images were used to test the dependence of SV registration on image noise, receive coil inhomogeneity, and RF needle artifacts. To quantitatively assess registration, we calculated the mean voxel displacement over a volume of interest between SV registration and volume-to-volume registration, which was previously shown to be quite accurate. More than 800 registration experiments were performed. For transverse image slices covering the prostate, the SV registration algorithm was 100 successful with an error of <2 mm, and the average and standard deviation was only 0.4 mm spl plusmn 0.2 mm. Visualizations such as combined sector display and contour overlay showed excellent registration of the prostate and other organs throughout the pelvis. Error was greater when an image slice was obtained at other orientations and positions, mostly because of inconsistent image content such as that from variable rectal and bladder filling. These preliminary experiments indicate that MR SV registration is sufficiently accurate to aid image-guided therapy."
]
} |
1611.07289 | 2953267074 | In this paper we present a novel method for the correction of motion artifacts that are present in fet al Magnetic Resonance Imaging (MRI) scans of the whole uterus. Contrary to current slice-to-volume registration (SVR) methods, requiring an inflexible anatomical enclosure of a single investigated organ, the proposed patch-to-volume reconstruction (PVR) approach is able to reconstruct a large field of view of non-rigidly deforming structures. It relaxes rigid motion assumptions by introducing a specific amount of redundant information that is exploited with parallelized patch-wise optimization, super-resolution, and automatic outlier rejection. We further describe and provide an efficient parallel implementation of PVR allowing its execution within reasonable time on commercially available graphics processing units (GPU), enabling its use in the clinical practice. We evaluate PVR's computational overhead compared to standard methods and observe improved reconstruction accuracy in presence of affine motion artifacts of approximately 30 compared to conventional SVR in synthetic experiments. Furthermore, we have evaluated our method qualitatively and quantitatively on real fet al MRI data subject to maternal breathing and sudden fet al movements. We evaluate peak-signal-to-noise ratio (PSNR), structural similarity index (SSIM), and cross correlation (CC) with respect to the originally acquired data and provide a method for visual inspection of reconstruction uncertainty. With these experiments we demonstrate successful application of PVR motion compensation to the whole uterus, the human fetus, and the human placenta. | @cite_17 introduces the acquisition of many thin slices to provide sufficient sampling of the region of interest. Cross correlation is used as a cost function for the SVR steps assuming that the data have consistent contrast properties. After that, multilevel B-splines are applied to the volumetric reconstruction for data interpolation, which has the advantage of reducing blurring of the reconstructed image supported by including the thin slices. | {
"cite_N": [
"@cite_17"
],
"mid": [
"2166966312"
],
"abstract": [
"Motion degrades magnetic resonance (MR) images and prevents acquisition of self-consistent and high-quality volume images. A novel methodology, Snapshot magnetic resonance imaging (MRI) with volume reconstruction (SVR) has been developed for imaging moving subjects at high resolution and high signal-to-noise ratio (SNR). The method combines registered 2D slices from sequential dynamic single-shot scans. The SVR approach requires that the anatomy in question is not changing shape or size and is moving at a rate that allows snapshot images to be acquired. After imaging the target volume repeatedly to guarantee sufficient sampling every where, a robust slice-to-volume registration method has been implemented that achieves alignment of each slice within 0.3 mm in the examples tested. Multilevel scattered interpolation has been used to obtain high-fidelity reconstruction with root-mean-square (rms) error that is less than the noise level in the images. The SVR method has been performed successfully for brain studies on subjects that cannot stay still, and in some cases were moving substantially during scanning. For example, awake neonates, deliberately moved adults and, especially, on fetuses, for which no conventional high-resolution 3D method is currently available. Fine structure of the in-utero fet al brain is clearly revealed for the first time and substantial SNR improvement is realized by having many individually acquired slices contribute to each voxel in the reconstructed image."
]
} |
1611.07289 | 2953267074 | In this paper we present a novel method for the correction of motion artifacts that are present in fet al Magnetic Resonance Imaging (MRI) scans of the whole uterus. Contrary to current slice-to-volume registration (SVR) methods, requiring an inflexible anatomical enclosure of a single investigated organ, the proposed patch-to-volume reconstruction (PVR) approach is able to reconstruct a large field of view of non-rigidly deforming structures. It relaxes rigid motion assumptions by introducing a specific amount of redundant information that is exploited with parallelized patch-wise optimization, super-resolution, and automatic outlier rejection. We further describe and provide an efficient parallel implementation of PVR allowing its execution within reasonable time on commercially available graphics processing units (GPU), enabling its use in the clinical practice. We evaluate PVR's computational overhead compared to standard methods and observe improved reconstruction accuracy in presence of affine motion artifacts of approximately 30 compared to conventional SVR in synthetic experiments. Furthermore, we have evaluated our method qualitatively and quantitatively on real fet al MRI data subject to maternal breathing and sudden fet al movements. We evaluate peak-signal-to-noise ratio (PSNR), structural similarity index (SSIM), and cross correlation (CC) with respect to the originally acquired data and provide a method for visual inspection of reconstruction uncertainty. With these experiments we demonstrate successful application of PVR motion compensation to the whole uterus, the human fetus, and the human placenta. | @cite_33 were the first to introduce a mathematical model for super-resolution (SR) volume reconstruction from slice acquisitions of fet al brains. The main difference to previous reconstruction methods is that it includes knowledge of the slice acquisition model and the SR reconstruction is performed based on maximum likelihood and a robust M-estimation minimization for an error norm function. A regularization term is also added to the cost function in order to enforce a solution when the number of acquired samples is not high enough for solving the reconstruction problem. | {
"cite_N": [
"@cite_33"
],
"mid": [
"2135264048"
],
"abstract": [
"Fast magnetic resonance imaging slice acquisition techniques such as single shot fast spin echo are routinely used in the presence of uncontrollable motion. These techniques are widely used for fet al magnetic resonance imaging (MRI) and MRI of moving subjects and organs. Although high-quality slices are frequently acquired by these techniques, inter-slice motion leads to severe motion artifacts that are apparent in out-of-plane views. Slice sequential acquisitions do not enable 3-D volume representation. In this study, we have developed a novel technique based on a slice acquisition model, which enables the reconstruction of a volumetric image from multiple-scan slice acquisitions. The super-resolution volume reconstruction is formulated as an inverse problem of finding the underlying structure generating the acquired slices. We have developed a robust M-estimation solution which minimizes a robust error norm function between the model-generated slices and the acquired slices. The accuracy and robustness of this novel technique has been quantitatively assessed through simulations with digital brain phantom images as well as high-resolution newborn images. We also report here successful application of our new technique for the reconstruction of volumetric fet al brain MRI from clinically acquired data."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | Kautz @cite_14 developed and used metrics to evaluate how new practices and tools for configuration and change management were affecting the software process at three SMEs. This work considers that the key to successful software measurement is to make metrics meaningful and to tailor them to a particular organisation. We confirm that observation in the context of software measurement. | {
"cite_N": [
"@cite_14"
],
"mid": [
"2075108560"
],
"abstract": [
"The key to successful measurement programs is to make the metrics meaningful and tailor them to the organization-however small it might be. Here the author explains how he helped three small companies reduce the time spent handling change requests, correcting errors, and generating system versions and production releases. Both the developers and customers appreciated the improved practices."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | Von @cite_16 investigated the assessment of software processes in SMEs to improve these processes. They developed MARES, a set of guidelines for conducting an ISO IEC 15504-conforming software process assessment, focused on small companies. We perceive the usage guidelines we reported as a potential bridge between automated static analysis and more general guidelines for software process improvement. | {
"cite_N": [
"@cite_16"
],
"mid": [
"2123004246"
],
"abstract": [
"A first step toward process improvement is identifying the strengths and weaknesses of an organization's software processes to determine effective improvement actions. An assessment can help an organization examine its processes against a reference model to determine the processes' capability or the organization's maturity, to meet quality, cost, and schedule goals, but small companies have difficulty running them. MARES, a set of guidelines for conducting 15504-conformant software process assessment focuses on small companies"
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | Hofer @cite_12 states that only 10 Returning to automated static analysis techniques, to the best of our knowledge, multiple techniques have never been applied in a study in an SME context. However, there are several publications in which such techniques were investigated separately and in other contexts: | {
"cite_N": [
"@cite_12"
],
"mid": [
"2151019728"
],
"abstract": [
"In this paper we present the results of an empirical study conducted within small and very small software developing enterprises in Austria. The primary goals of this survey were to find out how they are structured, to learn more about their characteristics, and how they organize software development. Furthermore we evaluate methods, processes, techniques and tools used, persons and roles involved, and finally problems occurring during software development. Additionally we place a special focus on quality management issues and discuss which motivations small and very small enterprises can have to install such a quality management system."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | @cite_28 report on application of function clone detection to a large telecommunication software system. As opposed to that, we do not limit clone detection to the comparison of functions but compare arbitrary code fragments with each other. In this work we also did not analyse large systems. Nevertheless, we came to the similar conclusion that clone detection has potential to improve software quality. | {
"cite_N": [
"@cite_28"
],
"mid": [
"2115534035"
],
"abstract": [
"The aim of the experiment presented in this paper is to present an insight into the evaluation of the potential benefits of introducing a function clone detection technology in an industrial software development process. To take advantage of function clone detection, two modifications to the software development process are presented. Our experiment consists of evaluating the impact that these proposed changes would have had on a specific software system if they had been applied over a 3 year period (involving 10000 person-months), where 6 subsequent versions of the software under study were released. The software under study is a large telecommunication system. In total 89 million lines of code have been analyzed. A first result showed that, against our expectations, a significant number of clones are being removed from the system over time. However, this effort is insufficient to prevent the growth of the overall number of clones in the system. In this context the first process change would have added value. We have also found that the second process change would have provided programmers with a significant number of opportunities for correcting problems before customers experienced them. This result shows a potential for improving the software system quality and customer satisfaction"
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | Lanubile and Mallardo @cite_18 performed research on finding clones in web applications developed using markup and programming languages. As mentioned earlier, our approach is technically limited in analysing such software systems. Introducing a semi-automatic approach presented by Lanubile and Mallardo could remove this limitation. | {
"cite_N": [
"@cite_18"
],
"mid": [
"2097557721"
],
"abstract": [
"Many Web applications use a mixture of HTML and scripting language code as the front-end to business services. Analogously to traditional applications, redundant code is introduced by copy-and-paste practices. Code duplication is a pathological form of software reuse because of its effects on the maintenance of large software systems. This paper describes how a simple semi-automated approach can be used to identity cloned functions within scripting code of Web applications. The results obtained from applying our approach to three Web applications show that the approach is useful for a fast selection of script function clones, and can be applied to prevent clone spreading or to remove redundant scripting code."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | @cite_2 evaluate the accuracy and value of FindBugs findings and discuss but not solve the problem of properly filtering false positives. They use the term for what we call smells and pedantry. We confirm their conclusions on the usefulness of findings and believe that an application of bug pattern detection has to undergo calibration guided by the staff of a software project. Moreover, by answering RQ2, we contribute to Foster's, Hicks' and Pugh's @cite_5 question Are the defects reported by [static analysis] tools important?''. | {
"cite_N": [
"@cite_5",
"@cite_2"
],
"mid": [
"2120981580",
"2125343911"
],
"abstract": [
"At the University of Maryland, we have been working to improve the reliability and security of software by developing new, effective static analysis tools. These tools scan software for bug patterns or show that the software is free from a particular class of defects. There are two themes common to our different projects: 1. Our ultimate focus is on utility: can a programmer actually improve the quality of his or her software using an analysis tool? The important first step toward answering this question is to engineer tools so that they can analyze existing, nontrivial programs, and to carefully report the results of such analyses experimentally. The desire to better understand a more human-centered notion of utility underlies much of our future work. 2. We release all of our tools open source. This allows other researchers to verify our results, and to reuse some or all of our implementations, which often required significant effort to engineer. We believe that releasing source code is important for accelerating the pace of research results software quality, and just as importantly allows feedback from the wider community. In this research group presentation, we summarize some recent work and sketch future directions.",
"Static analysis tools for software defect detection are becoming widely used in practice. However, there is little public information regarding the experimental evaluation of the accuracy and value of the warnings these tools report. In this paper, we discuss the warnings found by FindBugs, a static analysis tool that finds defects in Java programs. We discuss the kinds of warnings generated and the classification of warnings into false positives, trivial bugs and serious bugs. We also provide some insight into why static analysis tools often detect true but trivial bugs, and some information about defect warnings across the development lifetime of software release. We report data on the defect warnings in Sun's Java 6 JRE, in Sun's Glassfish JEE server, and in portions of Google's Java codebase. Finally, we report on some experiences from incorporating static analysis into the software development process at Google."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | Ferzund, Ahsan and Wotawa @cite_9 report on the effectiveness of rules for smell detection. The rules they developed are based on machine learning and source file statistics provided by static code metrics. They used training information from two software projects including bug databases. We did not address the estimation of rule effectiveness but focused on their selection and application. | {
"cite_N": [
"@cite_9"
],
"mid": [
"131006540"
],
"abstract": [
"Open Source Softwares provide a rich resource of empirical research in software engineering. Static code metrics are a good indicator of software quality and maintainability. In this work we have tried to answer the question whether bug predictors obtained from one project can be applied to a different project with reasonable accuracy. Two open source projects Firefox and Apache HTTP Server (AHS) are used for this study. Static code metrics are calculated for both projects using in-house software and the bug information is obtained from bug databases of these projects. The source code files are classified as clean or buggy using the Decision tree classifier. The classifier is trained on metrics and bug data of Firefox and tested on Apache HTTP Server and vice versa. The results obtained vary with different releases of these projects and can be as good as 92 of the files correctly classified and as poor as 68 of the files correctly classified by the trained classifier."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | @cite_24 similarly applied FindBugs and PMD to two industrial projects. They could not find defects reported from the field that are covered by bug pattern detection. However, our results show that this technique indeed captures critical defects that may eventually occur in the field. @cite_1 conducted an industrial case study on architecture conformance with three participating software engineers. They conclude that this technique should be integrated into the software engineering process and applied continuously. We think that the procedure we presented is able to satisfy their needs, because it explicitly focuses on continuous integration. | {
"cite_N": [
"@cite_24",
"@cite_1"
],
"mid": [
"2141790816",
"2087988340"
],
"abstract": [
"Automated static analysis is a promising technique to detect defects in software. However, although considerable effort has been spent for developing sophisticated detection possibilities, the effectiveness and efficiency has not been treated in equal detail. This paper presents the results of two industrial case studies in which two tools based on bug patterns for Java are applied and evaluated. First, the economic implications of the tools are analysed. It is estimated that only 3-4 potential field defects need to be detected for the tools to be cost-efficient. Second, the capabilities of detecting field defects are investigated. No field defects have been found that could have been detected by the tools. Third, the identification of fault-prone classes based on the results of such tools is investigated and found to be possible. Finally, methodological consequences are derived from the results and experiences in order to improve the use of bug pattern tools in practice.",
"A software designer often has little control over, or means of checking, whether his design is being adhered to, once the implementation begins. This 'architectural drift', where the original design of the system and the as-implemented design of the system diverge, can cause serious problems for evolution, maintenance and the comprehensibility of a system if it remains undocumented or uncorrected. The earlier such discrepancies can be identified, the better. This paper reports on a case study designed to evaluate a proposed approach for monitoring architectural drift during software development. The proposed approach differs significantly from existing design conformance approaches in that it has been designed for application within a forward engineering context. The findings show the utility of the approach in identifying architectural drift but also illustrate a number of improvements that could be made to the technique to further promote architecture conformance."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | @cite_10 illustrate their experience in an industrial project and the huge effort that is needed to keep the architectural model in conformance with the implementation. However, they tried to reach this goal in a manual way. Our results show that automation can dramatically reduce efforts. | {
"cite_N": [
"@cite_10"
],
"mid": [
"2166771505"
],
"abstract": [
"A basic idea of model driven development (MDD) is to capture all important design information in a set of formal or semi formal models that are automatically kept consistent by tools. This paper reports on industrial experience from use of MDD and shows that the approach needs improvements regarding the architecture since there are no suggested ways to formalize design rules which are an important part of the architecture. Instead, one has to rely on time consuming and error prone manual interpretations, reviews and reworkings to keep the system consistent with the architecture. To reap the full benefits of MDD it is therefore important to find ways of formalizing design rules to make it possible to allow automatic enforcement of the architecture on the system model."
]
} |
1611.07549 | 291824086 | Today’s small and medium-sized enterprises (SMEs) in the software industry are faced with major challenges. While having to work efficiently using limited resources they have to perform quality assurance on their code to avoid the risk of further effort for bug fixes or compensations. Automated static analysis can reduce this risk because it promises little effort for running an analysis. We report on our experience in analysing five projects from and with SMEs by three different static analysis techniques: code clone detection, bug pattern detection and architecture conformance analysis. We found that the effort that was needed to introduce those techniques was small (mostly below one person-hour), that we can detect diverse defects in production code and that the participating companies perceived the usefulness of the presented techniques as well as our analysis results high enough to include the techniques in their quality assurance. | Feilkas, Ratiu and Juergens @cite_19 analysed three .NET platform projects of Munich Re very similar to our procedure, but they analysed the effects of the loss of architectural knowledge. Compared to our results they report a much higher effort of about five days to apply the technique, mainly because of time consuming discussions. We think that the lower effort we are reporting is mainly caused by the fact that we were collaborating with small enterprises and experienced a lower communication overhead. | {
"cite_N": [
"@cite_19"
],
"mid": [
"2132092215"
],
"abstract": [
"Architecture defines the components of a system and their dependencies. The knowledge about how the architecture is intended to be implemented is essential to keep the system structure coherent and thereby comprehensible. In practice, this architectural knowledge is explicitly formulated only in the documentation (if at all), which usually gets outdated very soon. This leads to a growing amount of implicit knowledge during evolution that is especially volatile in projects with high developer fluctuation."
]
} |
1611.07509 | 2949355442 | Anti-discrimination is an increasingly important task in data science. In this paper, we investigate the problem of discovering both direct and indirect discrimination from the historical data, and removing the discriminatory effects before the data is used for predictive analysis (e.g., building classifiers). We make use of the causal network to capture the causal structure of the data. Then we model direct and indirect discrimination as the path-specific effects, which explicitly distinguish the two types of discrimination as the causal effects transmitted along different paths in the network. Based on that, we propose an effective algorithm for discovering direct and indirect discrimination, as well as an algorithm for precisely removing both types of discrimination while retaining good data utility. Different from previous works, our approaches can ensure that the predictive models built from the modified data will not incur discrimination in decision making. Experiments using real datasets show the effectiveness of our approaches. | Proposed methods for discrimination removal are either based on data preprocessing @cite_17 @cite_15 or algorithm tweaking @cite_27 @cite_2 @cite_8 . In a recent work @cite_24 , the authors first ensure no direct discrimination by completely removing the protected attribute @math from data, and then modify all the non-protected attributes to ensure that @math cannot be predicted from the non-protected attributes. As a result, indirect discrimination is removed since the decision @math has no connection with @math via the non-protected attributes. However, as shown in our experiment results, this approach cannot ensure that predictions made by the classifier built on the modified data do not contain discrimination. In addition, it suffers significant utility loss as it removes all the connections between @math and @math . | {
"cite_N": [
"@cite_8",
"@cite_24",
"@cite_27",
"@cite_2",
"@cite_15",
"@cite_17"
],
"mid": [
"2040825624",
"2952092676",
"2150997454",
"2097246321",
"2893559225",
""
],
"abstract": [
"With the spread of data mining technologies and the accumulation of social data, such technologies and data are being used for determinations that seriously affect people's lives. For example, credit scoring is frequently determined based on the records of past credit data together with statistical prediction techniques. Needless to say, such determinations must be socially and legally fair from a viewpoint of social responsibility, namely, it must be unbiased and nondiscriminatory in sensitive features, such as race, gender, religion, and so on. Several researchers have recently begun to attempt the development of analysis techniques that are aware of social fairness or discrimination. They have shown that simply avoiding the use of sensitive features is insufficient for eliminating biases in determinations, due to the indirect influence of sensitive information. From a privacy-preserving viewpoint, this can be interpreted as hiding sensitive information when classification results are observed. In this paper, we first discuss three causes of unfairness in machine learning. We then propose a regularization approach that is applicable to any prediction algorithm with probabilistic discriminative models. We further apply this approach to logistic regression and empirically show its effectiveness and efficiency.",
"What does it mean for an algorithm to be biased? In U.S. law, unintentional bias is encoded via disparate impact, which occurs when a selection process has widely different outcomes for different groups, even as it appears to be neutral. This legal determination hinges on a definition of a protected class (ethnicity, gender, religious practice) and an explicit description of the process. When the process is implemented using computers, determining disparate impact (and hence bias) is harder. It might not be possible to disclose the process. In addition, even if the process is open, it might be hard to elucidate in a legal setting how the algorithm makes its decisions. Instead of requiring access to the algorithm, we propose making inferences based on the data the algorithm uses. We make four contributions to this problem. First, we link the legal notion of disparate impact to a measure of classification accuracy that while known, has received relatively little attention. Second, we propose a test for disparate impact based on analyzing the information leakage of the protected class from the other data attributes. Third, we describe methods by which data might be made unbiased. Finally, we present empirical evidence supporting the effectiveness of our test for disparate impact and our approach for both masking bias and preserving relevant information in the data. Interestingly, our approach resembles some actual selection practices that have recently received legal scrutiny.",
"Recently, the following discrimination aware classification problem was introduced: given a labeled dataset and an attribute B, find a classifier with high predictive accuracy that at the same time does not discriminate on the basis of the given attribute B. This problem is motivated by the fact that often available historic data is biased due to discrimination, e.g., when B denotes ethnicity. Using the standard learners on this data may lead to wrongfully biased classifiers, even if the attribute B is removed from training data. Existing solutions for this problem consist in “cleaning away” the discrimination from the dataset before a classifier is learned. In this paper we study an alternative approach in which the non-discrimination constraint is pushed deeply into a decision tree learner by changing its splitting criterion and pruning strategy. Experimental evaluation shows that the proposed approach advances the state-of-the-art in the sense that the learned decision trees have a lower discrimination than models provided by previous methods, with little loss in accuracy.",
"In this paper, we investigate how to modify the naive Bayes classifier in order to perform classification that is restricted to be independent with respect to a given sensitive attribute. Such independency restrictions occur naturally when the decision process leading to the labels in the data-set was biased; e.g., due to gender or racial discrimination. This setting is motivated by many cases in which there exist laws that disallow a decision that is partly based on discrimination. Naive application of machine learning techniques would result in huge fines for companies. We present three approaches for making the naive Bayes classifier discrimination-free: (i) modifying the probability of the decision being positive, (ii) training one model for every sensitive attribute value and balancing them, and (iii) adding a latent variable to the Bayesian model that represents the unbiased label and optimizing the model parameters for likelihood using expectation maximization. We present experiments for the three approaches on both artificial and real-life data.",
"",
""
]
} |
1611.07285 | 2552369821 | Given an image, we would like to learn to detect objects belonging to particular object categories. Common object detection methods train on large annotated datasets which are annotated in terms of bounding boxes that contain the object of interest. Previous works on object detection model the problem as a structured regression problem which ranks the correct bounding boxes more than the background ones. In this paper we develop algorithms which actively obtain annotations from human annotators for a small set of images, instead of all images, thereby reducing the annotation effort. Towards this goal, we make the following contributions: 1. We develop a principled version space based active learning method that solves for object detection as a structured prediction problem in a weakly supervised setting 2. We derive the relation between our proposed method with other active learning techniques such as maximum model change 3. Additionally, we propose two variants of the margin-based querying strategy 4. We analyse the results on standard object detection benchmarks that show that with only 20 of the data we can obtain more than 95 of the localization accuracy of full supervision. Our methods outperform random sampling and the classical uncertainty-based active learning algorithms like entropy | Object classification and detection are widely explored topics in computer vision. It started with recognizing objects using simple geometry @cite_19 to generating better features for object classification and detection using @cite_16 @cite_2 @cite_39 to using better classifiers @cite_24 @cite_13 @cite_50 to using Convolutional Neural Networks (CNN) like @cite_22 , R-CNN @cite_20 , Fast R-CNN @cite_46 , Faster R-CNN @cite_6 and SSD @cite_5 . However, all these methods make use of full supervision for object detection in terms of explicit bounding box information. | {
"cite_N": [
"@cite_22",
"@cite_6",
"@cite_39",
"@cite_24",
"@cite_19",
"@cite_50",
"@cite_2",
"@cite_5",
"@cite_46",
"@cite_16",
"@cite_13",
"@cite_20"
],
"mid": [
"2163605009",
"2953106684",
"",
"1525954826",
"1910706092",
"2538008885",
"2161969291",
"2193145675",
"",
"2124386111",
"2168356304",
"2102605133"
],
"abstract": [
"We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5 and 17.0 which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called \"dropout\" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3 , compared to 26.2 achieved by the second-best entry.",
"State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully convolutional network that simultaneously predicts object bounds and objectness scores at each position. The RPN is trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. We further merge RPN and Fast R-CNN into a single network by sharing their convolutional features---using the recently popular terminology of neural networks with 'attention' mechanisms, the RPN component tells the unified network where to look. For the very deep VGG-16 model, our detection system has a frame rate of 5fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015 competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning entries in several tracks. Code has been made publicly available.",
"",
"Sliding window classifiers are among the most successful and widely applied techniques for object localization. However, training is typically done in a way that is not specific to the localization task. First a binary classifier is trained using a sample of positive and negative examples, and this classifier is subsequently applied to multiple regions within test images. We propose instead to treat object localization in a principled way by posing it as a problem of predicting structured data: we model the problem not as binary classification, but as the prediction of the bounding box of objects located in images. The use of a joint-kernelframework allows us to formulate the training procedure as a generalization of an SVM, which can be solved efficiently. We further improve computational efficiency by using a branch-and-bound strategy for localization during both training and testing. Experimental evaluation on the PASCAL VOC and TU Darmstadt datasets show that the structured training procedure improves performance over binary training as well as the best previously published scores.",
"Recent advances in object recognition have emphasized the integration of intensity-derived features such as affine patches with associated geometric constraints leading to impressive performance in complex scenes. Over the four previous decades, the central paradigm of recognition was based on formal geometric object descriptions with a focus on the properties of such descriptions under perspective image formation. This paper will review the key advances of the geometric era and investigate the underlying causes of the movement away from formal geometry and prior models towards the use of statistical learning methods based on appearance features.",
"Our objective is to obtain a state-of-the art object category detector by employing a state-of-the-art image classifier to search for the object in all possible image sub-windows. We use multiple kernel learning of Varma and Ray (ICCV 2007) to learn an optimal combination of exponential χ2 kernels, each of which captures a different feature channel. Our features include the distribution of edges, dense and sparse visual words, and feature descriptors at different levels of spatial organization.",
"We study the question of feature sets for robust visual object recognition; adopting linear SVM based human detection as a test case. After reviewing existing edge and gradient based descriptors, we show experimentally that grids of histograms of oriented gradient (HOG) descriptors significantly outperform existing feature sets for human detection. We study the influence of each stage of the computation on performance, concluding that fine-scale gradients, fine orientation binning, relatively coarse spatial binning, and high-quality local contrast normalization in overlapping descriptor blocks are all important for good results. The new approach gives near-perfect separation on the original MIT pedestrian database, so we introduce a more challenging dataset containing over 1800 annotated human images with a large range of pose variations and backgrounds.",
"We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For (300 300 ) input, SSD achieves 74.3 mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for (512 512 ) input, SSD achieves 76.9 mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https: github.com weiliu89 caffe tree ssd.",
"",
"An object recognition system has been developed that uses a new class of local image features. The features are invariant to image scaling, translation, and rotation, and partially invariant to illumination changes and affine or 3D projection. These features share similar properties with neurons in inferior temporal cortex that are used for object recognition in primate vision. Features are efficiently detected through a staged filtering approach that identifies stable points in scale space. Image keys are created that allow for local geometric deformations by representing blurred image gradients in multiple orientation planes and at multiple scales. The keys are used as input to a nearest neighbor indexing method that identifies candidate object matches. Final verification of each match is achieved by finding a low residual least squares solution for the unknown model parameters. Experimental results show that robust object recognition can be achieved in cluttered partially occluded images with a computation time of under 2 seconds.",
"We describe an object detection system based on mixtures of multiscale deformable part models. Our system is able to represent highly variable object classes and achieves state-of-the-art results in the PASCAL object detection challenges. While deformable part models have become quite popular, their value had not been demonstrated on difficult benchmarks such as the PASCAL data sets. Our system relies on new methods for discriminative training with partially labeled data. We combine a margin-sensitive approach for data-mining hard negative examples with a formalism we call latent SVM. A latent SVM is a reformulation of MI--SVM in terms of latent variables. A latent SVM is semiconvex, and the training problem becomes convex once latent information is specified for the positive examples. This leads to an iterative training algorithm that alternates between fixing latent values for positive examples and optimizing the latent SVM objective function.",
"Object detection performance, as measured on the canonical PASCAL VOC dataset, has plateaued in the last few years. The best-performing methods are complex ensemble systems that typically combine multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30 relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3 . Our approach combines two key insights: (1) one can apply high-capacity convolutional neural networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data is scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, yields a significant performance boost. Since we combine region proposals with CNNs, we call our method R-CNN: Regions with CNN features. We also present experiments that provide insight into what the network learns, revealing a rich hierarchy of image features. Source code for the complete system is available at http: www.cs.berkeley.edu rbg rcnn."
]
} |
1611.07285 | 2552369821 | Given an image, we would like to learn to detect objects belonging to particular object categories. Common object detection methods train on large annotated datasets which are annotated in terms of bounding boxes that contain the object of interest. Previous works on object detection model the problem as a structured regression problem which ranks the correct bounding boxes more than the background ones. In this paper we develop algorithms which actively obtain annotations from human annotators for a small set of images, instead of all images, thereby reducing the annotation effort. Towards this goal, we make the following contributions: 1. We develop a principled version space based active learning method that solves for object detection as a structured prediction problem in a weakly supervised setting 2. We derive the relation between our proposed method with other active learning techniques such as maximum model change 3. Additionally, we propose two variants of the margin-based querying strategy 4. We analyse the results on standard object detection benchmarks that show that with only 20 of the data we can obtain more than 95 of the localization accuracy of full supervision. Our methods outperform random sampling and the classical uncertainty-based active learning algorithms like entropy | Active learning @cite_10 has been widely used for object classification @cite_27 @cite_8 @cite_43 @cite_44 @cite_48 @cite_32 @cite_29 @cite_35 , where given an image we want to label the presence or absence of an object of a particular class, and video annotation @cite_7 @cite_42 @cite_38 @cite_9 . However, this problem has not been much explored for object detection which requires structured prediction of the correct bounding box. We discuss a few related works that have explored this problem. In @cite_4 , the authors propose a part-based detector for object detection amenable to SVM, and show how to identify its most uncertain instances using the simple margin of @cite_3 in sub-linear time with a hashing-based solution. @cite_1 proposes an incremental learning approach for some fixed data collection, like medical images, that continuously updates an object detector and detection thresholds as a user interactively corrects annotations proposed by the detector. Unlike @cite_1 , in the proposed approach, the annotation costs are assumed to be equal for every image. Active learning should not be confused with active detection methods like @cite_25 where the discriminative ability of an already trained base classifier is refined at test time using minimum human supervision. | {
"cite_N": [
"@cite_35",
"@cite_38",
"@cite_4",
"@cite_7",
"@cite_8",
"@cite_48",
"@cite_29",
"@cite_42",
"@cite_9",
"@cite_32",
"@cite_3",
"@cite_1",
"@cite_44",
"@cite_43",
"@cite_27",
"@cite_10",
"@cite_25"
],
"mid": [
"2008989859",
"1560354729",
"2027953712",
"1855646318",
"",
"1975672287",
"1510526001",
"2117907414",
"2023618850",
"2201054263",
"2426031434",
"2030297513",
"2037820867",
"2027248184",
"2098742124",
"2903158431",
"2127788596"
],
"abstract": [
"As a way to relieve the tedious work of manual annotation, active learning plays important roles in many applications of visual concept recognition. In typical active learning scenarios, the number of labelled data in the seed set is usually small. However, most existing active learning algorithms only exploit the labelled data, which often suffers from over-fitting due to the small number of labelled examples. Besides, while much progress has been made in binary class active learning, little research attention has been focused on multi-class active learning. In this paper, we propose a semi-supervised batch mode multi-class active learning algorithm for visual concept recognition. Our algorithm exploits the whole active pool to evaluate the uncertainty of the data. Considering that uncertain data are always similar to each other, we propose to make the selected data as diverse as possible, for which we explicitly impose a diversity constraint on the objective function. As a multi-class active learning algorithm, our algorithm is able to exploit uncertainty across multiple classes. An efficient algorithm is used to optimize the objective function. Extensive experiments on action recognition, object classification, scene recognition, and event detection demonstrate its advantages.",
"Manually segmenting and labeling objects in video sequences is quite tedious, yet such annotations are valuable for learning-based approaches to object and activity recognition. While automatic label propagation can help, existing methods simply propagate annotations from arbitrarily selected frames (e.g., the first one) and so may fail to best leverage the human effort invested. We define an active frame selection problem: select k frames for manual labeling, such that automatic pixel-level label propagation can proceed with minimal expected error. We propose a solution that directly ties a joint frame selection criterion to the predicted errors of a flow-based random field propagation model. It selects the set of k frames that together minimize the total mislabeling risk over the entire sequence. We derive an efficient dynamic programming solution to optimize the criterion. Further, we show how to automatically determine how many total frames k should be labeled in order to minimize the total manual effort spent labeling and correcting propagation errors. We demonstrate our method's clear advantages over several baselines, saving hours of human effort per video.",
"Active learning and crowdsourcing are promising ways to efficiently build up training sets for object recognition, but thus far techniques are tested in artificially controlled settings. Typically the vision researcher has already determined the dataset's scope, the labels “actively” obtained are in fact already known, and or the crowd-sourced collection process is iteratively fine-tuned. We present an approach for live learning of object detectors, in which the system autonomously refines its models by actively requesting crowd-sourced annotations on images crawled from the Web. To address the technical issues such a large-scale system entails, we introduce a novel part-based detector amenable to linear classifiers, and show how to identify its most uncertain instances in sub-linear time with a hashing-based solution. We demonstrate the approach with experiments of unprecedented scale and autonomy, and show it successfully improves the state-of-the-art for the most challenging objects in the PASCAL benchmark. In addition, we show our detector competes well with popular nonlinear classifiers that are much more expensive to train.",
"High-speed smooth and accurate visual tracking of objects in arbitrary, unstructured environments is essential for robotics and human motion analysis. However, building a system that can adapt to arbitrary objects and a wide range of lighting conditions is a challenging problem, especially if hard real-time constraints apply like in robotics scenarios. In this work, we introduce a method for learning a discriminative object tracking system based on the recent structured regression framework for object localization. Using a kernel function that allows fast evaluation on the GPU, the resulting system can process video streams at speed of 100 frames per second or more. Consecutive frames in high speed video sequences are typically very redundant, and for training an object detection system, it is sufficient to have training labels from only a subset of all images. We propose an active learning method that select training examples in a data-driven way, thereby minimizing the required number of training labeling. Experiments on realistic data show that the active learning is superior to previously used methods for dataset subsampling for this task.",
"",
"Recently active learning has attracted a lot of attention in computer vision field, as it is time and cost consuming to prepare a good set of labeled images for vision data analysis. Most existing active learning approaches employed in computer vision adopt most uncertainty measures as instance selection criteria. Although most uncertainty query selection strategies are very effective in many circumstances, they fail to take information in the large amount of unlabeled instances into account and are prone to querying outliers. In this paper, we present a novel adaptive active learning approach that combines an information density measure and a most uncertainty measure together to select critical instances to label for image classifications. Our experiments on two essential tasks of computer vision, object recognition and scene recognition, demonstrate the efficacy of the proposed approach.",
"Pairwise coupling is a popular multi-class classification method that combines all comparisons for each pair of classes. This paper presents two approaches for obtaining class probabilities. Both methods can be reduced to linear systems and are easy to implement. We show conceptually and experimentally that the proposed approaches are more stable than the two existing popular methods: voting and the method by Hastie and Tibshirani (1998)",
"We introduce a novel active learning framework for video annotation. By judiciously choosing which frames a user should annotate, we can obtain highly accurate tracks with minimal user effort. We cast this problem as one of active learning, and show that we can obtain excellent performance by querying frames that, if annotated, would produce a large expected change in the estimated object track. We implement a constrained tracker and compute the expected change for putative annotations with efficient dynamic programming algorithms. We demonstrate our framework on four datasets, including two benchmark datasets constructed with key frame annotations obtained by Amazon Mechanical Turk. Our results indicate that we could obtain equivalent labels for a small fraction of the original cost.",
"We describe an information-driven active selection approach to determine which detectors to deploy at which location in which frame of a video to minimize semantic class label uncertainty at every pixel, with the smallest computational cost that ensures a given uncertainty bound. We show minimal performance reduction compared to a \"paragon\" algorithm running all detectors at all locations in all frames, at a small fraction of the computational cost. Our method can handle uncertainty in the labeling mechanism, so it can handle both \"oracles\" (manual annotation) or noisy detectors (automated annotation).",
"Active learning is an effective way to relieve the tedious work of manual annotation in many applications of visual recognition. However, less research attention has been focused on multi-class active learning. In this paper, we propose a novel Gaussian process classifier model with multiple annotators for multi-class visual recognition. Expectation propagation (EP) is adopted for efficient approximate Bayesian inference of our probabilistic model for classification. Based on the EP approximation inference, a generalized Expectation Maximization (GEM) algorithm is derived to estimate both the parameters for instances and the quality of each individual annotator. Also, we incorporate the idea of reinforcement learning to actively select both the informative samples and the high-quality annotators, which better explores the trade-off between exploitation and exploration. The experiments clearly demonstrate the efficacy of the proposed model.",
"Support vector machines have met with significant success in numerous real-world learning tasks. However, like most machine learning algorithms, they are generally applied using a randomly selected training set classified in advance. In many settings, we also have the option of using pool-based active learning. Instead of using a randomly selected training set, the learner has access to a pool of unlabeled instances and can request the labels for some number of them. We introduce a new algorithm for performing active learning with support vector machines, i.e., an algorithm for choosing which instances to request next. We provide a theoretical motivation for the algorithm using the notion of a version space. We present experimental results showing that employing our active learning method can significantly reduce the need for labeled training instances in both the standard inductive and transductive settings.",
"In recent years, the rise of digital image and video data available has led to an increasing demand for image annotation. In this paper, we propose an interactive object annotation method that incrementally trains an object detector while the user provides annotations. In the design of the system, we have focused on minimizing human annotation time rather than pure algorithm learning performance. To this end, we optimize the detector based on a realistic annotation cost model based on a user study. Since our system gives live feedback to the user by detecting objects on the fly and predicts the potential annotation costs of unseen images, data can be efficiently annotated by a single user without excessive waiting time. In contrast to popular tracking-based methods for video annotation, our method is suitable for both still images and video. We have evaluated our interactive annotation approach on three datasets, ranging from surveillance, television, to cell microscopy.",
"We present an active learning framework that predicts the tradeoff between the effort and information gain associated with a candidate image annotation, thereby ranking unlabeled and partially labeled images according to their expected \"net worth\" to an object recognition system. We develop a multi-label multiple-instance approach that accommodates realistic images containing multiple objects and allows the category-learner to strategically choose what annotations it receives from a mixture of strong and weak labels. Since the annotation cost can vary depending on an image's complexity, we show how to improve the active selection by directly predicting the time required to segment an unlabeled image. Our approach accounts for the fact that the optimal use of manual effort may call for a combination of labels at multiple levels of granularity, as well as accurate prediction of manual effort. As a result, it is possible to learn more accurate category models with a lower total expenditure of annotation effort. Given a small initial pool of labeled data, the proposed method actively improves the category models with minimal manual intervention.",
"Discriminative methods for visual object category recognition are typically non-probabilistic, predicting class labels but not directly providing an estimate of uncertainty. Gaussian Processes (GPs) provide a framework for deriving regression techniques with explicit uncertainty models; we show here how Gaussian Processes with covariance functions defined based on a Pyramid Match Kernel (PMK) can be used for probabilistic object category recognition. Our probabilistic formulation provides a principled way to learn hyperparameters, which we utilize to learn an optimal combination of multiple covariance functions. It also offers confidence estimates at test points, and naturally allows for an active learning paradigm in which points are optimally selected for interactive labeling. We show that with an appropriate combination of kernels a significant boost in classification performance is possible. Further, our experiments indicate the utility of active learning with probabilistic predictive models, especially when the amount of training data labels that may be sought for a category is ultimately very small.",
"Most methods for learning object categories require large amounts of labeled training data. However, obtaining such data can be a difficult and time-consuming endeavor. We have developed a novel, entropy-based ldquoactive learningrdquo approach which makes significant progress towards this problem. The main idea is to sequentially acquire labeled data by presenting an oracle (the user) with unlabeled images that will be particularly informative when labeled. Active learning adaptively prioritizes the order in which the training examples are acquired, which, as shown by our experiments, can significantly reduce the overall number of training examples required to reach near-optimal performance. At first glance this may seem counter-intuitive: how can the algorithm know whether a group of unlabeled images will be informative, when, by definition, there is no label directly associated with any of the images? Our approach is based on choosing an image to label that maximizes the expected amount of information we gain about the set of unlabeled images. The technique is demonstrated in several contexts, including improving the efficiency of Web image-search queries and open-world visual learning by an autonomous agent. Experiments on a large set of 140 visual object categories taken directly from text-based Web image searches show that our technique can provide large improvements (up to 10 x reduction in the number of training examples needed) over baseline techniques.",
"",
"Efficient detection of multiple object instances is one of the fundamental challenges in computer vision. For certain object categories, even the best automatic systems are yet unable to produce high-quality detection results, and fully manual annotation would be an expensive process. How can detection algorithms interplay with human expert annotators? To make the best use of scarce (human) labeling resources, one needs to decide when to invoke the expert, such that the best possible performance can be achieved while requiring a minimum amount of supervision. In this paper, we propose a principled approach to active object detection, and show that for a rich class of base detectors algorithms, one can derive a natural sequential decision problem for deciding when to invoke expert supervision. We further show that the objective function satisfies adaptive submodularity, which allows us to derive strong performance guarantees for our algorithm. We demonstrate the proposed algorithm on three real-world tasks, including a problem for biodiversity monitoring from micro UAVs in the Sumatra rain forest. Our results show that active detection not only outperforms its passive counterpart; for certain tasks, it also works significantly better than straightforward application of existing active learning techniques. To the best of our knowledge, our approach is the first to rigorously address the active detection problem from both empirical and theoretical perspectives."
]
} |
1611.07285 | 2552369821 | Given an image, we would like to learn to detect objects belonging to particular object categories. Common object detection methods train on large annotated datasets which are annotated in terms of bounding boxes that contain the object of interest. Previous works on object detection model the problem as a structured regression problem which ranks the correct bounding boxes more than the background ones. In this paper we develop algorithms which actively obtain annotations from human annotators for a small set of images, instead of all images, thereby reducing the annotation effort. Towards this goal, we make the following contributions: 1. We develop a principled version space based active learning method that solves for object detection as a structured prediction problem in a weakly supervised setting 2. We derive the relation between our proposed method with other active learning techniques such as maximum model change 3. Additionally, we propose two variants of the margin-based querying strategy 4. We analyse the results on standard object detection benchmarks that show that with only 20 of the data we can obtain more than 95 of the localization accuracy of full supervision. Our methods outperform random sampling and the classical uncertainty-based active learning algorithms like entropy | Unlike heuristic approaches like entropy, version space approaches have a strong theoretical foundation @cite_3 . Through this paper we provide, to the best of our knowledge, the first work that uses version spaces in the context of object detection. | {
"cite_N": [
"@cite_3"
],
"mid": [
"2426031434"
],
"abstract": [
"Support vector machines have met with significant success in numerous real-world learning tasks. However, like most machine learning algorithms, they are generally applied using a randomly selected training set classified in advance. In many settings, we also have the option of using pool-based active learning. Instead of using a randomly selected training set, the learner has access to a pool of unlabeled instances and can request the labels for some number of them. We introduce a new algorithm for performing active learning with support vector machines, i.e., an algorithm for choosing which instances to request next. We provide a theoretical motivation for the algorithm using the notion of a version space. We present experimental results showing that employing our active learning method can significantly reduce the need for labeled training instances in both the standard inductive and transductive settings."
]
} |
1611.07233 | 2557216475 | Lossy image compression algorithms are pervasively used to reduce the size of images transmitted over the web and recorded on data storage media. However, we pay for their high compression rate with visual artifacts degrading the user experience. Deep convolutional neural networks have become a widespread tool to address high-level computer vision tasks very successfully. Recently, they have found their way into the areas of low-level computer vision and image processing to solve regression problems mostly with relatively shallow networks. We present a novel 12-layer deep convolutional network for image compression artifact suppression with hierarchical skip connections and a multi-scale loss function. We achieve a boost of up to 1.79 dB in PSNR over ordinary JPEG and an improvement of up to 0.36 dB over the best previous ConvNet result. We show that a network trained for a specific quality factor (QF) is resilient to the QF used to compress the input image — a single network trained for QF 60 provides a PSNR gain of more than 1.5 dB over the wide QF range from 40 to 76. | Traditional approaches to suppress compression artifacts can be split into three categories. Various types of intelligent edge-aware denoising such as SA-DCT @cite_29 @cite_3 , BM3D @cite_14 have been proposed to address this task during the late 2000s. In recent years, dictionary-based sparse recovery algorithms such as DicTV @cite_34 , RTF @cite_28 , S-D2 @cite_9 , @math @cite_6 , DDCN @cite_22 have achieved outstanding results by directly addressing the deficiencies such as ringing and blocking very specific to JPEG. These algorithms explicitly attempt to optimally reverse the effect of DCT-domain quantization using learned dictionaries very specific to the applied compressor and quantization tables. | {
"cite_N": [
"@cite_14",
"@cite_22",
"@cite_28",
"@cite_29",
"@cite_9",
"@cite_3",
"@cite_6",
"@cite_34"
],
"mid": [
"2056370875",
"2519021537",
"",
"2135065661",
"1946766895",
"2147218302",
"2466611277",
"1978280057"
],
"abstract": [
"We propose a novel image denoising strategy based on an enhanced sparse representation in transform domain. The enhancement of the sparsity is achieved by grouping similar 2D image fragments (e.g., blocks) into 3D data arrays which we call \"groups.\" Collaborative Altering is a special procedure developed to deal with these 3D groups. We realize it using the three successive steps: 3D transformation of a group, shrinkage of the transform spectrum, and inverse 3D transformation. The result is a 3D estimate that consists of the jointly filtered grouped image blocks. By attenuating the noise, the collaborative filtering reveals even the finest details shared by grouped blocks and, at the same time, it preserves the essential unique features of each individual block. The filtered blocks are then returned to their original positions. Because these blocks are overlapping, for each pixel, we obtain many different estimates which need to be combined. Aggregation is a particular averaging procedure which is exploited to take advantage of this redundancy. A significant improvement is obtained by a specially developed collaborative Wiener filtering. An algorithm based on this novel denoising strategy and its efficient implementation are presented in full detail; an extension to color-image denoising is also developed. The experimental results demonstrate that this computationally scalable algorithm achieves state-of-the-art denoising performance in terms of both peak signal-to-noise ratio and subjective visual quality.",
"We propose a highly accurate approach to remove artifacts of JPEG-compressed images. Our approach jointly learns a very deep convolutional network in both DCT and pixel domains. The dual-domain representation can make full use of DCT-domain prior knowledge of JPEG compression, which is usually lacking in traditional network-based approaches. At the same time, it can also benefit from the prowess and the efficiency of the deep feed-forward architecture, in comparison to capacity-limited sparse-coding-based approaches. Two simple strategies, i.e., Adam and residual learning, are adopted to train the very deep network and later proved to be a success. Extensive experiments demonstrate the large improvements of our approach over the state of the arts.",
"",
"The shape-adaptive discrete cosine transform (SA-DCT) transform can be computed on a support of arbitrary shape, but retains a computational complexity comparable to that of the usual separable block-DCT (B-DCT). Despite the near-optimal decorrelation and energy compaction properties, application of the SA-DCT has been rather limited, targeted nearly exclusively to video compression. In this paper, we present a novel approach to image filtering based on the SA-DCT. We use the SA-DCT in conjunction with the Anisotropic Local Polynomial Approximation-Intersection of Confidence Intervals technique, which defines the shape of the transform's support in a pointwise adaptive manner. The thresholded or attenuated SA-DCT coefficients are used to reconstruct a local estimate of the signal within the adaptive-shape support. Since supports corresponding to different points are in general overlapping, the local estimates are averaged together using adaptive weights that depend on the region's statistics. This approach can be used for various image-processing tasks. In this paper, we consider, in particular, image denoising and image deblocking and deringing from block-DCT compression. A special structural constraint in luminance-chrominance space is also proposed to enable an accurate filtering of color images. Simulation experiments show a state-of-the-art quality of the final estimate, both in terms of objective criteria and visual appearance. Thanks to the adaptive support, reconstructed edges are clean, and no unpleasant ringing artifacts are introduced by the fitted transform",
"Arguably the most common cause of image degradation is compression. This papers presents a novel approach to restoring JPEG-compressed images. The main innovation is in the approach of exploiting residual redundancies of JPEG code streams and sparsity properties of latent images. The restoration is a sparse coding process carried out jointy in the DCT and. pixel domains. The prowess of the proposed approach is directly restoring DCT coefficients of the latent image to prevent the spreading of quantization errors into the pixel domain, and at the same time using on-line machine-learnt local spatial features to regulate the solution of the underlying inverse problem. Experimental results are encouraging and show the promise of the new approach in significantly improving the quality of DCT-coded images.",
"We present an high-quality image deblocking algorithm based on the shape-adaptive DCT (SA-DCT) [20, 21]. The SA-DCT is a low-complexity transform which can be computed on a support of arbitrary shape. This transform has been adopted by the MPEG-4 standard [13] and it is found implemented in modern video hardware.",
"In this paper, we design a Deep Dual-Domain (D3) based fast restoration model to remove artifacts of JPEG compressed images. It leverages the large learning capacity of deep networks, as well as the problem-specific expertise that was hardly incorporated in the past design of deep architectures. For the latter, we take into consideration both the prior knowledge of the JPEG compression scheme, and the successful practice of the sparsity-based dual-domain approach. We further design the One-Step Sparse Inference (1-SI) module, as an efficient and lightweighted feed-forward approximation of sparse coding. Extensive experiments verify the superiority of the proposed D3 model over several state-of-the-art methods. Specifically, our best model is capable of outperforming the latest deep model for around 1 dB in PSNR, and is 30 times faster.",
"The JPEG compression method is among the most successful compression schemes since it readily provides good compressed results at a rather high compression ratio. However, the decompressed result of the standard JPEG decompression scheme usually contains some visible artifacts, such as blocking artifacts and Gibbs artifacts (ringing), especially when the compression ratio is rather high. In this paper, a novel artifact reducing approach for the JPEG decompression is proposed via sparse and redundant representations over a learned dictionary. Indeed, an effective two-step algorithm is developed. The first step involves dictionary learning and the second step involves the total variation regularization for decompressed images. Numerical experiments are performed to demonstrate that the proposed method outperforms the total variation and weighted total variation decompression methods in the measure of peak of signal to noise ratio, and structural similarity."
]
} |
1611.07233 | 2557216475 | Lossy image compression algorithms are pervasively used to reduce the size of images transmitted over the web and recorded on data storage media. However, we pay for their high compression rate with visual artifacts degrading the user experience. Deep convolutional neural networks have become a widespread tool to address high-level computer vision tasks very successfully. Recently, they have found their way into the areas of low-level computer vision and image processing to solve regression problems mostly with relatively shallow networks. We present a novel 12-layer deep convolutional network for image compression artifact suppression with hierarchical skip connections and a multi-scale loss function. We achieve a boost of up to 1.79 dB in PSNR over ordinary JPEG and an improvement of up to 0.36 dB over the best previous ConvNet result. We show that a network trained for a specific quality factor (QF) is resilient to the QF used to compress the input image — a single network trained for QF 60 provides a PSNR gain of more than 1.5 dB over the wide QF range from 40 to 76. | This work was inspired by single-image super-resolution ConvNets, which are a special case of compression artifact removal, where the compression is a simple sub-sampling operation. Several networks have shown to be very successful at this task, such as SRCNN @cite_27 or DRCN @cite_35 . They use different training procedures and approaches for network construction, but both ConvNets are a simple sequence of convolution and point-wise non-linearity layers. | {
"cite_N": [
"@cite_35",
"@cite_27"
],
"mid": [
"2949079773",
"54257720"
],
"abstract": [
"We propose an image super-resolution method (SR) using a deeply-recursive convolutional network (DRCN). Our network has a very deep recursive layer (up to 16 recursions). Increasing recursion depth can improve performance without introducing new parameters for additional convolutions. Albeit advantages, learning a DRCN is very hard with a standard gradient descent method due to exploding vanishing gradients. To ease the difficulty of training, we propose two extensions: recursive-supervision and skip-connection. Our method outperforms previous methods by a large margin.",
"We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) [15] that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage."
]
} |
1611.07233 | 2557216475 | Lossy image compression algorithms are pervasively used to reduce the size of images transmitted over the web and recorded on data storage media. However, we pay for their high compression rate with visual artifacts degrading the user experience. Deep convolutional neural networks have become a widespread tool to address high-level computer vision tasks very successfully. Recently, they have found their way into the areas of low-level computer vision and image processing to solve regression problems mostly with relatively shallow networks. We present a novel 12-layer deep convolutional network for image compression artifact suppression with hierarchical skip connections and a multi-scale loss function. We achieve a boost of up to 1.79 dB in PSNR over ordinary JPEG and an improvement of up to 0.36 dB over the best previous ConvNet result. We show that a network trained for a specific quality factor (QF) is resilient to the QF used to compress the input image — a single network trained for QF 60 provides a PSNR gain of more than 1.5 dB over the wide QF range from 40 to 76. | Recently, two important works have been published, which apply ConvNets for compression artifact suppression: AR-CNN @cite_19 @cite_8 and the approach presented in @cite_33 . The former starts from the architecture presented in SRCNN. In order to overcome convergence problems, they use transfer-learning from the 4-layer network retrained for artifact reduction to a deeper 5-layer network, as well as between networks trained for different JPEG quality factors (QFs) and datasets. In @cite_33 a residual structure extends the simple stacking of convolutional, non-linearity and pooling layers, such that the network is only trained to produce an increment compensating for the distortions. Furthermore, skip elements where some feature maps are bypassing one or multiple layers and are then concatenated to the feature maps at a later stage were introduced. Additionally, they do not use a plain MSE loss function but also include an additional term to emphasize edges. | {
"cite_N": [
"@cite_19",
"@cite_33",
"@cite_8"
],
"mid": [
"2142683286",
"2345337169",
"2526792573"
],
"abstract": [
"Lossy compression introduces complex compression artifacts, particularly the blocking artifacts, ringing effects and blurring. Existing algorithms either focus on removing blocking artifacts and produce blurred output, or restores sharpened images that are accompanied with ringing effects. Inspired by the deep convolutional networks (DCN) on super-resolution, we formulate a compact and efficient network for seamless attenuation of different compression artifacts. We also demonstrate that a deeper model can be effectively trained with the features learned in a shallow network. Following a similar \"easy to hard\" idea, we systematically investigate several practical transfer settings and show the effectiveness of transfer learning in low level vision problems. Our method shows superior performance than the state-of-the-arts both on the benchmark datasets and the real-world use cases (i.e. Twitter).",
"This paper shows that it is possible to train large and deep convolutional neural networks (CNN) for JPEG compression artifacts reduction, and that such networks can provide significantly better reconstruction quality compared to previously used smaller networks as well as to any other state-of-the-art methods. We were able to train networks with 8 layers in a single step and in relatively short time by combining residual learning, skip architecture, and symmetric weight initialization. We provide further insights into convolution networks for JPEG artifact reduction by evaluating three different objectives, generalization with respect to training dataset size, and generalization with respect to JPEG quality level.",
"Lossy compression introduces complex compression artifacts, particularly blocking artifacts, ringing effects and blurring. Existing algorithms either focus on removing blocking artifacts and produce blurred output, or restore sharpened images that are accompanied with ringing effects. Inspired by the success of deep convolutional networks (DCN) on superresolution, we formulate a compact and efficient network for seamless attenuation of different compression artifacts. To meet the speed requirement of real-world applications, we further accelerate the proposed baseline model by layer decomposition and joint use of large-stride convolutional and deconvolutional layers. This also leads to a more general CNN framework that has a close relationship with the conventional Multi-Layer Perceptron (MLP). Finally, the modified network achieves a speed up of 7.5 times with almost no performance loss compared to the baseline model. We also demonstrate that a deeper model can be effectively trained with features learned in a shallow network. Following a similar \"easy to hard\" idea, we systematically investigate three practical transfer settings and show the effectiveness of transfer learning in low-level vision problems. Our method shows superior performance than the state-of-the-art methods both on benchmark datasets and a real-world use case."
]
} |
1611.07571 | 2949347953 | Several machine learning tasks require to represent the data using only a sparse set of interest points. An ideal detector is able to find the corresponding interest points even if the data undergo a transformation typical for a given domain. Since the task is of high practical interest in computer vision, many hand-crafted solutions were proposed. In this paper, we ask a fundamental question: can we learn such detectors from scratch? Since it is often unclear what points are "interesting", human labelling cannot be used to find a truly unbiased solution. Therefore, the task requires an unsupervised formulation. We are the first to propose such a formulation: training a neural network to rank points in a transformation-invariant manner. Interest points are then extracted from the top bottom quantiles of this ranking. We validate our approach on two tasks: standard RGB image interest point detection and challenging cross-modal interest point detection between RGB and depth images. We quantitatively show that our unsupervised method performs better or on-par with baselines. | Currently, unsupervised learning comprises many directions: learning the distribution that best explains data (Gaussian Mixture Models learned via an EM-algorithm @cite_1 , Restricted Boltzmann Machines @cite_17 , Generative Adversarial Nets @cite_11 ), clustering, dimensionality reduction and unsupervised segmentation (kMeans @cite_18 , LLE @cite_24 , Isomap @cite_16 , PCA @cite_25 , Normalized Cuts @cite_26 , t-SNE @cite_2 ), learning to simulate task solvers (when a solution is provided by the solver and the task is automatically generated @cite_23 , @cite_37 ), and learning data representation suitable for further use in some other task (autoencoders @cite_29 , deep convolutional adversarial nets @cite_32 , learning by context prediction @cite_19 , learning from tracking in videos @cite_36 , metric learning @cite_33 , learning by predicting inpainting @cite_9 , learning by solving jigsaw puzzles @cite_31 ). | {
"cite_N": [
"@cite_18",
"@cite_26",
"@cite_37",
"@cite_33",
"@cite_36",
"@cite_29",
"@cite_9",
"@cite_1",
"@cite_17",
"@cite_32",
"@cite_24",
"@cite_19",
"@cite_23",
"@cite_2",
"@cite_31",
"@cite_16",
"@cite_25",
"@cite_11"
],
"mid": [
"",
"2121947440",
"2951384764",
"1909903157",
"219040644",
"2100495367",
"2342877626",
"1579271636",
"2116064496",
"2173520492",
"2053186076",
"2950187998",
"2089858332",
"2187089797",
"2949497014",
"2001141328",
"2148694408",
"2099471712"
],
"abstract": [
"",
"We propose a novel approach for solving the perceptual grouping problem in vision. Rather than focusing on local features and their consistencies in the image data, our approach aims at extracting the global impression of an image. We treat image segmentation as a graph partitioning problem and propose a novel global criterion, the normalized cut, for segmenting the graph. The normalized cut criterion measures both the total dissimilarity between the different groups as well as the total similarity within the groups. We show that an efficient computational technique based on a generalized eigenvalue problem can be used to optimize this criterion. We applied this approach to segmenting static images, as well as motion sequences, and found the results to be very encouraging.",
"Wooden blocks are a common toy for infants, allowing them to develop motor skills and gain intuition about the physical behavior of the world. In this paper, we explore the ability of deep feed-forward models to learn such intuitive physics. Using a 3D game engine, we create small towers of wooden blocks whose stability is randomized and render them collapsing (or remaining upright). This data allows us to train large convolutional network models which can accurately predict the outcome, as well as estimating the block trajectories. The models are also able to generalize in two important ways: (i) to new physical scenarios, e.g. towers with an additional block and (ii) to images of real wooden blocks, where it obtains a performance comparable to human subjects.",
"Detecting poorly textured objects and estimating their 3D pose reliably is still a very challenging problem. We introduce a simple but powerful approach to computing descriptors for object views that efficiently capture both the object identity and 3D pose. By contrast with previous manifold-based approaches, we can rely on the Euclidean distance to evaluate the similarity between descriptors, and therefore use scalable Nearest Neighbor search methods to efficiently handle a large number of objects under a large range of poses. To achieve this, we train a Convolutional Neural Network to compute these descriptors by enforcing simple similarity and dissimilarity constraints between the descriptors. We show that our constraints nicely untangle the images from different objects and different views into clusters that are not only well-separated but also structured as the corresponding sets of poses: The Euclidean distance between descriptors is large when the descriptors are from different objects, and directly related to the distance between the poses when the descriptors are from the same object. These important properties allow us to outperform state-of-the-art object views representations on challenging RGB and RGB-D data.",
"Is strong supervision necessary for learning a good visual representation? Do we really need millions of semantically-labeled images to train a Convolutional Neural Network (CNN)? In this paper, we present a simple yet surprisingly powerful approach for unsupervised learning of CNN. Specifically, we use hundreds of thousands of unlabeled videos from the web to learn visual representations. Our key idea is that visual tracking provides the supervision. That is, two patches connected by a track should have similar visual representation in deep feature space since they probably belong to same object or object part. We design a Siamese-triplet network with a ranking loss function to train this CNN representation. Without using a single image from ImageNet, just using 100K unlabeled videos and the VOC 2012 dataset, we train an ensemble of unsupervised networks that achieves 52 mAP (no bounding box regression). This performance comes tantalizingly close to its ImageNet-supervised counterpart, an ensemble which achieves a mAP of 54.4 . We also show that our unsupervised network can perform competitively in other tasks such as surface-normal estimation.",
"High-dimensional data can be converted to low-dimensional codes by training a multilayer neural network with a small central layer to reconstruct high-dimensional input vectors. Gradient descent can be used for fine-tuning the weights in such “autoencoder” networks, but this works well only if the initial weights are close to a good solution. We describe an effective way of initializing the weights that allows deep autoencoder networks to learn low-dimensional codes that work much better than principal components analysis as a tool to reduce the dimensionality of data.",
"We present an unsupervised visual feature learning algorithm driven by context-based pixel prediction. By analogy with auto-encoders, we propose Context Encoders -- a convolutional neural network trained to generate the contents of an arbitrary image region conditioned on its surroundings. In order to succeed at this task, context encoders need to both understand the content of the entire image, as well as produce a plausible hypothesis for the missing part(s). When training context encoders, we have experimented with both a standard pixel-wise reconstruction loss, as well as a reconstruction plus an adversarial loss. The latter produces much sharper results because it can better handle multiple modes in the output. We found that a context encoder learns a representation that captures not just appearance but also the semantics of visual structures. We quantitatively demonstrate the effectiveness of our learned features for CNN pre-training on classification, detection, and segmentation tasks. Furthermore, context encoders can be used for semantic inpainting tasks, either stand-alone or as initialization for non-parametric methods.",
"The important role of finite mixture models in the statistical analysis of data is underscored by the ever-increasing rate at which articles on mixture applications appear in the statistical and ge...",
"It is possible to combine multiple latent-variable models of the same data by multiplying their probability distributions together and then renormalizing. This way of combining individual \"expert\" models makes it hard to generate samples from the combined model but easy to infer the values of the latent variables of each expert, because the combination rule ensures that the latent variables of different experts are conditionally independent when given the data. A product of experts (PoE) is therefore an interesting candidate for a perceptual system in which rapid inference is vital and generation is unnecessary. Training a PoE by maximizing the likelihood of the data is difficult because it is hard even to approximate the derivatives of the renormalization term in the combination rule. Fortunately, a PoE can be trained using a different objective function called \"contrastive divergence\" whose derivatives with regard to the parameters can be approximated accurately and efficiently. Examples are presented of contrastive divergence learning using several types of expert on several types of data.",
"In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention. In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning. We introduce a class of CNNs called deep convolutional generative adversarial networks (DCGANs), that have certain architectural constraints, and demonstrate that they are a strong candidate for unsupervised learning. Training on various image datasets, we show convincing evidence that our deep convolutional adversarial pair learns a hierarchy of representations from object parts to scenes in both the generator and discriminator. Additionally, we use the learned features for novel tasks - demonstrating their applicability as general image representations.",
"Many areas of science depend on exploratory data analysis and visualization. The need to analyze large amounts of multivariate data raises the fundamental problem of dimensionality reduction: how to discover compact representations of high-dimensional data. Here, we introduce locally linear embedding (LLE), an unsupervised learning algorithm that computes low-dimensional, neighborhood-preserving embeddings of high-dimensional inputs. Unlike clustering methods for local dimensionality reduction, LLE maps its inputs into a single global coordinate system of lower dimensionality, and its optimizations do not involve local minima. By exploiting the local symmetries of linear reconstructions, LLE is able to learn the global structure of nonlinear manifolds, such as those generated by images of faces or documents of text. How do we judge similarity? Our mental representations of the world are formed by processing large numbers of sensory in",
"This work explores the use of spatial context as a source of free and plentiful supervisory signal for training a rich visual representation. Given only a large, unlabeled image collection, we extract random pairs of patches from each image and train a convolutional neural net to predict the position of the second patch relative to the first. We argue that doing well on this task requires the model to learn to recognize objects and their parts. We demonstrate that the feature representation learned using this within-image context indeed captures visual similarity across images. For example, this representation allows us to perform unsupervised visual discovery of objects like cats, people, and even birds from the Pascal VOC 2011 detection dataset. Furthermore, we show that the learned ConvNet can be used in the R-CNN framework and provides a significant boost over a randomly-initialized ConvNet, resulting in state-of-the-art performance among algorithms which use only Pascal-provided training set annotations.",
"Traditional fluid simulations require large computational resources even for an average sized scene with the main bottleneck being a very small time step size, required to guarantee the stability of the solution. Despite a large progress in parallel computing and efficient algorithms for pressure computation in the recent years, realtime fluid simulations have been possible only under very restricted conditions. In this paper we propose a novel machine learning based approach, that formulates physics-based fluid simulation as a regression problem, estimating the acceleration of every particle for each frame. We designed a feature vector, directly modelling individual forces and constraints from the Navier-Stokes equations, giving the method strong generalization properties to reliably predict positions and velocities of particles in a large time step setting on yet unseen test videos. We used a regression forest to approximate the behaviour of particles observed in the large training set of simulations obtained using a traditional solver. Our GPU implementation led to a speed-up of one to three orders of magnitude compared to the state-of-the-art position-based fluid solver and runs in real-time for systems with up to 2 million particles.",
"We present a new technique called “t-SNE” that visualizes high-dimensional data by giving each datapoint a location in a two or three-dimensional map. The technique is a variation of Stochastic Neighbor Embedding (Hinton and Roweis, 2002) that is much easier to optimize, and produces significantly better visualizations by reducing the tendency to crowd points together in the center of the map. t-SNE is better than existing techniques at creating a single map that reveals structure at many different scales. This is particularly important for high-dimensional data that lie on several different, but related, low-dimensional manifolds, such as images of objects from multiple classes seen from multiple viewpoints. For visualizing the structure of very large datasets, we show how t-SNE can use random walks on neighborhood graphs to allow the implicit structure of all of the data to influence the way in which a subset of the data is displayed. We illustrate the performance of t-SNE on a wide variety of datasets and compare it with many other non-parametric visualization techniques, including Sammon mapping, Isomap, and Locally Linear Embedding. The visualizations produced by t-SNE are significantly better than those produced by the other techniques on almost all of the datasets.",
"In this paper we study the problem of image representation learning without human annotation. By following the principles of self-supervision, we build a convolutional neural network (CNN) that can be trained to solve Jigsaw puzzles as a pretext task, which requires no manual labeling, and then later repurposed to solve object classification and detection. To maintain the compatibility across tasks we introduce the context-free network (CFN), a siamese-ennead CNN. The CFN takes image tiles as input and explicitly limits the receptive field (or context) of its early processing units to one tile at a time. We show that the CFN includes fewer parameters than AlexNet while preserving the same semantic learning capabilities. By training the CFN to solve Jigsaw puzzles, we learn both a feature mapping of object parts as well as their correct spatial arrangement. Our experimental evaluations show that the learned features capture semantically relevant content. Our proposed method for learning visual representations outperforms state of the art methods in several transfer learning benchmarks.",
"Scientists working with large volumes of high-dimensional data, such as global climate patterns, stellar spectra, or human gene distributions, regularly confront the problem of dimensionality reduction: finding meaningful low-dimensional structures hidden in their high-dimensional observations. The human brain confronts the same problem in everyday perception, extracting from its high-dimensional sensory inputs—30,000 auditory nerve fibers or 106 optic nerve fibers—a manageably small number of perceptually relevant features. Here we describe an approach to solving dimensionality reduction problems that uses easily measured local metric information to learn the underlying global geometry of a data set. Unlike classical techniques such as principal component analysis (PCA) and multidimensional scaling (MDS), our approach is capable of discovering the nonlinear degrees of freedom that underlie complex natural observations, such as human handwriting or images of a face under different viewing conditions. In contrast to previous algorithms for nonlinear dimensionality reduction, ours efficiently computes a globally optimal solution, and, for an important class of data manifolds, is guaranteed to converge asymptotically to the true structure.",
"Introduction * Properties of Population Principal Components * Properties of Sample Principal Components * Interpreting Principal Components: Examples * Graphical Representation of Data Using Principal Components * Choosing a Subset of Principal Components or Variables * Principal Component Analysis and Factor Analysis * Principal Components in Regression Analysis * Principal Components Used with Other Multivariate Techniques * Outlier Detection, Influential Observations and Robust Estimation * Rotation and Interpretation of Principal Components * Principal Component Analysis for Time Series and Other Non-Independent Data * Principal Component Analysis for Special Types of Data * Generalizations and Adaptations of Principal Component Analysis",
"We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to ½ everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples."
]
} |
1611.07571 | 2949347953 | Several machine learning tasks require to represent the data using only a sparse set of interest points. An ideal detector is able to find the corresponding interest points even if the data undergo a transformation typical for a given domain. Since the task is of high practical interest in computer vision, many hand-crafted solutions were proposed. In this paper, we ask a fundamental question: can we learn such detectors from scratch? Since it is often unclear what points are "interesting", human labelling cannot be used to find a truly unbiased solution. Therefore, the task requires an unsupervised formulation. We are the first to propose such a formulation: training a neural network to rank points in a transformation-invariant manner. Interest points are then extracted from the top bottom quantiles of this ranking. We validate our approach on two tasks: standard RGB image interest point detection and challenging cross-modal interest point detection between RGB and depth images. We quantitatively show that our unsupervised method performs better or on-par with baselines. | While some tasks actually have a non-human label (for example, in solver simulation we can obtain the solution by running a solver), others (for example, representation learning) have none at all. Instead, they try to find an auxiliary task which is hard enough in order to learn a representation that is useful for already existing tasks (classification, for example). Designing such a task is non-trivial, therefore only few successful approaches exist (for example, @cite_19 ). | {
"cite_N": [
"@cite_19"
],
"mid": [
"2950187998"
],
"abstract": [
"This work explores the use of spatial context as a source of free and plentiful supervisory signal for training a rich visual representation. Given only a large, unlabeled image collection, we extract random pairs of patches from each image and train a convolutional neural net to predict the position of the second patch relative to the first. We argue that doing well on this task requires the model to learn to recognize objects and their parts. We demonstrate that the feature representation learned using this within-image context indeed captures visual similarity across images. For example, this representation allows us to perform unsupervised visual discovery of objects like cats, people, and even birds from the Pascal VOC 2011 detection dataset. Furthermore, we show that the learned ConvNet can be used in the R-CNN framework and provides a significant boost over a randomly-initialized ConvNet, resulting in state-of-the-art performance among algorithms which use only Pascal-provided training set annotations."
]
} |
1611.07571 | 2949347953 | Several machine learning tasks require to represent the data using only a sparse set of interest points. An ideal detector is able to find the corresponding interest points even if the data undergo a transformation typical for a given domain. Since the task is of high practical interest in computer vision, many hand-crafted solutions were proposed. In this paper, we ask a fundamental question: can we learn such detectors from scratch? Since it is often unclear what points are "interesting", human labelling cannot be used to find a truly unbiased solution. Therefore, the task requires an unsupervised formulation. We are the first to propose such a formulation: training a neural network to rank points in a transformation-invariant manner. Interest points are then extracted from the top bottom quantiles of this ranking. We validate our approach on two tasks: standard RGB image interest point detection and challenging cross-modal interest point detection between RGB and depth images. We quantitatively show that our unsupervised method performs better or on-par with baselines. | One particular application of our method is interest point detection in images. Most of the existing image interest point detectors are hand-crafted to select particular visual elements like blobs, corners or edges. These include the DoG detector @cite_8 , the Harris corner detector @cite_21 and its affine-covariant version @cite_5 , the FAST corner detector @cite_13 and the MSER detector @cite_10 . Most recently, there also emerged methods that do supervised learning building upon a hand-crafted solution: LIFT @cite_30 aims to extract an SfM-surviving subset of DoG detections, TILDE @cite_35 uses DoG for collecting the training set, @cite_34 samples training points only where LoG filter gives large absolute-value response. Building upon a hand-crafted detector restricts those supervised approaches to a subset of their basic method detections --- which makes those approaches inapplicable in the cases where there is no good detector yet. Our unsupervised method instead learns the detector completely from scratch by optimizing for a repeatable ranking. | {
"cite_N": [
"@cite_13",
"@cite_30",
"@cite_35",
"@cite_8",
"@cite_21",
"@cite_5",
"@cite_34",
"@cite_10"
],
"mid": [
"2069295011",
"2949896259",
"1945298332",
"2151103935",
"2111308925",
"",
"2952052436",
""
],
"abstract": [
"Abstract This paper describes a new corner detection algorithm, based on the property of corners that the change of image intensity should be high in all directions. Consequently, the corner response function (CRF) is computed as a minimum change of intensity over all possible directions. To compute the intensity change in an arbitrary direction an interpixel approximation is used. A multigrid approach is employed to reduce the computational complexity and to improve the quality of the detected corners. This algorithm, and other popular corner detectors, were evaluated and compared on the basis of their consistency, accuracy and speed using a range of images and video sequences. It was found that our algorithm performs well compared to the other algorithms, but it is significantly faster to compute.",
"We introduce a novel Deep Network architecture that implements the full feature point handling pipeline, that is, detection, orientation estimation, and feature description. While previous works have successfully tackled each one of these problems individually, we show how to learn to do all three in a unified manner while preserving end-to-end differentiability. We then demonstrate that our Deep pipeline outperforms state-of-the-art methods on a number of benchmark datasets, without the need of retraining.",
"We introduce a learning-based approach to detect repeatable keypoints under drastic imaging changes of weather and lighting conditions to which state-of-the-art keypoint detectors are surprisingly sensitive. We first identify good keypoint candidates in multiple training images taken from the same viewpoint. We then train a regressor to predict a score map whose maxima are those points so that they can be found by simple non-maximum suppression. As there are no standard datasets to test the influence of these kinds of changes, we created our own, which we will make publicly available. We will show that our method significantly outperforms the state-of-the-art methods in such challenging conditions, while still achieving state-of-the-art performance on untrained standard datasets.",
"This paper presents a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different views of an object or scene. The features are invariant to image scale and rotation, and are shown to provide robust matching across a substantial range of affine distortion, change in 3D viewpoint, addition of noise, and change in illumination. The features are highly distinctive, in the sense that a single feature can be correctly matched with high probability against a large database of features from many images. This paper also describes an approach to using these features for object recognition. The recognition proceeds by matching individual features to a database of features from known objects using a fast nearest-neighbor algorithm, followed by a Hough transform to identify clusters belonging to a single object, and finally performing verification through least-squares solution for consistent pose parameters. This approach to recognition can robustly identify objects among clutter and occlusion while achieving near real-time performance.",
"The problem we are addressing in Alvey Project MMI149 is that of using computer vision to understand the unconstrained 3D world, in which the viewed scenes will in general contain too wide a diversity of objects for topdown recognition techniques to work. For example, we desire to obtain an understanding of natural scenes, containing roads, buildings, trees, bushes, etc., as typified by the two frames from a sequence illustrated in Figure 1. The solution to this problem that we are pursuing is to use a computer vision system based upon motion analysis of a monocular image sequence from a mobile camera. By extraction and tracking of image features, representations of the 3D analogues of these features can be constructed.",
"",
"Local covariant feature detection, namely the problem of extracting viewpoint invariant features from images, has so far largely resisted the application of machine learning techniques. In this paper, we propose the first fully general formulation for learning local covariant feature detectors. We propose to cast detection as a regression problem, enabling the use of powerful regressors such as deep neural networks. We then derive a covariance constraint that can be used to automatically learn which visual structures provide stable anchors for local feature detection. We support these ideas theoretically, proposing a novel analysis of local features in term of geometric transformations, and we show that all common and many uncommon detectors can be derived in this framework. Finally, we present empirical results on translation and rotation covariant detectors on standard feature benchmarks, showing the power and flexibility of the framework.",
""
]
} |
1611.07571 | 2949347953 | Several machine learning tasks require to represent the data using only a sparse set of interest points. An ideal detector is able to find the corresponding interest points even if the data undergo a transformation typical for a given domain. Since the task is of high practical interest in computer vision, many hand-crafted solutions were proposed. In this paper, we ask a fundamental question: can we learn such detectors from scratch? Since it is often unclear what points are "interesting", human labelling cannot be used to find a truly unbiased solution. Therefore, the task requires an unsupervised formulation. We are the first to propose such a formulation: training a neural network to rank points in a transformation-invariant manner. Interest points are then extracted from the top bottom quantiles of this ranking. We validate our approach on two tasks: standard RGB image interest point detection and challenging cross-modal interest point detection between RGB and depth images. We quantitatively show that our unsupervised method performs better or on-par with baselines. | Finally, a particularly challenging case in image interest point detection is the cross-modal one: the interest points should be repeatable among different image modalities. Several works mention this complex issue ( @cite_38 , @cite_14 , @cite_20 , @cite_6 ) but do not propose a general solution. Our approach, on the contrary, is general in a sense that the same learning procedure could be applied to different tasks: we show it to work for RGB RGB and RGB depth modality pairs. | {
"cite_N": [
"@cite_38",
"@cite_14",
"@cite_6",
"@cite_20"
],
"mid": [
"2219084768",
"1974450833",
"1646493568",
""
],
"abstract": [
"Many existing approaches for image-to-geometry registration assume that either a textured 3D model or a good initial guess of the 3D pose is available to bootstrap the registration process. In this paper we consider the registration of photographs to 3D models even when no texture information is available. This is very challenging as we cannot rely on texture gradients, and even shading gradients are hard to estimate since the lighting conditions are unknown. To that end, we propose average shading gradients, a rendering technique that estimates the average gradient magnitude over all lighting directions under Lambertian shading. We use this gradient representation as the building block of a registration pipeline based on matching sparse features. To cope with inevitable false matches due to the missing texture information and to increase robustness, the pose of the 3D model is estimated in two stages. Coarse pose hypotheses are first obtained from a single correct match each, subsequently refined using SIFT flow, and finally verified. We apply our algorithm to registering images of real-world objects to untextured 3D meshes of limited accuracy.",
"This paper presents a novel feature point descriptor for the multispectral image case: Far-Infrared and Visible Spectrum images. It allows matching interest points on images of the same scene but acquired in different spectral bands. Initially, points of interest are detected on both images through a SIFT-like based scale space representation. Then, these points are characterized using an Edge Oriented Histogram (EOH) descriptor. Finally, points of interest from multispectral images are matched by finding nearest couples using the information from the descriptor. The provided experimental results and comparisons with similar methods show both the validity of the proposed approach as well as the improvements it offers with respect to the current state-of-the-art.",
"We have presented a new problem -- the wide multiple baseline stereo (WxBS) -- which considers matching of images that simultaneously differ in more than one image acquisition factor such as viewpoint, illumination, sensor type or where object appearance changes significantly, e.g. over time. A new dataset with the ground truth for evaluation of matching algorithms has been introduced and will be made public. We have extensively tested a large set of popular and recent detectors and descriptors and show than the combination of RootSIFT and HalfRootSIFT as descriptors with MSER and Hessian-Affine detectors works best for many different nuisance factors. We show that simple adaptive thresholding improves Hessian-Affine, DoG, MSER (and possibly other) detectors and allows to use them on infrared and low contrast images. A novel matching algorithm for addressing the WxBS problem has been introduced. We have shown experimentally that the WxBS-M matcher dominantes the state-of-the-art methods both on both the new and existing datasets.",
""
]
} |
1611.07212 | 2949905205 | We present an attention-based model that reasons on human body shape and motion dynamics to identify individuals in the absence of RGB information, hence in the dark. Our approach leverages unique 4D spatio-temporal signatures to address the identification problem across days. Formulated as a reinforcement learning task, our model is based on a combination of convolutional and recurrent neural networks with the goal of identifying small, discriminative regions indicative of human identity. We demonstrate that our model produces state-of-the-art results on several published datasets given only depth images. We further study the robustness of our model towards viewpoint, appearance, and volumetric changes. Finally, we share insights gleaned from interpretable 2D, 3D, and 4D visualizations of our model's spatio-temporal attention. | A small number of studies have explored the applicability of deep neural networks to person identification. In @cite_14 , proposed a siamese convolutional neural network for similarity metric learning. In @cite_52 , proposed a similar approach by using filter pairs to model photometric and geometric transforms. Following these works, @cite_12 formulated the input as a triplet containing both correct and incorrect reference images. In @cite_32 , introduced cross-input neighborhood differences. | {
"cite_N": [
"@cite_14",
"@cite_32",
"@cite_52",
"@cite_12"
],
"mid": [
"",
"1928419358",
"1982925187",
"2048110836"
],
"abstract": [
"",
"In this work, we propose a method for simultaneously learning features and a corresponding similarity metric for person re-identification. We present a deep convolutional architecture with layers specially designed to address the problem of re-identification. Given a pair of images as input, our network outputs a similarity value indicating whether the two input images depict the same person. Novel elements of our architecture include a layer that computes cross-input neighborhood differences, which capture local relationships between the two input images based on mid-level features from each input image. A high-level summary of the outputs of this layer is computed by a layer of patch summary features, which are then spatially integrated in subsequent layers. Our method significantly outperforms the state of the art on both a large data set (CUHK03) and a medium-sized data set (CUHK01), and is resistant to over-fitting. We also demonstrate that by initially training on an unrelated large data set before fine-tuning on a small target data set, our network can achieve results comparable to the state of the art even on a small data set (VIPeR).",
"Person re-identification is to match pedestrian images from disjoint camera views detected by pedestrian detectors. Challenges are presented in the form of complex variations of lightings, poses, viewpoints, blurring effects, image resolutions, camera settings, occlusions and background clutter across camera views. In addition, misalignment introduced by the pedestrian detector will affect most existing person re-identification methods that use manually cropped pedestrian images and assume perfect detection. In this paper, we propose a novel filter pairing neural network (FPNN) to jointly handle misalignment, photometric and geometric transforms, occlusions and background clutter. All the key components are jointly optimized to maximize the strength of each component when cooperating with others. In contrast to existing works that use handcrafted features, our method automatically learns features optimal for the re-identification task from data. The learned filter pairs encode photometric transforms. Its deep architecture makes it possible to model a mixture of complex photometric and geometric transforms. We build the largest benchmark re-id dataset with 13, 164 images of 1, 360 pedestrians. Unlike existing datasets, which only provide manually cropped pedestrian images, our dataset provides automatically detected bounding boxes for evaluation close to practical applications. Our neural network significantly outperforms state-of-the-art methods on this dataset.",
"This paper introduces Pairwise Constrained Component Analysis (PCCA), a new algorithm for learning distance metrics from sparse pairwise similarity dissimilarity constraints in high dimensional input space, problem for which most existing distance metric learning approaches are not adapted. PCCA learns a projection into a low-dimensional space where the distance between pairs of data points respects the desired constraints, exhibiting good generalization properties in presence of high dimensional data. The paper also shows how to efficiently kernelize the approach. PCCA is experimentally validated on two challenging vision tasks, face verification and person re-identification, for which we obtain state-of-the-art results."
]
} |
1611.07212 | 2949905205 | We present an attention-based model that reasons on human body shape and motion dynamics to identify individuals in the absence of RGB information, hence in the dark. Our approach leverages unique 4D spatio-temporal signatures to address the identification problem across days. Formulated as a reinforcement learning task, our model is based on a combination of convolutional and recurrent neural networks with the goal of identifying small, discriminative regions indicative of human identity. We demonstrate that our model produces state-of-the-art results on several published datasets given only depth images. We further study the robustness of our model towards viewpoint, appearance, and volumetric changes. Finally, we share insights gleaned from interpretable 2D, 3D, and 4D visualizations of our model's spatio-temporal attention. | By measuring the sensitivity of output variables to variances in the input, attention models applied to image classification @cite_6 @cite_8 @cite_1 , image captioning @cite_64 @cite_82 @cite_39 , object detection @cite_68 , and tracking @cite_53 have demystified many aspects of convolutional and recurrent networks. These methods exploit the spatial structure of the input to understand intermediate network representations. Sequential data, on the other hand, requires temporal attention models to understand the order dependence of the input data. Recent papers in speech recognition @cite_57 , video captioning @cite_83 , and natural language processing @cite_10 @cite_65 @cite_49 explore the concept of attention in the temporal domain. | {
"cite_N": [
"@cite_64",
"@cite_8",
"@cite_53",
"@cite_1",
"@cite_65",
"@cite_6",
"@cite_39",
"@cite_57",
"@cite_83",
"@cite_49",
"@cite_68",
"@cite_10",
"@cite_82"
],
"mid": [
"",
"",
"2154071538",
"",
"",
"2952186574",
"",
"2127141656",
"2950307714",
"",
"1902237438",
"",
""
],
"abstract": [
"",
"",
"We discuss an attentional model for simultaneous object tracking and recognition that is driven by gaze data. Motivated by theories of perception, the model consists of two interacting pathways, identity and control, intended to mirror the what and where pathways in neuroscience models. The identity pathway models object appearance and performs classification using deep (factored)-restricted Boltzmann machines. At each point in time, the observations consist of foveated images, with decaying resolution toward the periphery of the gaze. The control pathway models the location, orientation, scale, and speed of the attended object. The posterior distribution of these states is estimated with particle filtering. Deeper in the control pathway, we encounter an attentional mechanism that learns to select gazes so as to minimize tracking uncertainty. Unlike in our previous work, we introduce gaze selection strategies that operate in the presence of partial information and on a continuous action space. We show that a straightforward extension of the existing approach to the partial information setting results in poor performance, and we propose an alternative method based on modeling the reward surface as a gaussian process. This approach gives good performance in the presence of partial information and allows us to expand the action space from a small, discrete set of fixation points to a continuous domain.",
"",
"",
"Large Convolutional Network models have recently demonstrated impressive classification performance on the ImageNet benchmark. However there is no clear understanding of why they perform so well, or how they might be improved. In this paper we address both issues. We introduce a novel visualization technique that gives insight into the function of intermediate feature layers and the operation of the classifier. We also perform an ablation study to discover the performance contribution from different model layers. This enables us to find model architectures that outperform Krizhevsky al on the ImageNet classification benchmark. We show our ImageNet model generalizes well to other datasets: when the softmax classifier is retrained, it convincingly beats the current state-of-the-art results on Caltech-101 and Caltech-256 datasets.",
"",
"Many real-world sequence learning tasks require the prediction of sequences of labels from noisy, unsegmented input data. In speech recognition, for example, an acoustic signal is transcribed into words or sub-word units. Recurrent neural networks (RNNs) are powerful sequence learners that would seem well suited to such tasks. However, because they require pre-segmented training data, and post-processing to transform their outputs into label sequences, their applicability has so far been limited. This paper presents a novel method for training RNNs to label unsegmented sequences directly, thereby solving both problems. An experiment on the TIMIT speech corpus demonstrates its advantages over both a baseline HMM and a hybrid HMM-RNN.",
"Recent progress in using recurrent neural networks (RNNs) for image description has motivated the exploration of their application for video description. However, while images are static, working with videos requires modeling their dynamic temporal structure and then properly integrating that information into a natural language description. In this context, we propose an approach that successfully takes into account both the local and global temporal structure of videos to produce descriptions. First, our approach incorporates a spatial temporal 3-D convolutional neural network (3-D CNN) representation of the short temporal dynamics. The 3-D CNN representation is trained on video action recognition tasks, so as to produce a representation that is tuned to human motion and behavior. Second we propose a temporal attention mechanism that allows to go beyond local temporal modeling and learns to automatically select the most relevant temporal segments given the text-generating RNN. Our approach exceeds the current state-of-art for both BLEU and METEOR metrics on the Youtube2Text dataset. We also present results on a new, larger and more challenging dataset of paired video and natural language descriptions.",
"",
"An attentional mechanism has lately been used to improve neural machine translation (NMT) by selectively focusing on parts of the source sentence during translation. However, there has been little work exploring useful architectures for attention-based NMT. This paper examines two simple and effective classes of attentional mechanism: a global approach which always attends to all source words and a local one that only looks at a subset of source words at a time. We demonstrate the effectiveness of both approaches on the WMT translation tasks between English and German in both directions. With local attention, we achieve a significant gain of 5.0 BLEU points over non-attentional systems that already incorporate known techniques such as dropout. Our ensemble model using different attention architectures yields a new state-of-the-art result in the WMT’15 English to German translation task with 25.9 BLEU points, an improvement of 1.0 BLEU points over the existing best system backed by NMT and an n-gram reranker. 1",
"",
""
]
} |
1611.07212 | 2949905205 | We present an attention-based model that reasons on human body shape and motion dynamics to identify individuals in the absence of RGB information, hence in the dark. Our approach leverages unique 4D spatio-temporal signatures to address the identification problem across days. Formulated as a reinforcement learning task, our model is based on a combination of convolutional and recurrent neural networks with the goal of identifying small, discriminative regions indicative of human identity. We demonstrate that our model produces state-of-the-art results on several published datasets given only depth images. We further study the robustness of our model towards viewpoint, appearance, and volumetric changes. Finally, we share insights gleaned from interpretable 2D, 3D, and 4D visualizations of our model's spatio-temporal attention. | Many deep learning models impose constraints on the input. Due to the high dimensionality of images ( high pixel count), preprocessing often includes resizing and or cropping the original input image @cite_72 . Videos are often truncated to a fixed length for training. Due to computational limitations, this loss of information is necessary to constrain runtimes. In the next section, we describe our model and how we balance this trade-off by employing visual glimpses" @cite_7 which process small 4D regions with high fidelity and grow to larger regions with lower detail. | {
"cite_N": [
"@cite_72",
"@cite_7"
],
"mid": [
"1928906481",
"2951527505"
],
"abstract": [
"Fine-grained classification is challenging because categories can only be discriminated by subtle and local differences. Variances in the pose, scale or rotation usually make the problem more difficult. Most fine-grained classification systems follow the pipeline of finding foreground object or object parts (where) to extract discriminative features (what).",
"Applying convolutional neural networks to large images is computationally expensive because the amount of computation scales linearly with the number of image pixels. We present a novel recurrent neural network model that is capable of extracting information from an image or video by adaptively selecting a sequence of regions or locations and only processing the selected regions at high resolution. Like convolutional neural networks, the proposed model has a degree of translation invariance built-in, but the amount of computation it performs can be controlled independently of the input image size. While the model is non-differentiable, it can be trained using reinforcement learning methods to learn task-specific policies. We evaluate our model on several image classification tasks, where it significantly outperforms a convolutional neural network baseline on cluttered images, and on a dynamic visual control problem, where it learns to track a simple object without an explicit training signal for doing so."
]
} |
1611.07675 | 2951159095 | Automatically generating natural language descriptions of videos plays a fundamental challenge for computer vision community. Most recent progress in this problem has been achieved through employing 2-D and or 3-D Convolutional Neural Networks (CNN) to encode video content and Recurrent Neural Networks (RNN) to decode a sentence. In this paper, we present Long Short-Term Memory with Transferred Semantic Attributes (LSTM-TSA)---a novel deep architecture that incorporates the transferred semantic attributes learnt from images and videos into the CNN plus RNN framework, by training them in an end-to-end manner. The design of LSTM-TSA is highly inspired by the facts that 1) semantic attributes play a significant contribution to captioning, and 2) images and videos carry complementary semantics and thus can reinforce each other for captioning. To boost video captioning, we propose a novel transfer unit to model the mutually correlated attributes learnt from images and videos. Extensive experiments are conducted on three public datasets, i.e., MSVD, M-VAD and MPII-MD. Our proposed LSTM-TSA achieves to-date the best published performance in sentence generation on MSVD: 52.8 and 74.0 in terms of BLEU@4 and CIDEr-D. Superior results when compared to state-of-the-art methods are also reported on M-VAD and MPII-MD. | Attributes are properties observed in visual content with rich semantic cues and have been widely studied in computer vision for improving the efficacy of visual recognition @cite_4 . Following this elegant recipe, several recent works have attempted to inject attributes into sequence learning for image caption generation. Fang @cite_10 firstly use Multiple Instance Learning to train attributes detector and then generate sentence through a maximum-entropy language model based on the outputs of attributes detector. Later in @cite_26 , this framework is further developed with a larger range of attributes, additionally including celebrities and landmarks, to enrich the generated sentence. More recently, in @cite_14 , high-level concepts attributes are shown to obtain clear improvements on image captioning task when injected into existing state-of-the-art RNN-based model and such visual attributes are also utilized as semantic attention in @cite_19 to enhance image captioning. | {
"cite_N": [
"@cite_14",
"@cite_26",
"@cite_4",
"@cite_19",
"@cite_10"
],
"mid": [
"2404394533",
"2952238956",
"",
"2953022248",
"2949769367"
],
"abstract": [
"Much of the recent progress in Vision-to-Language (V2L) problems has been achieved through a combination of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). This approach does not explicitly represent high-level semantic concepts, but rather seeks to progress directly from image features to text. We propose here a method of incorporating high-level concepts into the very successful CNN-RNN approach, and show that it achieves a significant improvement on the state-of-the-art performance in both image captioning and visual question answering. We also show that the same mechanism can be used to introduce external semantic information and that doing so further improves performance. In doing so we provide an analysis of the value of high level semantic information in V2L problems.",
"We present an image caption system that addresses new challenges of automatically describing images in the wild. The challenges include high quality caption quality with respect to human judgments, out-of-domain data handling, and low latency required in many applications. Built on top of a state-of-the-art framework, we developed a deep vision model that detects a broad range of visual concepts, an entity recognition model that identifies celebrities and landmarks, and a confidence model for the caption output. Experimental results show that our caption engine outperforms previous state-of-the-art systems significantly on both in-domain dataset (i.e. MS COCO) and out of-domain datasets.",
"",
"Automatically generating a natural language description of an image has attracted interests recently both because of its importance in practical applications and because it connects two major artificial intelligence fields: computer vision and natural language processing. Existing approaches are either top-down, which start from a gist of an image and convert it into words, or bottom-up, which come up with words describing various aspects of an image and then combine them. In this paper, we propose a new algorithm that combines both approaches through a model of semantic attention. Our algorithm learns to selectively attend to semantic concept proposals and fuse them into hidden states and outputs of recurrent neural networks. The selection and fusion form a feedback connecting the top-down and bottom-up computation. We evaluate our algorithm on two public benchmarks: Microsoft COCO and Flickr30K. Experimental results show that our algorithm significantly outperforms the state-of-the-art approaches consistently across different evaluation metrics.",
"This paper presents a novel approach for automatically generating image descriptions: visual detectors, language models, and multimodal similarity models learnt directly from a dataset of image captions. We use multiple instance learning to train visual detectors for words that commonly occur in captions, including many different parts of speech such as nouns, verbs, and adjectives. The word detector outputs serve as conditional inputs to a maximum-entropy language model. The language model learns from a set of over 400,000 image descriptions to capture the statistics of word usage. We capture global semantics by re-ranking caption candidates using sentence-level features and a deep multimodal similarity model. Our system is state-of-the-art on the official Microsoft COCO benchmark, producing a BLEU-4 score of 29.1 . When human judges compare the system captions to ones written by other people on our held-out test set, the system captions have equal or better quality 34 of the time."
]
} |
1611.06949 | 2950371778 | Dense captioning is a newly emerging computer vision topic for understanding images with dense language descriptions. The goal is to densely detect visual concepts (e.g., objects, object parts, and interactions between them) from images, labeling each with a short descriptive phrase. We identify two key challenges of dense captioning that need to be properly addressed when tackling the problem. First, dense visual concept annotations in each image are associated with highly overlapping target regions, making accurate localization of each visual concept challenging. Second, the large amount of visual concepts makes it hard to recognize each of them by appearance alone. We propose a new model pipeline based on two novel ideas, joint inference and context fusion, to alleviate these two challenges. We design our model architecture in a methodical manner and thoroughly evaluate the variations in architecture. Our final model, compact and efficient, achieves state-of-the-art accuracy on Visual Genome for dense captioning with a relative gain of 73 compared to the previous best algorithm. Qualitative experiments also reveal the semantic capabilities of our model in dense captioning. | Recent image captioning models often utilize a convolutional neural network (CNN) @cite_38 as an image encoder and a recurrent neural network (RNN) @cite_41 as a decoder for predicting a sentence @cite_0 @cite_33 @cite_3 . RNNs have been widely used in language modeling @cite_37 @cite_7 @cite_26 @cite_23 . Some image captioning approaches, though targeted at a global description, also build relationships with local visual elements. Karpathy al @cite_33 @cite_4 learn an embedding with a latent alignment between image regions and word phrases. Fang al @cite_43 first detect words from images using multiple instance learning, then incorporate the words in a maximum entropy language model. A soft-attention mechanism is also proposed to cast attention over different image regions when predicting each word @cite_19 @cite_13 . | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_26",
"@cite_4",
"@cite_33",
"@cite_7",
"@cite_41",
"@cite_3",
"@cite_0",
"@cite_43",
"@cite_19",
"@cite_23",
"@cite_13"
],
"mid": [
"2310919327",
"2132339004",
"179875071",
"2953276893",
"2951805548",
"",
"1971129545",
"2951912364",
"2951183276",
"2949769367",
"1785460851",
"196214544",
"2950178297"
],
"abstract": [
"",
"A goal of statistical language modeling is to learn the joint probability function of sequences of words in a language. This is intrinsically difficult because of the curse of dimensionality: a word sequence on which the model will be tested is likely to be different from all the word sequences seen during training. Traditional but very successful approaches based on n-grams obtain generalization by concatenating very short overlapping sequences seen in the training set. We propose to fight the curse of dimensionality by learning a distributed representation for words which allows each training sentence to inform the model about an exponential number of semantically neighboring sentences. The model learns simultaneously (1) a distributed representation for each word along with (2) the probability function for word sequences, expressed in terms of these representations. Generalization is obtained because a sequence of words that has never been seen before gets high probability if it is made of words that are similar (in the sense of having a nearby representation) to words forming an already seen sentence. Training such large models (with millions of parameters) within a reasonable time is itself a significant challenge. We report on experiments using neural networks for the probability function, showing on two text corpora that the proposed approach significantly improves on state-of-the-art n-gram models, and that the proposed approach allows to take advantage of longer contexts.",
"A new recurrent neural network based language model (RNN LM) with applications to speech recognition is presented. Results indicate that it is possible to obtain around 50 reduction of perplexity by using mixture of several RNN LMs, compared to a state of the art backoff language model. Speech recognition experiments show around 18 reduction of word error rate on the Wall Street Journal task when comparing models trained on the same amount of data, and around 5 on the much harder NIST RT05 task, even when the backoff model is trained on much more data than the RNN LM. We provide ample empirical evidence to suggest that connectionist language models are superior to standard n-gram techniques, except their high computational (training) complexity. Index Terms: language modeling, recurrent neural networks, speech recognition",
"We introduce a model for bidirectional retrieval of images and sentences through a multi-modal embedding of visual and natural language data. Unlike previous models that directly map images or sentences into a common embedding space, our model works on a finer level and embeds fragments of images (objects) and fragments of sentences (typed dependency tree relations) into a common space. In addition to a ranking objective seen in previous work, this allows us to add a new fragment alignment objective that learns to directly associate these fragments across modalities. Extensive experimental evaluation shows that reasoning on both the global level of images and sentences and the finer level of their respective fragments significantly improves performance on image-sentence retrieval tasks. Additionally, our model provides interpretable predictions since the inferred inter-modal fragment alignment is explicit.",
"We present a model that generates natural language descriptions of images and their regions. Our approach leverages datasets of images and their sentence descriptions to learn about the inter-modal correspondences between language and visual data. Our alignment model is based on a novel combination of Convolutional Neural Networks over image regions, bidirectional Recurrent Neural Networks over sentences, and a structured objective that aligns the two modalities through a multimodal embedding. We then describe a Multimodal Recurrent Neural Network architecture that uses the inferred alignments to learn to generate novel descriptions of image regions. We demonstrate that our alignment model produces state of the art results in retrieval experiments on Flickr8K, Flickr30K and MSCOCO datasets. We then show that the generated descriptions significantly outperform retrieval baselines on both full images and on a new dataset of region-level annotations.",
"",
"Abstract Backpropagation is often viewed as a method for adapting artificial neural networks to classify patterns. Based on parts of the book by Rumelhart and colleagues, many authors equate backpropagation with the generalized delta rule applied to fully-connected feedforward networks. This paper will summarize a more general formulation of backpropagation, developed in 1974, which does more justice to the roots of the method in numerical analysis and statistics, and also does more justice to creative approaches expressed by neural modelers in the past year or two. It will discuss applications of backpropagation to forecasting over time (where errors have been halved by using methods other than least squares), to optimization, to sensitivity analysis, and to brain research. This paper will go on to derive a generalization of backpropagation to recurrent systems (which input their own output), such as hybrids of perceptron-style networks and Grossberg Hopfield networks. Unlike the proposal of Rumelhart, Hinton, and Williams, this generalization does not require the storage of intermediate iterations to deal with continuous recurrence. This generalization was applied in 1981 to a model of natural gas markets, where it located sources of forecast uncertainty related to the use of least squares to estimate the model parameters in the first place.",
"Automatically describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision and natural language processing. In this paper, we present a generative model based on a deep recurrent architecture that combines recent advances in computer vision and machine translation and that can be used to generate natural sentences describing an image. The model is trained to maximize the likelihood of the target description sentence given the training image. Experiments on several datasets show the accuracy of the model and the fluency of the language it learns solely from image descriptions. Our model is often quite accurate, which we verify both qualitatively and quantitatively. For instance, while the current state-of-the-art BLEU-1 score (the higher the better) on the Pascal dataset is 25, our approach yields 59, to be compared to human performance around 69. We also show BLEU-1 score improvements on Flickr30k, from 56 to 66, and on SBU, from 19 to 28. Lastly, on the newly released COCO dataset, we achieve a BLEU-4 of 27.7, which is the current state-of-the-art.",
"Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent, or \"temporally deep\", are effective for tasks involving sequences, visual and otherwise. We develop a novel recurrent convolutional architecture suitable for large-scale visual learning which is end-to-end trainable, and demonstrate the value of these models on benchmark video recognition tasks, image description and retrieval problems, and video narration challenges. In contrast to current models which assume a fixed spatio-temporal receptive field or simple temporal averaging for sequential processing, recurrent convolutional models are \"doubly deep\"' in that they can be compositional in spatial and temporal \"layers\". Such models may have advantages when target concepts are complex and or training data are limited. Learning long-term dependencies is possible when nonlinearities are incorporated into the network state updates. Long-term RNN models are appealing in that they directly can map variable-length inputs (e.g., video frames) to variable length outputs (e.g., natural language text) and can model complex temporal dynamics; yet they can be optimized with backpropagation. Our recurrent long-term models are directly connected to modern visual convnet models and can be jointly trained to simultaneously learn temporal dynamics and convolutional perceptual representations. Our results show such models have distinct advantages over state-of-the-art models for recognition or generation which are separately defined and or optimized.",
"This paper presents a novel approach for automatically generating image descriptions: visual detectors, language models, and multimodal similarity models learnt directly from a dataset of image captions. We use multiple instance learning to train visual detectors for words that commonly occur in captions, including many different parts of speech such as nouns, verbs, and adjectives. The word detector outputs serve as conditional inputs to a maximum-entropy language model. The language model learns from a set of over 400,000 image descriptions to capture the statistics of word usage. We capture global semantics by re-ranking caption candidates using sentence-level features and a deep multimodal similarity model. Our system is state-of-the-art on the official Microsoft COCO benchmark, producing a BLEU-4 score of 29.1 . When human judges compare the system captions to ones written by other people on our held-out test set, the system captions have equal or better quality 34 of the time.",
"Recent progress on automatic generation of image captions has shown that it is possible to describe the most salient information conveyed by images with accurate and meaningful sentences. In this paper, we propose an image caption system that exploits the parallel structures between images and sentences. In our model, the process of generating the next word, given the previously generated ones, is aligned with the visual perception experience where the attention shifting among the visual regions imposes a thread of visual ordering. This alignment characterizes the flow of \"abstract meaning\", encoding what is semantically shared by both the visual scene and the text description. Our system also makes another novel modeling contribution by introducing scene-specific contexts that capture higher-level semantic information encoded in an image. The contexts adapt language models for word generation to specific scene types. We benchmark our system and contrast to published results on several popular datasets. We show that using either region-based attention or scene-specific contexts improves systems without those components. Furthermore, combining these two modeling ingredients attains the state-of-the-art performance.",
"Recurrent Neural Networks (RNNs) are very powerful sequence models that do not enjoy widespread use because it is extremely difficult to train them properly. Fortunately, recent advances in Hessian-free optimization have been able to overcome the difficulties associated with training RNNs, making it possible to apply them successfully to challenging sequence problems. In this paper we demonstrate the power of RNNs trained with the new Hessian-Free optimizer (HF) by applying them to character-level language modeling tasks. The standard RNN architecture, while effective, is not ideally suited for such tasks, so we introduce a new RNN variant that uses multiplicative (or \"gated\") connections which allow the current input character to determine the transition matrix from one hidden state vector to the next. After training the multiplicative RNN with the HF optimizer for five days on 8 high-end Graphics Processing Units, we were able to surpass the performance of the best previous single method for character-level language modeling – a hierarchical non-parametric sequence model. To our knowledge this represents the largest recurrent neural network application to date.",
"Inspired by recent work in machine translation and object detection, we introduce an attention based model that automatically learns to describe the content of images. We describe how we can train this model in a deterministic manner using standard backpropagation techniques and stochastically by maximizing a variational lower bound. We also show through visualization how the model is able to automatically learn to fix its gaze on salient objects while generating the corresponding words in the output sequence. We validate the use of attention with state-of-the-art performance on three benchmark datasets: Flickr8k, Flickr30k and MS COCO."
]
} |
1611.06949 | 2950371778 | Dense captioning is a newly emerging computer vision topic for understanding images with dense language descriptions. The goal is to densely detect visual concepts (e.g., objects, object parts, and interactions between them) from images, labeling each with a short descriptive phrase. We identify two key challenges of dense captioning that need to be properly addressed when tackling the problem. First, dense visual concept annotations in each image are associated with highly overlapping target regions, making accurate localization of each visual concept challenging. Second, the large amount of visual concepts makes it hard to recognize each of them by appearance alone. We propose a new model pipeline based on two novel ideas, joint inference and context fusion, to alleviate these two challenges. We design our model architecture in a methodical manner and thoroughly evaluate the variations in architecture. Our final model, compact and efficient, achieves state-of-the-art accuracy on Visual Genome for dense captioning with a relative gain of 73 compared to the previous best algorithm. Qualitative experiments also reveal the semantic capabilities of our model in dense captioning. | Recent object detection algorithms based on deep learning often show a two-stage paradigm: region proposal and detection @cite_30 @cite_1 @cite_11 . Faster R-CNN @cite_11 is the most related to our work, as it utilizes a Region Proposal Network (RPN) to generate region proposals and a detection network to predict object categories and bounding box offsets. The two networks can share convolutional features and can be trained with an approximate fast joint training algorithm. A recent improvement to faster R-CNN is the incorporation of context information using a four-direction RNN on the convolutional feature map @cite_15 . Visual context can greatly help tasks such as object detection @cite_15 @cite_34 @cite_12 and semantic segmentation @cite_12 . Another direction is to remove the RPN and directly produce detection results @cite_2 @cite_31 to further speed up the algorithm. | {
"cite_N": [
"@cite_30",
"@cite_31",
"@cite_1",
"@cite_2",
"@cite_15",
"@cite_34",
"@cite_12",
"@cite_11"
],
"mid": [
"",
"",
"2102605133",
"2193145675",
"",
"2141364309",
"2125215748",
"2953106684"
],
"abstract": [
"",
"",
"Object detection performance, as measured on the canonical PASCAL VOC dataset, has plateaued in the last few years. The best-performing methods are complex ensemble systems that typically combine multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30 relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3 . Our approach combines two key insights: (1) one can apply high-capacity convolutional neural networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data is scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, yields a significant performance boost. Since we combine region proposals with CNNs, we call our method R-CNN: Regions with CNN features. We also present experiments that provide insight into what the network learns, revealing a rich hierarchy of image features. Source code for the complete system is available at http: www.cs.berkeley.edu rbg rcnn.",
"We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For (300 300 ) input, SSD achieves 74.3 mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for (512 512 ) input, SSD achieves 76.9 mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https: github.com weiliu89 caffe tree ssd.",
"",
"This paper presents an empirical evaluation of the role of context in a contemporary, challenging object detection task - the PASCAL VOC 2008. Previous experiments with context have mostly been done on home-grown datasets, often with non-standard baselines, making it difficult to isolate the contribution of contextual information. In this work, we present our analysis on a standard dataset, using top-performing local appearance detectors as baseline. We evaluate several different sources of context and ways to utilize it. While we employ many contextual cues that have been used before, we also propose a few novel ones including the use of geographic context and a new approach for using object spatial support.",
"In this paper we study the role of context in existing state-of-the-art detection and segmentation approaches. Towards this goal, we label every pixel of PASCAL VOC 2010 detection challenge with a semantic category. We believe this data will provide plenty of challenges to the community, as it contains 520 additional classes for semantic segmentation and object detection. Our analysis shows that nearest neighbor based approaches perform poorly on semantic segmentation of contextual classes, showing the variability of PASCAL imagery. Furthermore, improvements of exist ing contextual models for detection is rather modest. In order to push forward the performance in this difficult scenario, we propose a novel deformable part-based model, which exploits both local context around each candidate detection as well as global context at the level of the scene. We show that this contextual reasoning significantly helps in detecting objects at all scales.",
"State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet and Fast R-CNN have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully convolutional network that simultaneously predicts object bounds and objectness scores at each position. The RPN is trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. We further merge RPN and Fast R-CNN into a single network by sharing their convolutional features---using the recently popular terminology of neural networks with 'attention' mechanisms, the RPN component tells the unified network where to look. For the very deep VGG-16 model, our detection system has a frame rate of 5fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015 competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning entries in several tracks. Code has been made publicly available."
]
} |
1611.06949 | 2950371778 | Dense captioning is a newly emerging computer vision topic for understanding images with dense language descriptions. The goal is to densely detect visual concepts (e.g., objects, object parts, and interactions between them) from images, labeling each with a short descriptive phrase. We identify two key challenges of dense captioning that need to be properly addressed when tackling the problem. First, dense visual concept annotations in each image are associated with highly overlapping target regions, making accurate localization of each visual concept challenging. Second, the large amount of visual concepts makes it hard to recognize each of them by appearance alone. We propose a new model pipeline based on two novel ideas, joint inference and context fusion, to alleviate these two challenges. We design our model architecture in a methodical manner and thoroughly evaluate the variations in architecture. Our final model, compact and efficient, achieves state-of-the-art accuracy on Visual Genome for dense captioning with a relative gain of 73 compared to the previous best algorithm. Qualitative experiments also reveal the semantic capabilities of our model in dense captioning. | The task of dense captioning was first proposed in @cite_20 , in which a spatial transformer network @cite_28 is used to facilitate joint training of the whole network. A related application is also proposed to detect an arbitrary phrase in images using the dense captioning model. The experiments are conducted on the Visual Genome dataset @cite_35 , which provides not only region descriptions but also objects, attributes, question answering pairs, etc. Also closely related are other recent topics such as localizing a phrase in a specific image @cite_5 @cite_32 @cite_8 , generating an unambiguous description for a specific region in an image @cite_32 @cite_17 , or detecting visual relationships in images @cite_10 @cite_25 . | {
"cite_N": [
"@cite_35",
"@cite_8",
"@cite_28",
"@cite_32",
"@cite_5",
"@cite_10",
"@cite_25",
"@cite_20",
"@cite_17"
],
"mid": [
"2949474740",
"2505639562",
"2951005624",
"2144960104",
"2963735856",
"2594494421",
"2479423890",
"2963758027",
"2949107813"
],
"abstract": [
"Despite progress in perceptual tasks such as image classification, computers still perform poorly on cognitive tasks such as image description and question answering. Cognition is core to tasks that involve not just recognizing, but reasoning about our visual world. However, models used to tackle the rich content in images for cognitive tasks are still being trained using the same datasets designed for perceptual tasks. To achieve success at cognitive tasks, models need to understand the interactions and relationships between objects in an image. When asked \"What vehicle is the person riding?\", computers will need to identify the objects in an image as well as the relationships riding(man, carriage) and pulling(horse, carriage) in order to answer correctly that \"the person is riding a horse-drawn carriage\". In this paper, we present the Visual Genome dataset to enable the modeling of such relationships. We collect dense annotations of objects, attributes, and relationships within each image to learn these models. Specifically, our dataset contains over 100K images where each image has an average of 21 objects, 18 attributes, and 18 pairwise relationships between objects. We canonicalize the objects, attributes, relationships, and noun phrases in region descriptions and questions answer pairs to WordNet synsets. Together, these annotations represent the densest and largest dataset of image descriptions, objects, attributes, relationships, and question answers.",
"Referring expressions usually describe an object using properties of the object and relationships of the object with other objects. We propose a technique that integrates context between objects to understand referring expressions. Our approach uses an LSTM to learn the probability of a referring expression, with input features from a region and a context region. The context regions are discovered using multiple-instance learning (MIL) since annotations for context objects are generally not available for training. We utilize max-margin based MIL objective functions for training the LSTM. Experiments on the Google RefExp and UNC RefExp datasets show that modeling context between objects provides better performance than modeling only object properties. We also qualitatively show that our technique can ground a referring expression to its referred region along with the supporting context region.",
"Convolutional Neural Networks define an exceptionally powerful class of models, but are still limited by the lack of ability to be spatially invariant to the input data in a computationally and parameter efficient manner. In this work we introduce a new learnable module, the Spatial Transformer, which explicitly allows the spatial manipulation of data within the network. This differentiable module can be inserted into existing convolutional architectures, giving neural networks the ability to actively spatially transform feature maps, conditional on the feature map itself, without any extra training supervision or modification to the optimisation process. We show that the use of spatial transformers results in models which learn invariance to translation, scale, rotation and more generic warping, resulting in state-of-the-art performance on several benchmarks, and for a number of classes of transformations.",
"We propose a method that can generate an unambiguous description (known as a referring expression) of a specific object or region in an image, and which can also comprehend or interpret such an expression to infer which object is being described. We show that our method outperforms previous methods that generate descriptions of objects without taking into account other potentially ambiguous objects in the scene. Our model is inspired by recent successes of deep learning methods for image captioning, but while image captioning is difficult to evaluate, our task allows for easy objective evaluation. We also present a new large-scale dataset for referring expressions, based on MSCOCO. We have released the dataset and a toolbox for visualization and evaluation, see https: github.com mjhucla Google_Refexp_toolbox.",
"In this paper, we address the task of natural language object retrieval, to localize a target object within a given image based on a natural language query of the object. Natural language object retrieval differs from text-based image retrieval task as it involves spatial information about objects within the scene and global scene context. To address this issue, we propose a novel Spatial Context Recurrent ConvNet (SCRC) model as scoring function on candidate boxes for object retrieval, integrating spatial configurations and global scene-level contextual information into the network. Our model processes query text, local image descriptors, spatial configurations and global context features through a recurrent network, outputs the probability of the query text conditioned on each candidate box as a score for the box, and can transfer visual-linguistic knowledge from image captioning domain to our task. Experimental results demonstrate that our method effectively utilizes both local and global information, outperforming previous baseline methods significantly on different datasets and scenarios, and can exploit large scale vision and language datasets for knowledge transfer.",
"",
"Visual relationships capture a wide variety of interactions between pairs of objects in images (e.g. “man riding bicycle” and “man pushing bicycle”). Consequently, the set of possible relationships is extremely large and it is difficult to obtain sufficient training examples for all possible relationships. Because of this limitation, previous work on visual relationship detection has concentrated on predicting only a handful of relationships. Though most relationships are infrequent, their objects (e.g. “man” and “bicycle”) and predicates (e.g. “riding” and “pushing”) independently occur more frequently. We propose a model that uses this insight to train visual models for objects and predicates individually and later combines them together to predict multiple relationships per image. We improve on prior work by leveraging language priors from semantic word embeddings to finetune the likelihood of a predicted relationship. Our model can scale to predict thousands of types of relationships from a few examples. Additionally, we localize the objects in the predicted relationships as bounding boxes in the image. We further demonstrate that understanding relationships can improve content based image retrieval.",
"We introduce the dense captioning task, which requires a computer vision system to both localize and describe salient regions in images in natural language. The dense captioning task generalizes object detection when the descriptions consist of a single word, and Image Captioning when one predicted region covers the full image. To address the localization and description task jointly we propose a Fully Convolutional Localization Network (FCLN) architecture that processes an image with a single, efficient forward pass, requires no external regions proposals, and can be trained end-to-end with a single round of optimization. The architecture is composed of a Convolutional Network, a novel dense localization layer, and Recurrent Neural Network language model that generates the label sequences. We evaluate our network on the Visual Genome dataset, which comprises 94,000 images and 4,100,000 region-grounded captions. We observe both speed and accuracy improvements over baselines based on current state of the art approaches in both generation and retrieval settings.",
"Humans refer to objects in their environments all the time, especially in dialogue with other people. We explore generating and comprehending natural language referring expressions for objects in images. In particular, we focus on incorporating better measures of visual context into referring expression models and find that visual comparison to other objects within an image helps improve performance significantly. We also develop methods to tie the language generation process together, so that we generate expressions for all objects of a particular category jointly. Evaluation on three recent datasets - RefCOCO, RefCOCO+, and RefCOCOg, shows the advantages of our methods for both referring expression generation and comprehension."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | In 1965, Golub and Kahan @cite_1 provides a singular value solver based on an initial reduction to bidiagonal form. In [Th. 1] doi:10.1137 0702016 , the GE2BD step is done using a QR step on the first column, then an LQ step on the first row, then a QR step on the second column, etc. The steps are done one column at a time using Householder transformation. This algorithm is implemented as a Level-2 BLAS algorithm in as xGEBD2 . For an @math -by- @math matrix, the cost of this algorithm is (approximately) @math . | {
"cite_N": [
"@cite_1"
],
"mid": [
"2005423095"
],
"abstract": [
"A numerically stable and fairly fast scheme is described to compute the unitary matrices U and V which transform a given matrix A into a diagonal form @math , thus exhibiting A’s singular values on @math ’s diagonal. The scheme first transforms A to a bidiagonal matrix J, then diagonalizes J. The scheme described here is complicated but does not suffer from the computational difficulties which occasionally afflict some previously known methods. Some applications are mentioned, in particular the use of the pseudo-inverse @math to solve least squares problems in a way which dampens spurious oscillation and cancellation."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | Preprocessing the bidiagonalization with a QR factorization (preQR step). In 1982, Chan @cite_5 explains that, for tall-and-skinny matrices, in order to perform less flops, one can pre-process the bidiagonalization step (GE2BD) with a QR factorization. In other words, Chan propose to do preQR( @math , @math )+GE2BD( @math , @math ) instead of GE2BD( @math , @math ). A curiosity of this algorithm is that it introduces nonzeros where zeros were previously introduced; yet, there is a gain in term of flops. Chan proves that the crossover points when preQR( @math , @math )+GE2BD( @math , @math ) performs less flops than GE2BD( @math , @math ) is when @math is greater than @math . Chan also proved that, asymptotically, preQR( @math , @math )+GE2BD( @math , @math ) will perform half the flops than GE2BD( @math , @math ) for a fixed @math and @math going to infinity. If the singular vectors are sought, preQR has more overhead: (1) the crossover point is moved to more tall-and-skinny matrices, and there is less gain; also (2) there is some complication as far as storage goes. | {
"cite_N": [
"@cite_5"
],
"mid": [
"2117026915"
],
"abstract": [
"The most well-known and widely used algorithm for computing the Singular Value Decomposition (SVD) A --U V T of an m x n rectangular matrix A is the Golub-Reinsch algorithm (GR-SVD). In this paper, an improved version of the original GR-SVD algorithm is presented. The new algorithm works best for matrices with m >> n, but is more efficient even when m is only slightly greater than n (usually when m 2n) and in some cases can achieve as much as 50 percent savings. If the matrix U s exphcltly desired, then n 2 extra storage locations are required, but otherwise no extra storage is needed. The two main modifications are: (1) first triangularizing A by Householder transformations before bldmgonahzing it (thin idea seems to be widely known among some researchers in the field, but as far as can be determined, neither a detailed analysis nor an lmplementatmn has been published before), and (2) accumulating the left Givens transformations in GR-SVD on an n x n array instead of on an m x n array. A PFORT-verified FORTRAN Implementation m included. Comparisons with the EISPACK SVD routine are given."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | Two-step approach: GE2BND+BND2BD. In 1999, Gro er and Lang @cite_14 studied a two-step approach for GE2BD: (1) go from full to band (GE2BND), (2) then go from band to bidiagonal (BND2BD). In this scenario, GE2BND has most of the flops and performs using Level-3 BLAS kernels; BND2BD is not using Level-3 BLAS but it executes much less flops and operates on a smaller data footprint that might fit better in cache. There is a trade-off for the bandwidth to be chosen. If the bandwidth is too small, then the first step (GE2BND) will have the same issues as GE2BD. If the bandwidth is too large, then the second step BND2BD will have many flops and dominates the run time. | {
"cite_N": [
"@cite_14"
],
"mid": [
"2087716616"
],
"abstract": [
"Abstract Most methods for calculating the SVD (singular value decomposition) require to first bidiagonalize the matrix. The blocked reduction of a general, dense matrix to bidiagonal form, as implemented in ScaLAPACK, does about one half of the operations with BLAS3. By subdividing the reduction into two stages dense → banded and banded → bidiagonal with cubic and quadratic arithmetic costs, respectively, we are able to carry out a much higher portion of the calculations in matrix–matrix multiplications. Thus, higher performance can be expected. This paper presents and compares three parallel techniques for reducing a full matrix to banded form. (The second reduction stage is described in another paper [B. Lang, Parallel Comput. 22 (1996) 1–18]). Numerical experiments on the Intel Paragon and IBM SP 1 distributed memory parallel computers demonstrate that the two-stage reduction approach can be significantly superior if only the singular values are required."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | As mentioned earlier, when the singular vectors are sought (GESVD), the overhead of the method in term of flops is quite large. In 2013, Haidar, Kurzak, and Luszczek @cite_25 reported that despite the extra flops a two-step approach leads to speedup even when singular vectors are sought (GESVD). | {
"cite_N": [
"@cite_25"
],
"mid": [
"2156779053"
],
"abstract": [
"The enormous gap between the high-performance capabilities of today's CPUs and off-chip communication poses extreme challenges to the development of numerical software that is scalable and achieves high performance. In this article, we describe a successful methodology to address these challenges---starting with our algorithm design, through kernel optimization and tuning, and finishing with our programming model. All these lead to development of a scalable high-performance Singular Value Decomposition (SVD) solver. We developed a set of highly optimized kernels and combined them with advanced optimization techniques that feature fine-grain and cache-contained kernels, a task based approach, and hybrid execution and scheduling runtime, all of which significantly increase the performance of our SVD solver. Our results demonstrate a many-fold performance increase compared to currently available software. In particular, our software is two times faster than Intel's Math Kernel Library (MKL), a highly optimized implementation from the hardware vendor, when all the singular vectors are requested; it achieves a 5-fold speed-up when only 20 of the vectors are computed; and it is up to 10 times faster if only the singular values are required."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | Tiled Algorithms for the SVD. In the context of massive parallelism, and of reducing data movement, many dense linear algebra algorithms have been moved to so-called tiled algorithms''. In the tiled algorithm framework, algorithms operates on of the matrix, and tasks are scheduled thanks to a runtime. In the context of the SVD, tiled algorithms naturally leads to band bidiagonal form. In 2010, Ltaief, Kurzak and Dongarra @cite_30 present a tiled algorithm for GE2BND (to go from full to band bidiagonal form). In 2013 (technical report in 2011), Ltaief, Luszczek, Dongarra @cite_22 add the second step (BND2BD) and present a tiled algorithm for GE2VAL using GE2BND+BND2BD+BD2VAL. In 2012, Ltaief, Luszczek, and Dongarra @cite_36 improve the algorithm for tall and skinny matrices by using any'' tree instead of flat trees in the QR steps. In 2012, Haidar, Ltaief, Luszczek and Dongarra @cite_17 improve the BND2BD step of @cite_22 . Finally, in 2013, Haidar, Kurzak, and Luszczek @cite_25 consider the problem of computing singular vectors (GESVD) by performing GE2BND+BND2BD+BD2VAL+VAL2BD+BD2BND+BND2GE . They show that the two-step approach (from full to band, then band to bidiagonal) can be successfully used not only for computing singular values, but also for computing singular vectors. | {
"cite_N": [
"@cite_30",
"@cite_22",
"@cite_36",
"@cite_25",
"@cite_17"
],
"mid": [
"2107410315",
"2057204994",
"1581582143",
"2156779053",
"2107857731"
],
"abstract": [
"The objective of this paper is to extend, in the context of multicore architectures, the concepts of tile algorithms [, 2007] for Cholesky, LU, and QR factorizations to the family of two-sided factorizations. In particular, the bidiagonal reduction of a general, dense matrix is very often used as a preprocessing step for calculating the Singular Value Decomposition. Furthermore, in the Top500 list of June 2008, 98 percent of the fastest parallel systems in the world were based on multicores. This confronts the scientific software community with both a daunting challenge and a unique opportunity. The challenge arises from the disturbing mismatch between the design of systems based on this new chip architecture-hundreds of thousands of nodes, a million or more cores, reduced bandwidth and memory available to cores-and the components of the traditional software stack, such as numerical libraries, on which scientific applications have relied for their accuracy and performance. The many-core trend has even more exacerbated the problem, and it becomes critical to efficiently integrate existing or new numerical linear algebra algorithms suitable for such hardware. By exploiting the concept of tile algorithms in the multicore environment (i.e., high level of parallelism with fine granularity and high-performance data representation combined with a dynamic data-driven execution), the band bidiagonal reduction presented here achieves 94 Gflop s on a 12,000 × 12,000 matrix with 16 Intel Tigerton 2.4 GHz processors. The main drawback of the tile algorithms approach for the bidiagonal reduction is that the full reduction cannot be obtained in one stage. Other methods have to be considered to further reduce the band matrix to the required form.",
"This article presents a new high-performance bidiagonal reduction (BRD) for homogeneous multicore architectures. This article is an extension of the high-performance tridiagonal reduction implemented by the same authors [, IPDPS 2011] to the BRD case. The BRD is the first step toward computing the singular value decomposition of a matrix, which is one of the most important algorithms in numerical linear algebra due to its broad impact in computational science. The high performance of the BRD described in this article comes from the combination of four important features: (1) tile algorithms with tile data layout, which provide an efficient data representation in main memory; (2) a two-stage reduction approach that allows to cast most of the computation during the first stage (reduction to band form) into calls to Level 3 BLAS and reduces the memory traffic during the second stage (reduction from band to bidiagonal form) by using high-performance kernels optimized for cache reuse; (3) a data dependence translation layer that maps the general algorithm with column-major data layout into the tile data layout; and (4) a dynamic runtime system that efficiently schedules the newly implemented kernels across the processing units and ensures that the data dependencies are not violated. A detailed analysis is provided to understand the critical impact of the tile size on the total execution time, which also corresponds to the matrix bandwidth size after the reduction of the first stage. The performance results show a significant improvement over currently established alternatives. The new high-performance BRD achieves up to a 30-fold speedup on a 16-core Intel Xeon machine with a 12000× 12000 matrix size against the state-of-the-art open source and commercial numerical software packages, namely LAPACK, compiled with optimized and multithreaded BLAS from MKL as well as Intel MKL version 10.2.",
"The objective of this paper is to enhance the parallelism of the tile bidiagonal transformation using tree reduction on multicore architectures. First introduced by Ltaief et. al [LAPACK Working Note #247, 2011], the bidiagonal transformation using tile algorithms with a two-stage approach has shown very promising results on square matrices. However, for tall and skinny matrices, the inherent problem of processing the panel in a domino-like fashion generates unnecessary sequential tasks. By using tree reduction, the panel is horizontally split, which creates another dimension of parallelism and engenders many concurrent tasks to be dynamically scheduled on the available cores. The results reported in this paper are very encouraging. The new tile bidiagonal transformation, targeting tall and skinny matrices, outperforms the state-of-the-art numerical linear algebra libraries LAPACK V3.2 and Intel MKL ver. 10.3 by up to 29-fold speedup and the standard two-stage PLASMA BRD by up to 20-fold speedup, on an eight socket hexa-core AMD Opteron multicore shared-memory system.",
"The enormous gap between the high-performance capabilities of today's CPUs and off-chip communication poses extreme challenges to the development of numerical software that is scalable and achieves high performance. In this article, we describe a successful methodology to address these challenges---starting with our algorithm design, through kernel optimization and tuning, and finishing with our programming model. All these lead to development of a scalable high-performance Singular Value Decomposition (SVD) solver. We developed a set of highly optimized kernels and combined them with advanced optimization techniques that feature fine-grain and cache-contained kernels, a task based approach, and hybrid execution and scheduling runtime, all of which significantly increase the performance of our SVD solver. Our results demonstrate a many-fold performance increase compared to currently available software. In particular, our software is two times faster than Intel's Math Kernel Library (MKL), a highly optimized implementation from the hardware vendor, when all the singular vectors are requested; it achieves a 5-fold speed-up when only 20 of the vectors are computed; and it is up to 10 times faster if only the singular values are required.",
"We present new high performance numerical kernels combined with advanced optimization techniques that significantly increase the performance of parallel bidiagonal reduction. Our approach is based on developing efficient fine-grained computational tasks as well as reducing overheads associated with their high-level scheduling during the so-called bulge chasing procedure that is an essential phase of a scalable bidiagonalization procedure. In essence, we coalesce multiple tasks in a way that reduces the time needed to switch execution context between the scheduler and useful computational tasks. At the same time, we maintain the crucial information about the tasks and their data dependencies between the coalescing groups. This is the necessary condition to preserve numerical correctness of the computation. We show our annihilation strategy based on multiple applications of single orthogonal reflectors. Despite non-trivial characteristics in computational complexity and memory access patterns, our optimization approach smoothly applies to the annihilation scenario. The coalescing positively influences another equally important aspect of the bulge chasing stage: the memory reuse. For the tasks within the coalescing groups, the data is retained in high levels of the cache hierarchy and, as a consequence, operations that are normally memory-bound increase their ratio of computation to off-chip communication and become compute-bound which renders them amenable to efficient execution on multicore architectures. The performance for the new two-stage bidiagonal reduction is staggering. Our implementation results in up to 50-fold and 12-fold improvement ( 130 Gflop s) compared to the equivalent routines from LAPACK V3.2 and Intel MKL V10.3, respectively, on an eight socket hexa-core AMD Opteron multicore shared-memory system with a matrix size of 24000 × 24000. Last but not least, we provide a comprehensive study on the impact of the coalescing group size in terms of cache utilization and power consumption in the context of this new two-stage bidiagonal reduction."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | BND2BD step. The algorithm in for BND2BD is xGBBRD. In 1996, Lang @cite_33 improved the sequential version of the algorithm and developed a parallel distributed algorithm. Recently, PLASMA released an efficient multi-threaded implementation @cite_17 @cite_22 . We also note that Rajamanickam @cite_15 recently worked on this step. | {
"cite_N": [
"@cite_22",
"@cite_15",
"@cite_33",
"@cite_17"
],
"mid": [
"2057204994",
"",
"2053869601",
"2107857731"
],
"abstract": [
"This article presents a new high-performance bidiagonal reduction (BRD) for homogeneous multicore architectures. This article is an extension of the high-performance tridiagonal reduction implemented by the same authors [, IPDPS 2011] to the BRD case. The BRD is the first step toward computing the singular value decomposition of a matrix, which is one of the most important algorithms in numerical linear algebra due to its broad impact in computational science. The high performance of the BRD described in this article comes from the combination of four important features: (1) tile algorithms with tile data layout, which provide an efficient data representation in main memory; (2) a two-stage reduction approach that allows to cast most of the computation during the first stage (reduction to band form) into calls to Level 3 BLAS and reduces the memory traffic during the second stage (reduction from band to bidiagonal form) by using high-performance kernels optimized for cache reuse; (3) a data dependence translation layer that maps the general algorithm with column-major data layout into the tile data layout; and (4) a dynamic runtime system that efficiently schedules the newly implemented kernels across the processing units and ensures that the data dependencies are not violated. A detailed analysis is provided to understand the critical impact of the tile size on the total execution time, which also corresponds to the matrix bandwidth size after the reduction of the first stage. The performance results show a significant improvement over currently established alternatives. The new high-performance BRD achieves up to a 30-fold speedup on a 16-core Intel Xeon machine with a 12000× 12000 matrix size against the state-of-the-art open source and commercial numerical software packages, namely LAPACK, compiled with optimized and multithreaded BLAS from MKL as well as Intel MKL version 10.2.",
"",
"A parallel algorithm for reducing banded matrices to bidiagonal form is presented. In contrast to the rotation-based Â?standard approachÂ?, our algorithm is based on Householder transforms, therefore exhibiting considerably higher data locality (BLAS level 2 instead of level 1). The update of the transformation matrices which involves the vast majority of the operations can even be blocked to allow the use of level 3 BLAS. Thus, our algorithm will outperform the standard method on a serial computer with a distinct memory hierarchy. In addition, the algorithm can be efficiently implemented in a distributed memory environment, as is demonstrated by numerical results on the Intel Paragon.",
"We present new high performance numerical kernels combined with advanced optimization techniques that significantly increase the performance of parallel bidiagonal reduction. Our approach is based on developing efficient fine-grained computational tasks as well as reducing overheads associated with their high-level scheduling during the so-called bulge chasing procedure that is an essential phase of a scalable bidiagonalization procedure. In essence, we coalesce multiple tasks in a way that reduces the time needed to switch execution context between the scheduler and useful computational tasks. At the same time, we maintain the crucial information about the tasks and their data dependencies between the coalescing groups. This is the necessary condition to preserve numerical correctness of the computation. We show our annihilation strategy based on multiple applications of single orthogonal reflectors. Despite non-trivial characteristics in computational complexity and memory access patterns, our optimization approach smoothly applies to the annihilation scenario. The coalescing positively influences another equally important aspect of the bulge chasing stage: the memory reuse. For the tasks within the coalescing groups, the data is retained in high levels of the cache hierarchy and, as a consequence, operations that are normally memory-bound increase their ratio of computation to off-chip communication and become compute-bound which renders them amenable to efficient execution on multicore architectures. The performance for the new two-stage bidiagonal reduction is staggering. Our implementation results in up to 50-fold and 12-fold improvement ( 130 Gflop s) compared to the equivalent routines from LAPACK V3.2 and Intel MKL V10.3, respectively, on an eight socket hexa-core AMD Opteron multicore shared-memory system with a matrix size of 24000 × 24000. Last but not least, we provide a comprehensive study on the impact of the coalescing group size in terms of cache utilization and power consumption in the context of this new two-stage bidiagonal reduction."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | BD2VAL step. Much research has been done and is done on this kernel. Much software exists. In , to compute the singular values and optionally the singular vectors of a bidiagonal matrix, the routine xBDSQR uses the Golub-Kahan QR algorithm @cite_1 ; the routine xBDSDC uses the divide-and-conquer algorithm @cite_8 ; and the routine xBDSVX uses bisection and inverse iteration algorithm. Recent research was trying to apply the MRRR (Multiple Relatively Robust Representations) method @cite_18 to the problem. | {
"cite_N": [
"@cite_18",
"@cite_1",
"@cite_8"
],
"mid": [
"2058443643",
"2005423095",
"2169378749"
],
"abstract": [
"We describe the design and implementation of a new algorithm for computing the singular value decomposition (SVD) of a real bidiagonal matrix. This algorithm uses ideas developed by Grosser and Lang that extend Parlett’s and Dhillon’s multiple relatively robust representations (MRRR) algorithm for the tridiagonal symmetric eigenproblem. One key feature of our new implementation is that @math singular triplets can be computed using only @math storage units and floating point operations, where @math is the dimension of the matrix. The algorithm will be made available as routine xBDSCR in the upcoming new release of the LAPACK library.",
"A numerically stable and fairly fast scheme is described to compute the unitary matrices U and V which transform a given matrix A into a diagonal form @math , thus exhibiting A’s singular values on @math ’s diagonal. The scheme first transforms A to a bidiagonal matrix J, then diagonalizes J. The scheme described here is complicated but does not suffer from the computational difficulties which occasionally afflict some previously known methods. Some applications are mentioned, in particular the use of the pseudo-inverse @math to solve least squares problems in a way which dampens spurious oscillation and cancellation.",
"The authors present a stable and efficient divide-and-conquer algorithm for computing the singular value decomposition (SVD) of a lower bidiagonal matrix. Previous divide-and-conquer algorithms all suffer from a potential loss of orthogonality among the computed singular vectors unless extended precision arithmetic is used. A generalization that computes the SVD of a lower banded matrix is also presented."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | BND2BD+BD2VAL steps in this paper. This paper does not focus neither on BND2BD nor BD2VAL. As far as we are concerned, we can use any of the methods mentioned above. The faster these two steps are, the better for us. For this study, during the experimental section, for BND2BD, we use the PLASMA multi-threaded implementation @cite_17 @cite_22 and, for BD2VAL, we use LAPACK xBDSQR . | {
"cite_N": [
"@cite_22",
"@cite_17"
],
"mid": [
"2057204994",
"2107857731"
],
"abstract": [
"This article presents a new high-performance bidiagonal reduction (BRD) for homogeneous multicore architectures. This article is an extension of the high-performance tridiagonal reduction implemented by the same authors [, IPDPS 2011] to the BRD case. The BRD is the first step toward computing the singular value decomposition of a matrix, which is one of the most important algorithms in numerical linear algebra due to its broad impact in computational science. The high performance of the BRD described in this article comes from the combination of four important features: (1) tile algorithms with tile data layout, which provide an efficient data representation in main memory; (2) a two-stage reduction approach that allows to cast most of the computation during the first stage (reduction to band form) into calls to Level 3 BLAS and reduces the memory traffic during the second stage (reduction from band to bidiagonal form) by using high-performance kernels optimized for cache reuse; (3) a data dependence translation layer that maps the general algorithm with column-major data layout into the tile data layout; and (4) a dynamic runtime system that efficiently schedules the newly implemented kernels across the processing units and ensures that the data dependencies are not violated. A detailed analysis is provided to understand the critical impact of the tile size on the total execution time, which also corresponds to the matrix bandwidth size after the reduction of the first stage. The performance results show a significant improvement over currently established alternatives. The new high-performance BRD achieves up to a 30-fold speedup on a 16-core Intel Xeon machine with a 12000× 12000 matrix size against the state-of-the-art open source and commercial numerical software packages, namely LAPACK, compiled with optimized and multithreaded BLAS from MKL as well as Intel MKL version 10.2.",
"We present new high performance numerical kernels combined with advanced optimization techniques that significantly increase the performance of parallel bidiagonal reduction. Our approach is based on developing efficient fine-grained computational tasks as well as reducing overheads associated with their high-level scheduling during the so-called bulge chasing procedure that is an essential phase of a scalable bidiagonalization procedure. In essence, we coalesce multiple tasks in a way that reduces the time needed to switch execution context between the scheduler and useful computational tasks. At the same time, we maintain the crucial information about the tasks and their data dependencies between the coalescing groups. This is the necessary condition to preserve numerical correctness of the computation. We show our annihilation strategy based on multiple applications of single orthogonal reflectors. Despite non-trivial characteristics in computational complexity and memory access patterns, our optimization approach smoothly applies to the annihilation scenario. The coalescing positively influences another equally important aspect of the bulge chasing stage: the memory reuse. For the tasks within the coalescing groups, the data is retained in high levels of the cache hierarchy and, as a consequence, operations that are normally memory-bound increase their ratio of computation to off-chip communication and become compute-bound which renders them amenable to efficient execution on multicore architectures. The performance for the new two-stage bidiagonal reduction is staggering. Our implementation results in up to 50-fold and 12-fold improvement ( 130 Gflop s) compared to the equivalent routines from LAPACK V3.2 and Intel MKL V10.3, respectively, on an eight socket hexa-core AMD Opteron multicore shared-memory system with a matrix size of 24000 × 24000. Last but not least, we provide a comprehensive study on the impact of the coalescing group size in terms of cache utilization and power consumption in the context of this new two-stage bidiagonal reduction."
]
} |
1611.06892 | 2539842690 | We consider algorithms for going from a full'' matrix to a condensed band bidiagonal'' form using orthogonal transformations. We use the framework of algorithms by tiles''. Within this framework, we study: (i) the tiled bidiagonalization algorithm , which is a tiled version of the standard scalar bidiagonalization algorithm; and (ii) the R-bidiagonalization algorithm , which is a tiled version of the algorithm which consists in first performing the QR factorization of the initial matrix, then performing the band-bidiagonalization of the R-factor. For both bidiagonalization algorithms and , we use four main types of reduction trees, namely , , , and a newly introduced auto-adaptive tree, . We provide a study of critical path lengths for these tiled algorithms, which shows that (i) has a shorter critical path length than for tall and skinny matrices, and (ii) based schemes are much better than earlier proposed variants with unbounded resources. We provide experiments on a single multicore node, and on a few multicore nodes of a parallel distributed shared-memory system, to show the superiority of the new algorithms on a variety of matrix sizes, matrix shapes and core counts. | Approach to compute the SVD without going by bidiagonal form. We note that going to bidiagonal form is not a necessary step to compute the singular value decomposition of a matrix. There are approaches to compute the SVD of a matrix which avoid the reduction to bidiagonal form. In the direct method'' category, we can quote Jacobi SVD @cite_2 @cite_34 , QDWHSVD @cite_35 ; one can also think to think the singular value decomposition directly from the block bidiagonal form although not much work has been done in this direction so far. Also, we can also refer to all iterative methods'' for computing the singular values. The goal of these iterative methods is in general to compute a few singular triplets. As far as software packages, we can quote SVDPACK @cite_9 , PROPACK @cite_23 , and PRIMME SVD @cite_11 . As far methods, we can quote @cite_19 @cite_32 @cite_12 . | {
"cite_N": [
"@cite_35",
"@cite_9",
"@cite_32",
"@cite_19",
"@cite_23",
"@cite_2",
"@cite_34",
"@cite_12",
"@cite_11"
],
"mid": [
"2059076129",
"1918870341",
"2021710646",
"1747539222",
"2154202969",
"2020828929",
"",
"1965497533",
"2462387008"
],
"abstract": [
"Spectral divide and conquer algorithms solve the eigenvalue problem for all the eigenvalues and eigenvectors by recursively computing an invariant subspace for a subset of the spectrum and using it to decouple the problem into two smaller subproblems. A number of such algorithms have been developed over the last 40 years, often motivated by parallel computing and, most recently, with the aim of achieving minimal communication costs. However, none of the existing algorithms has been proved to be backward stable, and they all have a significantly higher arithmetic cost than the standard algorithms currently used. We present new spectral divide and conquer algorithms for the symmetric eigenvalue problem and the singular value decomposition that are backward stable, achieve lower bounds on communication costs recently derived by Ballard, Demmel, Holtz, and Schwartz, and have operation counts within a small constant factor of those for the standard algorithms. The new algorithms are built on the polar decompos...",
"We present four numerical methods for computing the singular value decomposition SVD of large sparse matrices on a multiprocessor architecture. We emphasize Lanczos and subspace iteration-based methods for determining several of the largest singular triplets singular values and corresponding left- and right-singular vectors for sparse matrices arising from two practical applications: information retrieval and seismic reflection tomography. The target architectures for our implementations are the CRAY-2S 4-128 and Alliant FX 80. The sparse SVD problem is well motivated by recent information-retrieval techniques in which dominant singular values and their corresponding singular vectors of large sparse term-document matrices are desired, and by nonlinear inverse problems from seismic tomography applications which require approximate pseudo-inverses of large sparse Jacobian matrices. This research may help advance the development of future out-of-core sparse SVD methods, which can be used, for example, to handle extremely large sparse matrices 0 ? 106 rows or columns associated with extremely large databases in query-based information-retrieval applications.",
"We discuss a new method for the iterative computation of a portion of the singular values and vectors of a large sparse matrix. Similar to the Jacobi--Davidson method for the eigenvalue problem, we compute in each step a correction by (approximately) solving a correction equation. We give a few variants of this Jacobi--Davidson SVD (JDSVD) method with their theoretical properties. It is shown that the JDSVD can be seen as an accelerated (inexact) Newton scheme. We experimentally compare the method with some other iterative SVD methods.",
"The computation of a few singular triplets of large, sparse matrices is a challenging task, especially when the smallest magnitude singular values are needed in high accuracy. Most recent efforts try to address this problem through variations of the Lanczos bidiagonalization method, but they are still challenged even for medium matrix sizes due to the difficulty of the problem. We propose a novel SVD approach that can take advantage of preconditioning and of any well-designed eigensolver to compute both largest and smallest singular triplets. Accuracy and efficiency is achieved through a hybrid, two-stage meta-method, PHSVDS. In the first stage, PHSVDS solves the normal equations up to the best achievable accuracy. If further accuracy is required, the method switches automatically to an eigenvalue problem with the augmented matrix. Thus it combines the advantages of the two stages, faster convergence and accuracy, respectively. For the augmented matrix, solving the interior eigenvalue is facilitated by pr...",
"A partial reorthogonalization procedure (BPRO) for maintaining semi-orthogonality among the left and right Lanczos vectors in the Lanczos bidiagonalization (LBD) is presented. The resulting algorithm is mathematically equivalent to the symmetric Lanczos algorithm with partial reorthogonalization (PRO) developed by Simon but works directly on the Lanczos bidiagonalization of A. For computing the singular values and vectors of a large sparse matrix with high accuracy, the BPRO algorithm uses only half the amount of storage and a factor of 3-4 less work compared to methods based on PRO applied to an equivalent symmetric system. Like PRO the algorithm presented here is based on simple recurrences which enable it to monitor the loss of orthogonality among the Lanczos vectors directly without forming inner products. These recurrences are used to develop a Lanczos bidiagonalization algorithm with partial reorthogonalization which has been implemented in a MATLAB package for sparse SVD and eigenvalue problems called PROPACK. Numerical experiments with the routines from PROPACK are conducted using a test problem from inverse helioseismology to illustrate the properties of the method. In addition a number of test matrices from the Harwell-Boeing collection are used to compare the accuracy and efficiency of the MATLAB implementations of BPRO and PRO with the svds routine in MATLAB 5.1, which uses an implicitly restarted Lanczos algorithm.",
"This paper is the result of concerted efforts to break the barrier between numerical accuracy and run-time efficiency in computing the fundamental decomposition of numerical linear algebra—the singular value decomposition (SVD) of general dense matrices. It is an unfortunate fact that the numerically most accurate one-sided Jacobi SVD algorithm is several times slower than generally less accurate bidiagonalization-based methods such as the QR or the divide-and-conquer algorithm. Our quest for a highly accurate and efficient SVD algorithm has led us to a new, superior variant of the Jacobi algorithm. The new algorithm has inherited all good high accuracy properties of the Jacobi algorithm, and it can outperform the QR algorithm.",
"",
"The bidiagonalization Lanczos method can be used for computing a few of the largest or smallest singular values and corresponding singular vectors of a large matrix, but the method may encounter some convergence problems. In this paper the convergence of the method is analyzed, showing why it may converge erratically and perhaps fail to converge. To correct this possible nonconvergence and improve the method, a refined bidiagonalization Lanczos method is proposed. The implicitly restarting technique due to Sorensen is applied to the method, and an implicitly restarted refined bidiagonalization Lanczos algorithm (IRRBL) is developed. A new selection of shifts is proposed for use within IRRBL, called refined shifts, and a reliable and efficient algorithm is developed for computing the refined shifts. Numerical experiments show that IRRBL can perform better than the implicitly restarted bidiagonalization Lanczos algorithm (IRBL) proposed by Larsen, in particular when the smallest singular triplets are desired.",
"The increasing number of applications requiring the solution of large scale singular value problems have rekindled interest in iterative methods for the SVD. Some promising recent ad- vances in large scale iterative methods are still plagued by slow convergence and accuracy limitations for computing smallest singular triplets. Furthermore, their current implementations in MATLAB cannot address the required large problems. Recently, we presented a preconditioned, two-stage method to effectively and accurately compute a small number of extreme singular triplets. In this research, we present a high-performance software, PRIMME SVDS, that implements our hybrid method based on the state-of-the-art eigensolver package PRIMME for both largest and smallest singular values. PRIMME SVDS fills a gap in production level software for computing the partial SVD, especially with preconditioning. The numerical experiments demonstrate its superior performance compared to other state-of-the-art software and its good parallel performance under strong and weak scaling."
]
} |
1611.06656 | 2949455921 | Deep residual networks have recently emerged as the state-of-the-art architecture in image segmentation and object detection. In this paper, we propose new image features (called ResFeats) extracted from the last convolutional layer of deep residual networks pre-trained on ImageNet. We propose to use ResFeats for diverse image classification tasks namely, object classification, scene classification and coral classification and show that ResFeats consistently perform better than their CNN counterparts on these classification tasks. Since the ResFeats are large feature vectors, we propose to use PCA for dimensionality reduction. Experimental results are provided to show the effectiveness of ResFeats with state-of-the-art classification accuracies on Caltech-101, Caltech-256 and MLC datasets and a significant performance improvement on MIT-67 dataset compared to the widely used CNN features. | Chatfield al @cite_25 evaluated the performance of CNN based methods for image classification and compared their methods with previous feature encoding methods. Their findings established that deeper CNN performed better than the shallower models of the same network trained on augmented data. VGGnet @cite_15 was used as the source CNN in their work. They improved the classification accuracies of popular datasets such as VOC, Caltech-101 and Caltech-256. He al @cite_26 used spatial pyramid pooling of CNN features to further improve the classification accuracy on the Caltech datasets and reported state-of-the-art object classification results. | {
"cite_N": [
"@cite_15",
"@cite_26",
"@cite_25"
],
"mid": [
"1686810756",
"2179352600",
"1994002998"
],
"abstract": [
"In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.",
"Existing deep convolutional neural networks (CNNs) require a fixed-size (e.g. 224×224) input image. This requirement is “artificial” and may hurt the recognition accuracy for the images or sub-images of an arbitrary size scale. In this work, we equip the networks with a more principled pooling strategy, “spatial pyramid pooling”, to eliminate the above requirement. The new network structure, called SPP-net, can generate a fixed-length representation regardless of image size scale. By removing the fixed-size limitation, we can improve all CNN-based image classification methods in general. Our SPP-net achieves state-of-the-art accuracy on the datasets of ImageNet 2012, Pascal VOC 2007, and Caltech101.",
"The latest generation of Convolutional Neural Networks (CNN) have achieved impressive results in challenging benchmarks on image recognition and object detection, significantly raising the interest of the community in these methods. Nevertheless, it is still unclear how different CNN methods compare with each other and with previous state-of-the-art shallow representations such as the Bag-of-Visual-Words and the Improved Fisher Vector. This paper conducts a rigorous evaluation of these new techniques, exploring different deep architectures and comparing them on a common ground, identifying and disclosing important implementation details. We identify several useful properties of CNN-based representations, including the fact that the dimensionality of the CNN output layer can be reduced significantly without having an adverse effect on performance. We also identify aspects of deep and shallow methods that can be successfully shared. In particular, we show that the data augmentation techniques commonly applied to CNN-based methods can also be applied to shallow methods, and result in an analogous performance boost. Source code and models to reproduce the experiments in the paper is made publicly available."
]
} |
1611.06689 | 2552317450 | Recently, the popularity of depth-sensors such as Kinect has made depth videos easily available while its advantages have not been fully exploited. This paper investigates, for gesture recognition, to explore the spatial and temporal information complementarily embedded in RGB and depth sequences. We propose a convolutional twostream consensus voting network (2SCVN) which explicitly models both the short-term and long-term structure of the RGB sequences. To alleviate distractions from background, a 3d depth-saliency ConvNet stream (3DDSN) is aggregated in parallel to identify subtle motion characteristics. These two components in an unified framework significantly improve the recognition accuracy. On the challenging Chalearn IsoGD benchmark, our proposed method outperforms the first place on the leader-board by a large margin (10.29 ) while also achieving the best result on RGBD-HuDaAct dataset (96.74 ). Both quantitative experiments and qualitative analysis shows the effectiveness of our proposed framework and codes will be released to facilitate future research. | Many hand-crafted features have been proposed for video analysis in the area of gesture recognition @cite_38 @cite_29 @cite_41 @cite_37 @cite_6 . For example, @cite_28 extracted a novel spatiotemporal feature named MFSK while @cite_17 proposed to calculate SIFT-like descriptors on 3D gradient and motion spaces respectively for RGB-D video recognition. @cite_10 recognized hand gestures via bag-of-words vector mapped from extracted keypoints using SIFT and a multiclass SVM was trained as gesture classifier. | {
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_41",
"@cite_28",
"@cite_29",
"@cite_6",
"@cite_10",
"@cite_17"
],
"mid": [
"2146221819",
"2008110769",
"",
"",
"",
"2112480082",
"2119656522",
"2042713909"
],
"abstract": [
"We present two real-time hidden Markov model-based systems for recognizing sentence-level continuous American sign language (ASL) using a single camera to track the user's unadorned hands. The first system observes the user from a desk mounted camera and achieves 92 percent word accuracy. The second system mounts the camera in a cap worn by the user and achieves 98 percent accuracy (97 percent with an unrestricted grammar). Both experiments use a 40-word lexicon.",
"In this paper we propose a hand gesture recognition system that relies on color and depth images, and on a small pose sensor on the human palm. Monocular and stereo vision systems have been used for human pose and gesture recognition, but with limited scope due to limitations on texture, illumination, etc. New RGB-Depth sensors, that reply on projected light such as the Microsoft Kinect, have overcome many of those limitations. However, the point clouds for hand gestures are still in many cases noisy and partially occluded, and hand gesture recognition is not trivial. Hand gesture recognition is much more complex than full body motion, since we can have the hands in any orientation and can not assume a standing body on a ground plane. In this work we propose to add a tiny pose sensor to the human palm, with a minute accelerometer and magnetometer that combined provide 3D angular pose, to reduce the search space and have a robust and computationally light recognition method. Starting with the full depth image point cloud, segmentation can be performed by taking into account the relative depth and hand orientation, as well as skin color. Identification is then performed by matching 3D voxel occupancy against a gesture template database. Preliminary results are presented for the recognition of Portuguese Sign Language alphabet, showing the validity of the approach.",
"",
"",
"",
"For one-shot learning gesture recognition, two important challenges are: how to extract distinctive features and how to learn a discriminative model from only one training sample per gesture class. For feature extraction, a new spatio-temporal feature representation called 3D enhanced motion scale-invariant feature transform (3D EMoSIFT) is proposed, which fuses RGB-D data. Compared with other features, the new feature set is invariant to scale and rotation, and has more compact and richer visual representations. For learning a discriminative model, all features extracted from training samples are clustered with the k-means algorithm to learn a visual codebook. Then, unlike the traditional bag of feature (BoF) models using vector quantization (VQ) to map each feature into a certain visual codeword, a sparse coding method named simulation orthogonal matching pursuit (SOMP) is applied and thus each feature can be represented by some linear combination of a small number of codewords. Compared with VQ, SOMP leads to a much lower reconstruction error and achieves better performance. The proposed approach has been evaluated on ChaLearn gesture database and the result has been ranked amongst the top best performing techniques on ChaLearn gesture challenge (round 2).",
"This paper presents a novel and real-time system for interaction with an application or video game via hand gestures. Our system includes detecting and tracking bare hand in cluttered background using skin detection and hand posture contour comparison algorithm after face subtraction, recognizing hand gestures via bag-of-features and multiclass support vector machine (SVM) and building a grammar that generates gesture commands to control an application. In the training stage, after extracting the keypoints for every training image using the scale invariance feature transform (SIFT), a vector quantization technique will map keypoints from every training image into a unified dimensional histogram vector (bag-of-words) after K-means clustering. This histogram is treated as an input vector for a multiclass SVM to build the training classifier. In the testing stage, for every frame captured from a webcam, the hand is detected using our algorithm, then, the keypoints are extracted for every small image that contains the detected hand gesture only and fed into the cluster model to map them into a bag-of-words vector, which is finally fed into the multiclass SVM training classifier to recognize the hand gesture.",
"Human activity recognition based on RGB-D data has received more attention in recent years. We propose a spatiotemporal feature named three-dimensional (3D) sparse motion scale-invariant feature transform (SIFT) from RGB-D data for activity recognition. First, we build pyramids as scale space for each RGB and depth frame, and then use Shi-Tomasi corner detector and sparse optical flow to quickly detect and track robust keypoints around the motion pattern in the scale space. Subsequently, local patches around keypoints, which are extracted from RGB-D data, are used to build 3D gradient and motion spaces. Then SIFT-like descriptors are calculated on both 3D spaces, respectively. The proposed feature is invariant to scale, transition, and partial occlusions. More importantly, the running time of the proposed feature is fast so that it is well-suited for real-time applications. We have evaluated the proposed feature under a bag of words model on three public RGB-D datasets: one-shot learning Chalearn Gesture Dataset, Cornell Activity Dataset-60, and MSR Daily Activity 3D dataset. Experimental results show that the proposed feature outperforms other spatiotemporal features and are comparative to other state-of-the-art approaches, even though there is only one training sample for each class."
]
} |
1611.06689 | 2552317450 | Recently, the popularity of depth-sensors such as Kinect has made depth videos easily available while its advantages have not been fully exploited. This paper investigates, for gesture recognition, to explore the spatial and temporal information complementarily embedded in RGB and depth sequences. We propose a convolutional twostream consensus voting network (2SCVN) which explicitly models both the short-term and long-term structure of the RGB sequences. To alleviate distractions from background, a 3d depth-saliency ConvNet stream (3DDSN) is aggregated in parallel to identify subtle motion characteristics. These two components in an unified framework significantly improve the recognition accuracy. On the challenging Chalearn IsoGD benchmark, our proposed method outperforms the first place on the leader-board by a large margin (10.29 ) while also achieving the best result on RGBD-HuDaAct dataset (96.74 ). Both quantitative experiments and qualitative analysis shows the effectiveness of our proposed framework and codes will be released to facilitate future research. | Recently, the convolutinal neural networks @cite_4 have been introduced to the field of gesture and action recognition due to its rich capacity for representation @cite_14 @cite_1 @cite_39 . Additionally, the rapid emergence of depth-sensor has made it economically feasible to capture both color and depth videos, providing motion information as well as three-dimensional structual information. This significantly reduces motion ambiguity when projecting the three-dimensional motion onto the two-dimensional image plane @cite_11 @cite_26 @cite_13 . For example, @cite_14 proposes to use depth and intensity data with 3D convolutional networks for gesture recognition. @cite_1 proposes a multi-stream recurrent neural network that can be trained end to end without domain-specific hand engineering while @cite_39 combines 3DCNN with RNN for online gesture detection and classification. Ohn- @cite_43 first detects a hand in the region of interaction and then combines RGB and depth descriptor for classification. @cite_7 proposes a multi-modal architecture that operates at 3 temporal scales corresponding to dynamic poses for gesture localization. | {
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_26",
"@cite_7",
"@cite_1",
"@cite_39",
"@cite_43",
"@cite_13",
"@cite_11"
],
"mid": [
"1943054406",
"2310919327",
"2342311830",
"",
"2405546895",
"",
"1993666393",
"2080107612",
"1989665047"
],
"abstract": [
"Touchless hand gesture recognition systems are becoming important in automotive user interfaces as they improve safety and comfort. Various computer vision algorithms have employed color and depth cameras for hand gesture recognition, but robust classification of gestures from different subjects performed under widely varying lighting conditions is still challenging. We propose an algorithm for drivers' hand gesture recognition from challenging depth and intensity data using 3D convolutional neural networks. Our solution combines information from multiple spatial scales for the final prediction. It also employs spatio-temporal data augmentation for more effective training and to reduce potential overfitting. Our method achieves a correct classification rate of 77.5 on the VIVA challenge dataset.",
"",
"Recent approaches in depth-based human activity analysis achieved outstanding performance and proved the effectiveness of 3D representation for classification of action classes. Currently available depth-based and RGB+D-based action recognition benchmarks have a number of limitations, including the lack of training samples, distinct class labels, camera views and variety of subjects. In this paper we introduce a large-scale dataset for RGB+D human action recognition with more than 56 thousand video samples and 4 million frames, collected from 40 distinct subjects. Our dataset contains 60 different action classes including daily, mutual, and health-related actions. In addition, we propose a new recurrent neural network structure to model the long-term temporal correlation of the features for each body part, and utilize them for better action classification. Experimental results show the advantages of applying deep learning methods over state-of-the-art hand-crafted features on the suggested cross-subject and cross-view evaluation criteria for our dataset. The introduction of this large scale dataset will enable the community to apply, develop and adapt various data-hungry learning techniques for the task of depth-based and RGB+D-based human activity analysis.",
"",
"In this paper, we present a novel method for multimodal gesture recognition based on neural networks. Our multi-stream recurrent neural network MRNN is a completely data-driven model that can be trained from end to end without domain-specific hand engineering. The MRNN extends recurrent neural networks with Long Short-Term Memory cells LSTM-RNNs that facilitate the handling of variable-length gestures. We propose a recurrent approach for fusing multiple temporal modalities using multiple streams of LSTM-RNNs. In addition, we propose alternative fusion architectures and empirically evaluate the performance and robustness of these fusion strategies. Experimental results demonstrate that the proposed MRNN outperforms other state-of-the-art methods in the Sheffield Kinect Gesture SKIG dataset, and has significantly high robustness to noisy inputs.",
"",
"In this paper, we develop a vision-based system that employs a combined RGB and depth descriptor to classify hand gestures. The method is studied for a human-machine interface application in the car. Two interconnected modules are employed: one that detects a hand in the region of interaction and performs user classification, and another that performs gesture recognition. The feasibility of the system is demonstrated using a challenging RGBD hand gesture data set collected under settings of common illumination variation and occlusion.",
"In this paper, we propose a novel approach called class-specific maximization of mutual information (CSMMI) using a submodular method, which aims at learning a compact and discriminative dictionary for each class. Unlike traditional dictionary-based algorithms, which typically learn a shared dictionary for all of the classes, we unify the intraclass and interclass mutual information (MI) into an single objective function to optimize class-specific dictionary. The objective function has two aims: 1) maximizing the MI between dictionary items within a specific class (intrinsic structure) and 2) minimizing the MI between the dictionary items in a given class and those of the other classes (extrinsic structure). We significantly reduce the computational complexity of CSMMI by introducing an novel submodular method, which is one of the important contributions of this paper. This paper also contributes a state-of-the-art end-to-end system for action and gesture recognition incorporating CSMMI, with feature extraction, learning initial dictionary per each class by sparse coding, CSMMI via submodularity, and classification based on reconstruction errors. We performed extensive experiments on synthetic data and eight benchmark data sets. Our experimental results show that CSMMI outperforms shared dictionary methods and that our end-to-end system is competitive with other state-of-the-art approaches.",
"In this paper, we present a home-monitoring oriented human activity recognition benchmark database, based on the combination of a color video camera and a depth sensor. Our contributions are two-fold: 1) We have created a publicly releasable human activity video database (i.e., named as RGBD-HuDaAct), which contains synchronized color-depth video streams, for the task of human daily activity recognition. This database aims at encouraging more research efforts on human activity recognition based on multi-modality sensor combination (e.g., color plus depth). 2) Two multi-modality fusion schemes, which naturally combine color and depth information, have been developed from two state-of-the-art feature representation methods for action recognition, i.e., spatio-temporal interest points (STIPs) and motion history images (MHIs). These depth-extended feature representation methods are evaluated comprehensively and superior recognition performances over their uni-modality (e.g., color only) counterparts are demonstrated."
]
} |
1611.06689 | 2552317450 | Recently, the popularity of depth-sensors such as Kinect has made depth videos easily available while its advantages have not been fully exploited. This paper investigates, for gesture recognition, to explore the spatial and temporal information complementarily embedded in RGB and depth sequences. We propose a convolutional twostream consensus voting network (2SCVN) which explicitly models both the short-term and long-term structure of the RGB sequences. To alleviate distractions from background, a 3d depth-saliency ConvNet stream (3DDSN) is aggregated in parallel to identify subtle motion characteristics. These two components in an unified framework significantly improve the recognition accuracy. On the challenging Chalearn IsoGD benchmark, our proposed method outperforms the first place on the leader-board by a large margin (10.29 ) while also achieving the best result on RGBD-HuDaAct dataset (96.74 ). Both quantitative experiments and qualitative analysis shows the effectiveness of our proposed framework and codes will be released to facilitate future research. | In the camp of action recognition, @cite_9 extended CNNs into video classification on a large-scale dataset of 1 million videos (Sports-1M). @cite_34 embraced recurrent nerual networks to explicitly model the complex temporal dynamics. @cite_15 proposed to simultaneously extract the spatio-temporal features with deep 3D Convolutional Neural Networks (3D-CNN) followed by a SVM classifier for classification. @cite_30 designed an architecture that captures the complementary information on appearance and motion between frames. Based on which, @cite_40 studied several levels of granularity in feature abstraction to fuse spatial and temporal cues. | {
"cite_N": [
"@cite_30",
"@cite_9",
"@cite_40",
"@cite_15",
"@cite_34"
],
"mid": [
"2952186347",
"2308045930",
"2342662179",
"2952633803",
"2951183276"
],
"abstract": [
"We investigate architectures of discriminatively trained deep Convolutional Networks (ConvNets) for action recognition in video. The challenge is to capture the complementary information on appearance from still frames and motion between frames. We also aim to generalise the best performing hand-crafted features within a data-driven learning framework. Our contribution is three-fold. First, we propose a two-stream ConvNet architecture which incorporates spatial and temporal networks. Second, we demonstrate that a ConvNet trained on multi-frame dense optical flow is able to achieve very good performance in spite of limited training data. Finally, we show that multi-task learning, applied to two different action classification datasets, can be used to increase the amount of training data and improve the performance on both. Our architecture is trained and evaluated on the standard video actions benchmarks of UCF-101 and HMDB-51, where it is competitive with the state of the art. It also exceeds by a large margin previous attempts to use deep nets for video classification.",
"",
"Recent applications of Convolutional Neural Networks (ConvNets) for human action recognition in videos have proposed different solutions for incorporating the appearance and motion information. We study a number of ways of fusing ConvNet towers both spatially and temporally in order to best take advantage of this spatio-temporal information. We make the following findings: (i) that rather than fusing at the softmax layer, a spatial and temporal network can be fused at a convolution layer without loss of performance, but with a substantial saving in parameters, (ii) that it is better to fuse such networks spatially at the last convolutional layer than earlier, and that additionally fusing at the class prediction layer can boost accuracy, finally (iii) that pooling of abstract convolutional features over spatiotemporal neighbourhoods further boosts performance. Based on these studies we propose a new ConvNet architecture for spatiotemporal fusion of video snippets, and evaluate its performance on standard benchmarks where this architecture achieves state-of-the-art results.",
"We propose a simple, yet effective approach for spatiotemporal feature learning using deep 3-dimensional convolutional networks (3D ConvNets) trained on a large scale supervised video dataset. Our findings are three-fold: 1) 3D ConvNets are more suitable for spatiotemporal feature learning compared to 2D ConvNets; 2) A homogeneous architecture with small 3x3x3 convolution kernels in all layers is among the best performing architectures for 3D ConvNets; and 3) Our learned features, namely C3D (Convolutional 3D), with a simple linear classifier outperform state-of-the-art methods on 4 different benchmarks and are comparable with current best methods on the other 2 benchmarks. In addition, the features are compact: achieving 52.8 accuracy on UCF101 dataset with only 10 dimensions and also very efficient to compute due to the fast inference of ConvNets. Finally, they are conceptually very simple and easy to train and use.",
"Models based on deep convolutional networks have dominated recent image interpretation tasks; we investigate whether models which are also recurrent, or \"temporally deep\", are effective for tasks involving sequences, visual and otherwise. We develop a novel recurrent convolutional architecture suitable for large-scale visual learning which is end-to-end trainable, and demonstrate the value of these models on benchmark video recognition tasks, image description and retrieval problems, and video narration challenges. In contrast to current models which assume a fixed spatio-temporal receptive field or simple temporal averaging for sequential processing, recurrent convolutional models are \"doubly deep\"' in that they can be compositional in spatial and temporal \"layers\". Such models may have advantages when target concepts are complex and or training data are limited. Learning long-term dependencies is possible when nonlinearities are incorporated into the network state updates. Long-term RNN models are appealing in that they directly can map variable-length inputs (e.g., video frames) to variable length outputs (e.g., natural language text) and can model complex temporal dynamics; yet they can be optimized with backpropagation. Our recurrent long-term models are directly connected to modern visual convnet models and can be jointly trained to simultaneously learn temporal dynamics and convolutional perceptual representations. Our results show such models have distinct advantages over state-of-the-art models for recognition or generation which are separately defined and or optimized."
]
} |
1611.06737 | 2950951350 | Online Social Networks (OSNs), such as Facebook, provide users with tools to share information along with a set of privacy controls preferences to regulate the spread of information. Current privacy controls are efficient to protect content data. However, the complexity of tuning them undermine their efficiency when protecting contextual information (such as the social network structure) that many users believe being kept private. In this paper, we demonstrate the extent of the problem of information leakage in Facebook. In particular, we show the possibility of inferring, from the network "surrounding" a victim user, some information that the victim set as hidden. We developed a system, named OSSINT (Open Source Social Network INTelligence), on top of our previous tool SocialSpy, that is able to infer hidden information of a victim profile and retrieve private information from public one. OSSINT retrieves the friendship network of a victim and shows how it is possible to infer additional private information (e.g., user personal preferences and hobbies). Our proposed system OSSINT goes extra mile about the network topology information, i.e., predicting new friendships using the victim's friends of friends network (2-hop of distance from the victim profile), and hence possibly deduce private information of the full Facebook network. OSSINT correctly improved the previous results of SocialSpy predicting an average of 11 additional friendships with peaks of 20 new friends. Moreover, OSSINT, for the considered victim profiles demonstrated how it is possible to infer real life information such as current city, hometown, university, supposed being private. | An attack example is in @cite_22 . demonstrate how it is possible to retrieve the friends list of a victim user in Facebook, using public information. Results have an average of 25 there is the solutions category where we have works that try to mitigate the lack of information from OSNs. A solution is proposed by in @cite_13 with Social Privacy Protector software (SPP). The SPP software consists of two main parts, namely, a Firefox add-on and a Facebook application. The two parts provide Facebook users with three different layers of protection. The first layer enables Facebook users to easily control their profile privacy. The second layer notifies users of the number of applications installed on their profile which may impose a threat to their privacy. The third layer, a Facebook application, analyzes a user’s friends list. | {
"cite_N": [
"@cite_13",
"@cite_22"
],
"mid": [
"2952877703",
"2952991763"
],
"abstract": [
"The amount of personal information unwillingly exposed by users on online social networks is staggering, as shown in recent research. Moreover, recent reports indicate that these networks are infested with tens of millions of fake users profiles, which may jeopardize the users' security and privacy. To identify fake users in such networks and to improve users' security and privacy, we developed the Social Privacy Protector software for Facebook. This software contains three protection layers, which improve user privacy by implementing different methods. The software first identifies a user's friends who might pose a threat and then restricts this \"friend's\" exposure to the user's personal information. The second layer is an expansion of Facebook's basic privacy settings based on different types of social network usage profiles. The third layer alerts users about the number of installed applications on their Facebook profile, which have access to their private information. An initial version of the Social Privacy Protection software received high media coverage, and more than 3,000 users from more than twenty countries have installed the software, out of which 527 used the software to restrict more than nine thousand friends. In addition, we estimate that more than a hundred users accepted the software's recommendations and removed at least 1,792 Facebook applications from their profiles. By analyzing the unique dataset obtained by the software in combination with machine learning techniques, we developed classifiers, which are able to predict which Facebook profiles have high probabilities of being fake and therefore, threaten the user's well-being. Moreover, in this study, we present statistics on users' privacy settings and statistics of the number of applications installed on Facebook profiles...",
"Online Social Networks are becoming the most important \"places\" where people share information about their lives. With the increasing concern that users have about privacy, most social networks offer ways to control the privacy of the user. Unfortunately, we believe that current privacy settings are not as effective as users might think. In this paper, we highlight this problem focusing on one of the most popular social networks, Facebook. In particular, we show how easy it is to retrieve information that a user might have set as (and hence thought as) \"private\". As a case study, we focus on retrieving the list of friends for users that did set this information as \"hidden\" (to non-friends). We propose four different strategies to achieve this goal, and we evaluate them. The results of our thorough experiments show the feasibility of our strategies as well as their effectiveness: our approach is able to retrieve a significant percentage of the names of the \"hidden\" friends: i.e., some 25 on average, and more than 70 for some users."
]
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.