今日学术视野(2015.10.6)

cs.AI - 人工智能
cs.CL - 计算与语言
cs.CV - 机器视觉与模式识别
cs.DB - 数据库
cs.DS - 数据结构与算法
cs.IR - 信息检索
cs.IT - 信息论
cs.LG - 自动学习
cs.NE - 神经与进化计算
math.ST - 统计理论
quant-ph - 量子物理
stat.AP - 应用统计
stat.CO - 统计计算
stat.ME - 统计方法论
stat.ML - (统计)机器学习

• [cs.AI]Online Vision- and Action-Based Object Classification Using Both Symbolic and Subsymbolic Knowledge Representations
• [cs.CL]Automatic Taxonomy Extraction from Query Logs with no Additional Sources of Information
• [cs.CL]Response to Liu, Xu, and Liang (2015) and Ferrer-i-Cancho and Gómez-Rodríguez (2015) on Dependency Length Minimization
• [cs.CV]Effective Object Tracking in Unstructured Crowd Scenes
• [cs.CV]Human Action Recognition using Factorized Spatio-Temporal Convolutional Networks
• [cs.CV]Learning a Discriminative Model for the Perception of Realism in Composite Images
• [cs.CV]Local Higher-Order Statistics (LHS) describing images with statistics of local non-binarized pixel patterns
• [cs.CV]Off-the-Grid Recovery of Piecewise Constant Images from Few Fourier Samples
• [cs.DB]Exposing the Probabilistic Causal Structure of Discrimination
• [cs.DS]Implementing Efficient All Solutions SAT Solvers
• [cs.IR]A Complex Network Approach for Collaborative Recommendation
• [cs.IT]Minimax Lower Bounds for Noisy Matrix Completion Under Sparse Factor Models
• [cs.LG]Minimax Binary Classifier Aggregation with General Losses
• [cs.LG]Multi-armed Bandits with Application to 5G Small Cells
• [cs.NE]An Asynchronous Implementation of the Limited Memory CMA-ES
• [math.ST]A Selective Approach to Internal Inference
• [math.ST]Model Selection and Multiple Testing - A Bayesian and Empirical Bayes Overview and some New Results
• [quant-ph]Autonomous Perceptron Neural Network Inspired from Quantum computing
• [stat.AP]Bayesian modeling of networks in complex business intelligence problems
• [stat.AP]Microwave Surveillance based on Ghost Imaging and Distributed Antennas
• [stat.CO]A Bayesian approach to constrained single- and multi-objective optimization
• [stat.CO]On Estimation of Parameter Uncertainty in Model-Based Clustering
• [stat.ME]The zero-inflated promotion cure rate regression model applied to fraud propensity in bank loan applications
• [stat.ML]Distributed Multitask Learning 

·····································

• [cs.AI]Online Vision- and Action-Based Object Classification Using Both Symbolic and Subsymbolic Knowledge Representations
Laura Steinert, Jens Hoefinghoff, Josef Pauli
//arxiv.org/abs/1510.00604v1 

If a robot is supposed to roam an environment and interact with objects, it is often necessary to know all possible objects in advance, so that a database with models of all objects can be generated for visual identification. However, this constraint cannot always be fulfilled. Due to that reason, a model based object recognition cannot be used to guide the robot’s interactions. Therefore, this paper proposes a system that analyzes features of encountered objects and then uses these features to compare unknown objects to already known ones. From the resulting similarity appropriate actions can be derived. Moreover, the system enables the robot to learn object categories by grouping similar objects or by splitting existing categories. To represent the knowledge a hybrid form is used, consisting of both symbolic and subsymbolic representations. 

• [cs.CL]Automatic Taxonomy Extraction from Query Logs with no Additional Sources of Information
Miguel Fernandez-Fernandez, Daniel Gayo Avello
//arxiv.org/abs/1510.00618v1 

Search engine logs store detailed information on Web users interactions. Thus, as more and more people use search engines on a daily basis, important trails of users common knowledge are being recorded in those files. Previous research has shown that it is possible to extract concept taxonomies from full text documents, while other scholars have proposed methods to obtain similar queries from query logs. We propose a mixture of both lines of research, that is, mining query logs not to find related queries nor query hierarchies, but actual term taxonomies that could be used to improve search engine effectiveness and efficiency. As a result, in this study we have developed a method that combines lexical heuristics with a supervised classification model to successfully extract hyponymy relations from specialization search patterns revealed from log missions, with no additional sources of information, and in a language independent way. 

• [cs.CL]Response to Liu, Xu, and Liang (2015) and Ferrer-i-Cancho and Gómez-Rodríguez (2015) on Dependency Length Minimization
Richard Futrell, Kyle Mahowald, Edward Gibson
//arxiv.org/abs/1510.00436v1 

We address recent criticisms (Liu et al., 2015; Ferrer-i-Cancho and G\‘omez-Rodr\'iguez, 2015) of our work on empirical evidence of dependency length minimization across languages (Futrell et al., 2015). First, we acknowledge error in failing to acknowledge Liu (2008)’s previous work on corpora of 20 languages with similar aims. A correction will appear in PNAS. Nevertheless, we argue that our work provides novel, strong evidence for dependency length minimization as a universal quantitative property of languages, beyond this previous work, because it provides baselines which focus on word order preferences. Second, we argue that our choices of baselines were appropriate because they control for alternative theories. 

• [cs.CV]Effective Object Tracking in Unstructured Crowd Scenes
Ishan Jindal, Shanmuganathan Raman
//arxiv.org/abs/1510.00479v1 

In this paper, we are presenting a rotation variant Oriented Texture Curve (OTC) descriptor based mean shift algorithm for tracking an object in an unstructured crowd scene. The proposed algorithm works by first obtaining the OTC features for a manually selected object target, then a visual vocabulary is created by using all the OTC features of the target. The target histogram is obtained using codebook encoding method which is then used in mean shift framework to perform similarity search. Results are obtained on different videos of challenging scenes and the comparison of the proposed approach with several state-of-the-art approaches are provided. The analysis shows the advantages and limitations of the proposed approach for tracking an object in unstructured crowd scenes. 

• [cs.CV]Human Action Recognition using Factorized Spatio-Temporal Convolutional Networks
Lin Sun, Kui Jia, Dit-Yan Yeung, Bertram E. Shi
//arxiv.org/abs/1510.00562v1 

Human actions in video sequences are three-dimensional (3D) spatio-temporal signals characterizing both the visual appearance and motion dynamics of the involved humans and objects. Inspired by the success of convolutional neural networks (CNN) for image classification, recent attempts have been made to learn 3D CNNs for recognizing human actions in videos. However, partly due to the high complexity of training 3D convolution kernels and the need for large quantities of training videos, only limited success has been reported. This has triggered us to investigate in this paper a new deep architecture which can handle 3D signals more effectively. Specifically, we propose factorized spatio-temporal convolutional networks (FstCN) that factorize the original 3D convolution kernel learning as a sequential process of learning 2D spatial kernels in the lower layers (called spatial convolutional layers), followed by learning 1D temporal kernels in the upper layers (called temporal convolutional layers). We introduce a novel transformation and permutation operator to make factorization in FstCN possible. Moreover, to address the issue of sequence alignment, we propose an effective training and inference strategy based on sampling multiple video clips from a given action video sequence. We have tested FstCN on two commonly used benchmark datasets (UCF-101 and HMDB-51). Without using auxiliary training videos to boost the performance, FstCN outperforms existing CNN based methods and achieves comparable performance with a recent method that benefits from using auxiliary training videos. 

• [cs.CV]Learning a Discriminative Model for the Perception of Realism in Composite Images
Jun-Yan Zhu, Philipp Krähenbühl, Eli Shechtman, Alexei A. Efros
//arxiv.org/abs/1510.00477v1 

What makes an image appear realistic? In this work, we are answering this question from a data-driven perspective by learning the perception of visual realism directly from large amounts of data. In particular, we train a Convolutional Neural Network (CNN) model that distinguishes natural photographs from automatically generated composite images. The model learns to predict visual realism of a scene in terms of color, lighting and texture compatibility, without any human annotations pertaining to it. Our model outperforms previous works that rely on hand-crafted heuristics, for the task of classifying realistic vs. unrealistic photos. Furthermore, we apply our learned model to compute optimal parameters of a compositing method, to maximize the visual realism score predicted by our CNN model. We demonstrate its advantage against existing methods via a human perception study. 

• [cs.CV]Local Higher-Order Statistics (LHS) describing images with statistics of local non-binarized pixel patterns
Gaurav Sharma, Frederic Jurie
//arxiv.org/abs/1510.00542v1 

We propose a new image representation for texture categorization and facial analysis, relying on the use of higher-order local differential statistics as features. It has been recently shown that small local pixel pattern distributions can be highly discriminative while being extremely efficient to compute, which is in contrast to the models based on the global structure of images. Motivated by such works, we propose to use higher-order statistics of local non-binarized pixel patterns for the image description. The proposed model does not require either (i) user specified quantization of the space (of pixel patterns) or (ii) any heuristics for discarding low occupancy volumes of the space. We propose to use a data driven soft quantization of the space, with parametric mixture models, combined with higher-order statistics, based on Fisher scores. We demonstrate that this leads to a more expressive representation which, when combined with discriminatively learned classifiers and metrics, achieves state-of-the-art performance on challenging texture and facial analysis datasets, in low complexity setup. Further, it is complementary to higher complexity features and when combined with them improves performance. 

• [cs.CV]Off-the-Grid Recovery of Piecewise Constant Images from Few Fourier Samples
Greg Ongie, Mathews Jacob
//arxiv.org/abs/1510.00384v1 

We introduce a method to recover a continuous domain representation of a piecewise constant two-dimensional image from few low-pass Fourier samples. Assuming the edge set of the image is localized to the zero set of a trigonometric polynomial, we show the Fourier coefficients of the partial derivatives of the image satisfy a linear annihilation relation. We present necessary and sufficient conditions for unique recovery of the image from finite low-pass Fourier samples using the annihilation relation. We also propose a practical two-stage recovery algorithm which is robust to model-mismatch and noise. In the first stage we estimate a continuous domain representation of the edge set of the image. In the second stage we perform an extrapolation in Fourier domain by a least squares two-dimensional linear prediction, which recovers the exact Fourier coefficients of the underlying image. We demonstrate our algorithm on the super-resolution recovery of MRI phantoms and real MRI data from low-pass Fourier samples, and show the superiority of the method over standard approaches for single-image super-resolution MRI. 

• [cs.DB]Exposing the Probabilistic Causal Structure of Discrimination
Francesco Bonchi, Sara Hajian, Bud Mishra, Daniele Ramazzotti
//arxiv.org/abs/1510.00552v1 

\emph{Discrimination discovery} from data is an important task aiming at identifying patterns of illegal and unethical discriminatory activities against protected-by-law groups, e.g.,~ethnic minorities. While any legally-valid proof of discrimination requires evidence of causality, the state-of-the-art methods are essentially correlation-based, albeit, as it is well known, correlation does not imply causation. In this paper we take a principled causal approach to the data mining problem of discrimination detection in databases. Following Suppes’ \emph{probabilistic causation theory}, we define a method to extract, from a dataset of historical decision records, the causal structures existing among the attributes in the data. The result is a type of constrained Bayesian network, which we dub \emph{Suppes-Bayes Causal Network} (\SBCN). Next, we develop a toolkit of methods based on random walks on top of the \SBCN, addressing different anti-discrimination legal concepts, such as direct and indirect discrimination, group and individual discrimination, genuine requirement, and favoritism. Our experiments on real-world datasets confirm the inferential power of our approach in all these different tasks. 

• [cs.DS]Implementing Efficient All Solutions SAT Solvers
Takahisa Toda, Takehide Soh
//arxiv.org/abs/1510.00523v1 

All solutions SAT (AllSAT for short) is a variant of propositional satisfiability problem. Despite its significance, AllSAT has been relatively unexplored compared to other variants. We thus survey and discuss major techniques of AllSAT solvers. We faithfully implement them and conduct comprehensive experiments using a large number of instances and various types of solvers including one of the few public softwares. The experiments reveal solver’s characteristics. Our implemented solvers are made publicly available so that other researchers can easily develop their solver by modifying our codes and compare it with existing methods. 

• [cs.IR]A Complex Network Approach for Collaborative Recommendation
Ranveer Singh, Bidyut Kr. Patra, Bibhas Adhikari
//arxiv.org/abs/1510.00585v1 

Collaborative filtering (CF) is the most widely used and successful approach for personalized service recommendations. Among the collaborative recommendation approaches, neighborhood based approaches enjoy a huge amount of popularity, due to their simplicity, justifiability, efficiency and stability. Neighborhood based collaborative filtering approach finds K nearest neighbors to an active user or K most similar rated items to the target item for recommendation. Traditional similarity measures use ratings of co-rated items to find similarity between a pair of users. Therefore, traditional similarity measures cannot compute effective neighbors in sparse dataset. In this paper, we propose a two-phase approach, which generates user-user and item-item networks using traditional similarity measures in the first phase. In the second phase, two hybrid approaches HB1, HB2, which utilize structural similarity of both the network for finding K nearest neighbors and K most similar items to a target items are introduced. To show effectiveness of the measures, we compared performances of neighborhood based CFs using state-of-the-art similarity measures with our proposed structural similarity measures based CFs. Recommendation results on a set of real data show that proposed measures based CFs outperform existing measures based CFs in various evaluation metrics. 

• [cs.IT]Minimax Lower Bounds for Noisy Matrix Completion Under Sparse Factor Models
Abhinav V. Sambasivan, Jarvis D. Haupt
//arxiv.org/abs/1510.00701v1 

This paper examines fundamental error characteristics for a general class of matrix completion problems, where matrix of interest is a product of two a priori unknown matrices, one of which is sparse, and the observations are noisy. Our main contributions come in the form of minimax lower bounds for the expected per-element squared error for these problems under several noise/corruption models; specifically, we analyze scenarios where the corruptions are characterized by additive Gaussian noise or additive heavier-tailed (Laplace) noise, Poisson-distributed observations, and highly-quantized (e.g., one-bit) observations. Our results establish that the error bounds derived in (Soni et al., 2014) for complexity-regularized maximum likelihood estimators achieve, up to multiplicative constant and logarithmic factors, the minimax error rates in each of these noise scenarios, provided the sparse factor exhibits linear sparsity. 

• [cs.LG]Minimax Binary Classifier Aggregation with General Losses
Akshay Balsubramani, Yoav Freund
//arxiv.org/abs/1510.00452v1 

We develop a worst-case analysis of aggregation of binary classifier ensembles in a transductive setting, for a broad class of losses including but not limited to all convex surrogates. The result is a family of parameter-free ensemble aggregation algorithms, which are as efficient as linear learning and prediction for convex risk minimization but work without any relaxations whatsoever on many nonconvex losses like the 0-1 loss. The prediction algorithms take a familiar form, applying “link functions” to a generalized notion of ensemble margin, but without the assumptions typically made in margin-based learning - all this structure follows from a minimax interpretation of loss minimization. 

• [cs.LG]Multi-armed Bandits with Application to 5G Small Cells
Setareh Maghsudi, Ekram Hossain
//arxiv.org/abs/1510.00627v1 

Due to the pervasive demand for mobile services, next generation wireless networks are expected to be able to deliver high date rates while wireless resources become more and more scarce. This requires the next generation wireless networks to move towards new networking paradigms that are able to efficiently support resource-demanding applications such as personalized mobile services. Examples of such paradigms foreseen for the emerging fifth generation (5G) cellular networks include very densely deployed small cells and device-to-device communications. For 5G networks, it will be imperative to search for spectrum and energy-efficient solutions to the resource allocation problems that i) are amenable to distributed implementation, ii) are capable of dealing with uncertainty and lack of information, and iii) can cope with users' selfishness. The core objective of this article is to investigate and to establish the potential of multi-armed bandit (MAB) framework to address this challenge. In particular, we provide a brief tutorial on bandit problems, including different variations and solution approaches. Furthermore, we discuss recent applications as well as future research directions. In addition, we provide a detailed example of using an MAB model for energy-efficient small cell planning in 5G networks. 

• [cs.NE]An Asynchronous Implementation of the Limited Memory CMA-ES
Viktor Arkhipov, Maxim Buzdalov, Anatoly Shalyto
//arxiv.org/abs/1510.00419v1 

We present our asynchronous implementation of the LM-CMA-ES algorithm, which is a modern evolution strategy for solving complex large-scale continuous optimization problems. Our implementation brings the best results when the number of cores is relatively high and the computational complexity of the fitness function is also high. The experiments with benchmark functions show that it is able to overcome its origin on the Sphere function, reaches certain thresholds faster on the Rosenbrock and Ellipsoid function, and surprisingly performs much better than the original version on the Rastrigin function. 

• [math.ST]A Selective Approach to Internal Inference
Samuel M. Gross, Jonathan Taylor, Robert Tibshirani
//arxiv.org/abs/1510.00486v1 

A common goal in modern biostatistics is to form a biomarker signature from high dimensional gene expression data that is predictive of some outcome of interest. After learning this biomarker signature, an important question to answer is how well it predicts the response compared to classical predictors. This is challenging, because the biomarker signature is an internal predictor – one that has been learned using the same dataset on which we want to evaluate it’s significance. We propose a new method for approaching this problem based on the technique of selective inference. Simulations show that our method is able to properly control the level of the test, and that in certain settings we have more power than sample splitting. 

• [math.ST]Model Selection and Multiple Testing - A Bayesian and Empirical Bayes Overview and some New Results
Ritabrata Dutta, Malgortaza Bogdan, Jayanta K. Ghosh
//arxiv.org/abs/1510.00547v1 

We provide a brief overview of both Bayes and classical model selection. We argue tentatively that model selection has at least two major goals, that of finding the correct model or predicting well, and that in general both these goals may not be achieved in an optimum manner by a single model selection rule. We discuss, briefly but critically, through a study of well-known model selection rules like AIC, BIC, DIC and Lasso, how these different goals are pursued in each paradigm. We introduce some new definitions of consistency, results and conjectures about consistency in high dimensional model selection problems. Finally we discuss some new or recent results in Full Bayes and Empirical Bayes multiple testing, and cross-validation. We show that when the number of parameters tends to infinity at a smaller rate than sample size, then it is best from the point of view of consistency to use most of the data for inference and only a negligible proportion to make an improper prior proper. 

• [quant-ph]Autonomous Perceptron Neural Network Inspired from Quantum computing
M. Zidan, A. Sagheer, N. Metwally
//arxiv.org/abs/1510.00556v1 

Recently with the rapid development of technology, there are a lot of applications require to achieve low-cost learning in order to accomplish inexpensive computation. However the known computational power of classical artificial neural networks (CANN), they are not capable to provide low-cost learning due to many reasons such as linearity, complexity of architecture, etc. In contrast, quantum neural networks (QNN) may be representing a good computational alternate to CANN, based on the computational power of quantum bit (qubit) over the classical bit. In this paper, a new algorithm of quantum perceptron neural network based only on one neuron is introduced to overcome some limitations of the classical perceptron neural networks. The proposed algorithm is capable to construct its own set of activation operators that enough to accomplish the learning process in a limited number of iterations and, consequently, reduces the cost of computation. For evaluation purpose, we utilize the proposed algorithm to solve five problems using real and artificial data. It is shown throughout the paper that promising results are provided and compared favorably with other reported algorithms 

• [stat.AP]Bayesian modeling of networks in complex business intelligence problems
Daniele Durante, Sally Paganin, Bruno Scarpa, David B. Dunson
//arxiv.org/abs/1510.00646v1 

Complex network data problems are increasingly common in many fields of application. Our motivation is drawn from strategic marketing studies monitoring customer preferences for specific products, along with co-subscription networks encoding multi-buying behavior. Data are available for multiple agencies within the same insurance company, and our goal is to efficiently exploit co-subscription networks to inform targeted advertising of cross-selling campaigns to currently mono-product customers. We address this goal by developing a Bayesian hierarchical model, which groups agencies according to common customer preferences and co-subscription networks. Within each cluster, we efficiently model customer behaviors via a cluster-dependent mixture of latent eigenmodels. This formulation allows efficient targeting, while providing key information on mono- and multi-product buying behaviors within clusters, informing cross-selling marketing campaigns. We develop simple algorithms for tractable inference, and assess the performance in simulations and an application to business intelligence. 

• [stat.AP]Microwave Surveillance based on Ghost Imaging and Distributed Antennas
Xiaopeng Wang, Zihuai Lin
//arxiv.org/abs/1510.00474v1 

In this letter, we proposed a ghost imaging (GI) and distributed antennas based microwave surveillance scheme. By analyzing its imaging resolution and sampling requirement, the potential of employing microwave GI to achieve high-quality surveillance performance with low system complexity has been demonstrated. The theoretical analysis and effectiveness of the proposed microwave surveillance method are also validated via simulations. 

• [stat.CO]A Bayesian approach to constrained single- and multi-objective optimization
Paul Féliot, Julien Bect, Emmanuel Vazquez
//arxiv.org/abs/1510.00503v1 

This article addresses the problem of derivative-free (single- or multi-objective) optimization subject to multiple inequality constraints. Both the objective and constraint functions are assumed to be smooth, non-linear and expensive to evaluate. As a consequence, the number of evaluations that can be used to carry out the optimization is very limited, as in complex industrial design optimization problems. The method we propose to overcome this difficulty has its roots in the Bayesian and the multiobjective optimization literatures. More specifically, an extended domination rule is used to handle the constraints and a corresponding Bayesian expected hyper-volume improvement sampling criterion is proposed. This new criterion extends existing Bayesian sampling criteria to the multi-objective constrained case, and makes it possible to start the algorithm without an initial feasible point. The calculation and optimization of the criterion are performed using Sequential Monte Carlo techniques. In particular, an algorithm similar to the subset simulation method, which is well known in the field of structural reliability, is used to estimate the expected hyper-volume improvement criterion. The method, which we call BMOO (for Bayesian Multi-Objective Optimization), is compared to state-of-the-art algorithms for single-objective and multi-objective constrained optimization problems. 

• [stat.CO]On Estimation of Parameter Uncertainty in Model-Based Clustering
Adrian O'Hagan, Thomas Brendan Murphy, Isobel Claire Gormley
//arxiv.org/abs/1510.00551v1 

Mixture models are a popular tool in model-based clustering. Such a model is often fitted by a procedure that maximizes the likelihood, such as the EM algorithm. At convergence, the maximum likelihood parameter estimates are typically reported, but in most cases little emphasis is placed on the variability associated with these estimates. In part this may be due to the fact that standard errors are not directly calculated in the model-fitting algorithm, either because they are not required to fit the model, or because they are difficult to compute. The examination of standard errors in model-based clustering is therefore typically neglected. The widely used R package mclust has recently introduced bootstrap and weighted likelihood bootstrap methods to facilitate standard error estimation. This paper provides an empirical comparison of these methods (along with the jackknife method) for producing standard errors and confidence intervals for mixture parameters. These methods are illustrated and contrasted in both a simulation study and in the traditional Old Faithful data set. 

• [stat.ME]The zero-inflated promotion cure rate regression model applied to fraud propensity in bank loan applications
Francisco Louzada, Mauro R. de Oliveira Jr, Fernando F. Moreira
//arxiv.org/abs/1510.00443v1 

In this paper we extend the promotion cure rate model proposed by Chen et al (1999), by incorporating excess of zeros in the modelling. Despite allowing to relate the covariates to the fraction of cure, the current approach, which is based on a biological interpretation of the causes that trigger the event of interest, does not enable to relate the covariates to the fraction of zeros. The presence of zeros in survival data, unusual in medical studies, can frequently occur in banking loan portfolios, as presented in Louzada et al (2015), where they deal with propensity to fraud in lending loans in a major Brazilian bank. To illustrate the new cure rate survival method, the same real dataset analyzed in Louzada et al (2015) is fitted here, and the results are compared. 

• [stat.ML]Distributed Multitask Learning
Jialei Wang, Mladen Kolar, Nathan Srebro
//arxiv.org/abs/1510.00633v1 

We consider the problem of distributed multi-task learning, where each machine learns a separate, but related, task. Specifically, each machine learns a linear predictor in high-dimensional space,where all tasks share the same small support. We present a communication-efficient estimator based on the debiased lasso and show that it is comparable with the optimal centralized method.