May 6, 2019

2755 words 13 mins read

Paper Group ANR 300

Tinkering Under the Hood: Interactive Zero-Shot Learning with Net Surgery. Automatic Labelling of Topics with Neural Embeddings. Efficient Globally Convergent Stochastic Optimization for Canonical Correlation Analysis. Decentralized Dynamic Discriminative Dictionary Learning. How Image Degradations Affect Deep CNN-based Face Recognition?. Superpixe …

Tinkering Under the Hood: Interactive Zero-Shot Learning with Net Surgery


Title	Tinkering Under the Hood: Interactive Zero-Shot Learning with Net Surgery
Authors	Vivek Krishnan, Deva Ramanan
Abstract	We consider the task of visual net surgery, in which a CNN can be reconfigured without extra data to recognize novel concepts that may be omitted from the training set. While most prior work make use of linguistic cues for such “zero-shot” learning, we do so by using a pictorial language representation of the training set, implicitly learned by a CNN, to generalize to new classes. To this end, we introduce a set of visualization techniques that better reveal the activation patterns and relations between groups of CNN filters. We next demonstrate that knowledge of pictorial languages can be used to rewire certain CNN neurons into a part model, which we call a pictorial language classifier. We demonstrate the robustness of simple PLCs by applying them in a weakly supervised manner: labeling unlabeled concepts for visual classes present in the training data. Specifically we show that a PLC built on top of a CNN trained for ImageNet classification can localize humans in Graz-02 and determine the pose of birds in PASCAL-VOC without extra labeled data or additional training. We then apply PLCs in an interactive zero-shot manner, demonstrating that pictorial languages are expressive enough to detect a set of visual classes in MS-COCO that never appear in the ImageNet training set.
Tasks	Zero-Shot Learning
Published	2016-12-15
URL	http://arxiv.org/abs/1612.04901v1
PDF	http://arxiv.org/pdf/1612.04901v1.pdf
PWC	https://paperswithcode.com/paper/tinkering-under-the-hood-interactive-zero
Repo
Framework

Automatic Labelling of Topics with Neural Embeddings


Title	Automatic Labelling of Topics with Neural Embeddings
Authors	Shraey Bhatia, Jey Han Lau, Timothy Baldwin
Abstract	Topics generated by topic models are typically represented as list of terms. To reduce the cognitive overhead of interpreting these topics for end-users, we propose labelling a topic with a succinct phrase that summarises its theme or idea. Using Wikipedia document titles as label candidates, we compute neural embeddings for documents and words to select the most relevant labels for topics. Compared to a state-of-the-art topic labelling system, our methodology is simpler, more efficient, and finds better topic labels.
Tasks	Topic Models
Published	2016-12-16
URL	http://arxiv.org/abs/1612.05340v2
PDF	http://arxiv.org/pdf/1612.05340v2.pdf
PWC	https://paperswithcode.com/paper/automatic-labelling-of-topics-with-neural
Repo
Framework

Efficient Globally Convergent Stochastic Optimization for Canonical Correlation Analysis


Title	Efficient Globally Convergent Stochastic Optimization for Canonical Correlation Analysis
Authors	Weiran Wang, Jialei Wang, Dan Garber, Nathan Srebro
Abstract	We study the stochastic optimization of canonical correlation analysis (CCA), whose objective is nonconvex and does not decouple over training samples. Although several stochastic gradient based optimization algorithms have been recently proposed to solve this problem, no global convergence guarantee was provided by any of them. Inspired by the alternating least squares/power iterations formulation of CCA, and the shift-and-invert preconditioning method for PCA, we propose two globally convergent meta-algorithms for CCA, both of which transform the original problem into sequences of least squares problems that need only be solved approximately. We instantiate the meta-algorithms with state-of-the-art SGD methods and obtain time complexities that significantly improve upon that of previous work. Experimental results demonstrate their superior performance.
Tasks	Stochastic Optimization
Published	2016-04-07
URL	http://arxiv.org/abs/1604.01870v4
PDF	http://arxiv.org/pdf/1604.01870v4.pdf
PWC	https://paperswithcode.com/paper/efficient-globally-convergent-stochastic
Repo
Framework

Decentralized Dynamic Discriminative Dictionary Learning


Title	Decentralized Dynamic Discriminative Dictionary Learning
Authors	Alec Koppel, Garrett Warnell, Ethan Stump, Alejandro Ribeiro
Abstract	We consider discriminative dictionary learning in a distributed online setting, where a network of agents aims to learn a common set of dictionary elements of a feature space and model parameters while sequentially receiving observations. We formulate this problem as a distributed stochastic program with a non-convex objective and present a block variant of the Arrow-Hurwicz saddle point algorithm to solve it. Using Lagrange multipliers to penalize the discrepancy between them, only neighboring nodes exchange model information. We show that decisions made with this saddle point algorithm asymptotically achieve a first-order stationarity condition on average.
Tasks	Dictionary Learning
Published	2016-05-03
URL	http://arxiv.org/abs/1605.01107v1
PDF	http://arxiv.org/pdf/1605.01107v1.pdf
PWC	https://paperswithcode.com/paper/decentralized-dynamic-discriminative
Repo
Framework

How Image Degradations Affect Deep CNN-based Face Recognition?


Title	How Image Degradations Affect Deep CNN-based Face Recognition?
Authors	Samil Karahan, Merve Kilinc Yildirim, Kadir Kirtac, Ferhat Sukru Rende, Gultekin Butun, Hazim Kemal Ekenel
Abstract	Face recognition approaches that are based on deep convolutional neural networks (CNN) have been dominating the field. The performance improvements they have provided in the so called in-the-wild datasets are significant, however, their performance under image quality degradations have not been assessed, yet. This is particularly important, since in real-world face recognition applications, images may contain various kinds of degradations due to motion blur, noise, compression artifacts, color distortions, and occlusion. In this work, we have addressed this problem and analyzed the influence of these image degradations on the performance of deep CNN-based face recognition approaches using the standard LFW closed-set identification protocol. We have evaluated three popular deep CNN models, namely, the AlexNet, VGG-Face, and GoogLeNet. Results have indicated that blur, noise, and occlusion cause a significant decrease in performance, while deep CNN models are found to be robust to distortions, such as color distortions and change in color balance.
Tasks	Face Recognition
Published	2016-08-18
URL	http://arxiv.org/abs/1608.05246v1
PDF	http://arxiv.org/pdf/1608.05246v1.pdf
PWC	https://paperswithcode.com/paper/how-image-degradations-affect-deep-cnn-based
Repo
Framework

Superpixel-based Two-view Deterministic Fitting for Multiple-structure Data


Title	Superpixel-based Two-view Deterministic Fitting for Multiple-structure Data
Authors	Guobao Xiao, Hanzi Wang, Yan Yan, David Suter
Abstract	This paper proposes a two-view deterministic geometric model fitting method, termed Superpixel-based Deterministic Fitting (SDF), for multiple-structure data. SDF starts from superpixel segmentation, which effectively captures prior information of feature appearances. The feature appearances are beneficial to reduce the computational complexity for deterministic fitting methods. SDF also includes two original elements, i.e., a deterministic sampling algorithm and a novel model selection algorithm. The two algorithms are tightly coupled to boost the performance of SDF in both speed and accuracy. Specifically, the proposed sampling algorithm leverages the grouping cues of superpixels to generate reliable and consistent hypotheses. The proposed model selection algorithm further makes use of desirable properties of the generated hypotheses, to improve the conventional fit-and-remove framework for more efficient and effective performance. The key characteristic of SDF is that it can efficiently and deterministically estimate the parameters of model instances in multi-structure data. Experimental results demonstrate that the proposed SDF shows superiority over several state-of-the-art fitting methods for real images with single-structure and multiple-structure data.
Tasks	Model Selection
Published	2016-07-20
URL	http://arxiv.org/abs/1607.05839v1
PDF	http://arxiv.org/pdf/1607.05839v1.pdf
PWC	https://paperswithcode.com/paper/superpixel-based-two-view-deterministic
Repo
Framework

An Aggregate and Iterative Disaggregate Algorithm with Proven Optimality in Machine Learning


Title	An Aggregate and Iterative Disaggregate Algorithm with Proven Optimality in Machine Learning
Authors	Young Woong Park, Diego Klabjan
Abstract	We propose a clustering-based iterative algorithm to solve certain optimization problems in machine learning, where we start the algorithm by aggregating the original data, solving the problem on aggregated data, and then in subsequent steps gradually disaggregate the aggregated data. We apply the algorithm to common machine learning problems such as the least absolute deviation regression problem, support vector machines, and semi-supervised support vector machines. We derive model-specific data aggregation and disaggregation procedures. We also show optimality, convergence, and the optimality gap of the approximated solution in each iteration. A computational study is provided.
Tasks
Published	2016-07-05
URL	http://arxiv.org/abs/1607.01400v1
PDF	http://arxiv.org/pdf/1607.01400v1.pdf
PWC	https://paperswithcode.com/paper/an-aggregate-and-iterative-disaggregate
Repo
Framework

Kernel Bayesian Inference with Posterior Regularization


Title	Kernel Bayesian Inference with Posterior Regularization
Authors	Yang Song, Jun Zhu, Yong Ren
Abstract	We propose a vector-valued regression problem whose solution is equivalent to the reproducing kernel Hilbert space (RKHS) embedding of the Bayesian posterior distribution. This equivalence provides a new understanding of kernel Bayesian inference. Moreover, the optimization problem induces a new regularization for the posterior embedding estimator, which is faster and has comparable performance to the squared regularization in kernel Bayes’ rule. This regularization coincides with a former thresholding approach used in kernel POMDPs whose consistency remains to be established. Our theoretical work solves this open problem and provides consistency analysis in regression settings. Based on our optimizational formulation, we propose a flexible Bayesian posterior regularization framework which for the first time enables us to put regularization at the distribution level. We apply this method to nonparametric state-space filtering tasks with extremely nonlinear dynamics and show performance gains over all other baselines.
Tasks	Bayesian Inference
Published	2016-07-07
URL	http://arxiv.org/abs/1607.02011v2
PDF	http://arxiv.org/pdf/1607.02011v2.pdf
PWC	https://paperswithcode.com/paper/kernel-bayesian-inference-with-posterior
Repo
Framework

Stable-Unstable Semantics: Beyond NP with Normal Logic Programs


Title	Stable-Unstable Semantics: Beyond NP with Normal Logic Programs
Authors	Bart Bogaerts, Tomi Janhunen, Shahab Tasharrofi
Abstract	Standard answer set programming (ASP) targets at solving search problems from the first level of the polynomial time hierarchy (PH). Tackling search problems beyond NP using ASP is less straightforward. The class of disjunctive logic programs offers the most prominent way of reaching the second level of the PH, but encoding respective hard problems as disjunctive programs typically requires sophisticated techniques such as saturation or meta-interpretation. The application of such techniques easily leads to encodings that are inaccessible to non-experts. Furthermore, while disjunctive ASP solvers often rely on calls to a (co-)NP oracle, it may be difficult to detect from the input program where the oracle is being accessed. In other formalisms, such as Quantified Boolean Formulas (QBFs), the interface to the underlying oracle is more transparent as it is explicitly recorded in the quantifier prefix of a formula. On the other hand, ASP has advantages over QBFs from the modeling perspective. The rich high-level languages such as ASP-Core-2 offer a wide variety of primitives that enable concise and natural encodings of search problems. In this paper, we present a novel logic programming–based modeling paradigm that combines the best features of ASP and QBFs. We develop so-called combined logic programs in which oracles are directly cast as (normal) logic programs themselves. Recursive incarnations of this construction enable logic programming on arbitrarily high levels of the PH. We develop a proof-of-concept implementation for our new paradigm. This paper is under consideration for acceptance in TPLP.
Tasks
Published	2016-08-05
URL	http://arxiv.org/abs/1608.01835v3
PDF	http://arxiv.org/pdf/1608.01835v3.pdf
PWC	https://paperswithcode.com/paper/stable-unstable-semantics-beyond-np-with
Repo
Framework

Symmetry-aware Depth Estimation using Deep Neural Networks


Title	Symmetry-aware Depth Estimation using Deep Neural Networks
Authors	Guilin Liu, Chao Yang, Zimo Li, Duygu Ceylan, Qixing Huang
Abstract	Due to the abundance of 2D product images from the Internet, developing efficient and scalable algorithms to recover the missing depth information is central to many applications. Recent works have addressed the single-view depth estimation problem by utilizing convolutional neural networks. In this paper, we show that exploring symmetry information, which is ubiquitous in man made objects, can significantly boost the quality of such depth predictions. Specifically, we propose a new convolutional neural network architecture to first estimate dense symmetric correspondences in a product image and then propose an optimization which utilizes this information explicitly to significantly improve the quality of single-view depth estimations. We have evaluated our approach extensively, and experimental results show that this approach outperforms state-of-the-art depth estimation techniques.
Tasks	Depth Estimation
Published	2016-04-20
URL	http://arxiv.org/abs/1604.06079v2
PDF	http://arxiv.org/pdf/1604.06079v2.pdf
PWC	https://paperswithcode.com/paper/symmetry-aware-depth-estimation-using-deep
Repo
Framework

Network cross-validation by edge sampling


Title	Network cross-validation by edge sampling
Authors	Tianxi Li, Elizaveta Levina, Ji Zhu
Abstract	While many statistical models and methods are now available for network analysis, resampling network data remains a challenging problem. Cross-validation is a useful general tool for model selection and parameter tuning, but is not directly applicable to networks since splitting network nodes into groups requires deleting edges and destroys some of the network structure. Here we propose a new network resampling strategy based on splitting node pairs rather than nodes applicable to cross-validation for a wide range of network model selection tasks. We provide a theoretical justification for our method in a general setting and examples of how our method can be used in specific network model selection and parameter tuning tasks. Numerical results on simulated networks and on a citation network of statisticians show that this cross-validation approach works well for model selection.
Tasks	Model Selection
Published	2016-12-14
URL	http://arxiv.org/abs/1612.04717v6
PDF	http://arxiv.org/pdf/1612.04717v6.pdf
PWC	https://paperswithcode.com/paper/network-cross-validation-by-edge-sampling
Repo
Framework

Friction from Reflectance: Deep Reflectance Codes for Predicting Physical Surface Properties from One-Shot In-Field Reflectance


Title	Friction from Reflectance: Deep Reflectance Codes for Predicting Physical Surface Properties from One-Shot In-Field Reflectance
Authors	Hang Zhang, Kristin Dana, Ko Nishino
Abstract	Images are the standard input for vision algorithms, but one-shot infield reflectance measurements are creating new opportunities for recognition and scene understanding. In this work, we address the question of what reflectance can reveal about materials in an efficient manner. We go beyond the question of recognition and labeling and ask the question: What intrinsic physical properties of the surface can be estimated using reflectance? We introduce a framework that enables prediction of actual friction values for surfaces using one-shot reflectance measurements. This work is a first of its kind vision-based friction estimation. We develop a novel representation for reflectance disks that capture partial BRDF measurements instantaneously. Our method of deep reflectance codes combines CNN features and fisher vector pooling with optimal binary embedding to create codes that have sufficient discriminatory power and have important properties of illumination and spatial invariance. The experimental results demonstrate that reflectance can play a new role in deciphering the underlying physical properties of real-world scenes.
Tasks	Scene Understanding
Published	2016-03-25
URL	http://arxiv.org/abs/1603.07998v2
PDF	http://arxiv.org/pdf/1603.07998v2.pdf
PWC	https://paperswithcode.com/paper/friction-from-reflectance-deep-reflectance
Repo
Framework

Cross-Lingual Predicate Mapping Between Linked Data Ontologies


Title	Cross-Lingual Predicate Mapping Between Linked Data Ontologies
Authors	Gautam Singh, Saemi Jang, Mun Y. Yi
Abstract	Ontologies in different natural languages often differ in quality in terms of richness of schema or richness of internal links. This difference is markedly visible when comparing a rich English language ontology with a non-English language counterpart. Discovering alignment between them is a useful endeavor as it serves as a starting point in bridging the disparity. In particular, our work is motivated by the absence of inter-language links for predicates in the localised versions of DBpedia. In this paper, we propose and demonstrate an ad-hoc system to find possible owl:equivalentProperty links between predicates in ontologies of different natural languages. We seek to achieve this mapping by using pre-existing inter-language links of the resources connected by the given predicate. Thus, our methodology stresses on semantic similarity rather than lexical. Moreover, through an evaluation, we show that our system is capable of outperforming a baseline system that is similar to the one used in recent OAEI campaigns.
Tasks	Semantic Similarity, Semantic Textual Similarity
Published	2016-12-06
URL	http://arxiv.org/abs/1612.01892v1
PDF	http://arxiv.org/pdf/1612.01892v1.pdf
PWC	https://paperswithcode.com/paper/cross-lingual-predicate-mapping-between
Repo
Framework

Condorcet’s Jury Theorem for Consensus Clustering and its Implications for Diversity


Title	Condorcet’s Jury Theorem for Consensus Clustering and its Implications for Diversity
Authors	Brijnesh J. Jain
Abstract	Condorcet’s Jury Theorem has been invoked for ensemble classifiers to indicate that the combination of many classifiers can have better predictive performance than a single classifier. Such a theoretical underpinning is unknown for consensus clustering. This article extends Condorcet’s Jury Theorem to the mean partition approach under the additional assumptions that a unique ground-truth partition exists and sample partitions are drawn from a sufficiently small ball containing the ground-truth. As an implication of practical relevance, we question the claim that the quality of consensus clustering depends on the diversity of the sample partitions. Instead, we conjecture that limiting the diversity of the mean partitions is necessary for controlling the quality.
Tasks
Published	2016-04-26
URL	http://arxiv.org/abs/1604.07711v2
PDF	http://arxiv.org/pdf/1604.07711v2.pdf
PWC	https://paperswithcode.com/paper/condorcets-jury-theorem-for-consensus
Repo
Framework

3D Image Reconstruction from X-Ray Measurements with Overlap


Title	3D Image Reconstruction from X-Ray Measurements with Overlap
Authors	Maria Klodt, Raphael Hauser
Abstract	3D image reconstruction from a set of X-ray projections is an important image reconstruction problem, with applications in medical imaging, industrial inspection and airport security. The innovation of X-ray emitter arrays allows for a novel type of X-ray scanners with multiple simultaneously emitting sources. However, two or more sources emitting at the same time can yield measurements from overlapping rays, imposing a new type of image reconstruction problem based on nonlinear constraints. Using traditional linear reconstruction methods, respective scanner geometries have to be implemented such that no rays overlap, which severely restricts the scanner design. We derive a new type of 3D image reconstruction model with nonlinear constraints, based on measurements with overlapping X-rays. Further, we show that the arising optimization problem is partially convex, and present an algorithm to solve it. Experiments show highly improved image reconstruction results from both simulated and real-world measurements.
Tasks	Image Reconstruction
Published	2016-11-22
URL	http://arxiv.org/abs/1611.07390v1
PDF	http://arxiv.org/pdf/1611.07390v1.pdf
PWC	https://paperswithcode.com/paper/3d-image-reconstruction-from-x-ray
Repo
Framework