检索结果-内蒙古大学图书馆

Generalized Independent Component Analysis Over Finite Alphabets

IEEE TRANSACTIONS ON INFORMATION THEORY 2016年第2期62卷 1038-1053页

作者： Painsky, Amichai Rosset, Saharon Feder, Meir Tel Aviv Univ Dept Stat IL-6997801 Tel Aviv Israel Tel Aviv Univ Dept Elect Engn IL-6997801 Tel Aviv Israel

Independent component analysis (ICA) is a statistical method for transforming an observable multi-dimensional random vector into components that are as statistically independent as possible from each other. Usually, the ICA framework assumes a model according to which the observations are generated (such as a linear transformation with additive noise). ICA over finite fields is a special case of ICA in which both the observations and the independent components are over a finite alphabet. In this paper, we consider a generalization of this framework in which an observation vector is decomposed to its independent components (as much as possible) with no prior assumption on the way it was generated. This generalization is also known as Barlow's minimal redundancy representation problem and is considered an open problem. We propose several theorems and show that this hard problem can be accurately solved with a branch and bound search tree algorithm, or tightly approximated with a series of linear problems. Our contribution provides the first efficient set of solutions to Barlow's problem. The minimal redundancy representation (also known as factorial code) has many applications, mainly in the fields of neural networks and deep learning. The binary ICA is also shown to have applications in several domains, including medical diagnosis, multi-cluster assignment, network tomography, and internet resource management. In this paper, we show that this formulation further applies to multiple disciplines in source coding, such as predictive coding, distributed source coding, and coding of large alphabet sources.

关键词： Independent component analysis BICA ICA over galois field blind source separation minimal redundancy representation minimum entropy codes factorial codes predictive coding distributed source coding neural networks

来源：评论

学校读者我要写书评

暂无评论

The Curious Incident of Attention in Multisensory Integration: Bottom-up vs. Top-down

引用

MULTISENSORY RESEARCH 2016年第6-7期29卷 557-583页

作者： Macaluso, Emiliano Noppeney, Uta Talsma, Durk Vercillo, Tiziana Hartcher-O'Brien, Jess Adam, Ruth Fdn Santa Lucia Neuroimaging Lab Rome Italy Univ Birmingham Computat Neurosci & Cognit Robot Ctr Birmingham B15 2TT W Midlands England Univ Ghent Dept Expt Psychol Henri Dunantlaan 2 B-9000 Ghent Belgium Univ Nevada Dept Psychol Reno NV 89557 USA Univ Paris 06 Sorbonne Univ ISIR UMR 7222 F-75005 Paris France Univ Munich Klinikum Univ Munchen Inst Stroke & Dementia Res Munich Germany

The role attention plays in our experience of a coherent, multisensory world is still controversial. On the one hand, a subset of inputs may be selected for detailed processing and multisensory integration in a top-down manner, i.e., guidance of multisensory integration by attention. On the other hand, stimuli may be integrated in a bottom-up fashion according to low-level properties such as spatial coincidence, thereby capturing attention. Moreover, attention itself is multifaceted and can be described via both top-down and bottom-up mechanisms. Thus, the interaction between attention and multisensory integration is complex and situation-dependent. The authors of this opinion paper are researchers who have contributed to this discussion from behavioural, computational and neurophysiological perspectives. We posed a series of questions, the goal of which was to illustrate the interplay between bottom-up and top-down processes in various multisensory scenarios in order to clarify the standpoint taken by each author and with the hope of reaching a consensus. Although divergence of viewpoint emerges in the current responses, there is also considerable overlap: In general, it can be concluded that the amount of influence that attention exerts on MSI depends on the current task as well as prior knowledge and expectations of the observer. Moreover stimulus properties such as the reliability and salience also determine how open the processing is to influences of attention.

关键词： Bayesian causal inference endogenous predictive coding salience stimulus-driven

来源：评论

学校读者我要写书评

暂无评论

Empirical evaluations of active learning strategies in legal document review

Empirical evaluations of active learning strategies in legal...

引用

IEEE International Conference on Big Data

作者： Rishi Chhatwal Nathaniel Huber-Fliflet Robert Keeling Jianping Zhang Haozhen Zhao Esq. Legal AT&T Services Inc. Washington D.C. USA Legal Technology Solutions Navigant Consulting Inc. Washington D.C. USA Complex Commercial Litigation Sidley Austin LLP Washington D.C. USA

One type of machine learning, text classification, is now regularly applied in the legal matters involving voluminous document populations because it can reduce the time and expense associated with the review of those documents. One form of machine learning - Active Learning - has drawn attention from the legal community because it offers the potential to make the machine learning process even more effective. Active Learning, applied to legal documents, is considered a new technology in the legal domain and is continuously applied to all documents in a legal matter until an insignificant number of relevant documents are left for review. This implementation is slightly different than traditional implementations of Active Learning where the process stops once achieving acceptable model performance. The purpose of this paper is twofold: (i) to question whether Active Learning actually is a superior learning methodology and (ii) to highlight the ways that Active Learning can be most effectively applied to real legal industry data. Unlike other studies, our experiments were performed against large data sets taken from recent, real-world legal matters covering a variety of areas. We conclude that, although these experiments show the Active Learning strategy popularly used in legal document review can quickly identify informative training documents, it becomes less effective over time. In particular, our findings suggest this most popular form of Active Learning in the legal arena, where the highest-scoring documents are selected as training examples, is in fact not the most efficient approach in most instances. Ultimately, a different Active Learning strategy may be best suited to initiate the predictive modeling process but not to continue through the entire document review.

关键词： Law Training Electronic mail predictive coding predictive models Text categorization

来源：评论

学校读者我要写书评

暂无评论

Compression of multiple user photo galleries

引用

IMAGE AND VISION COMPUTING 2016年第0期53卷 68-75页

作者： Milani, Simone Univ Padua Dept Informat Engn Via Gradenigo 6-B I-35131 Padua Italy

The possibility of sharing multimedia contents in easy and ubiquitous way has brought to the creation of multiuser photo albums. Pictures and video sequences taken by different people attending common social events (e.g., concerts and sport competitions) are gathered together into huge sets of heterogeneous multimedia data. These databases require effective compression strategies that exploit the common visual information related to the scene but compensate effectively the differences depending on the acquiring viewpoints, camera models, and acquisition time instants. The paper presents a predictive coding strategy for multi-user photo gallery, which initially localizes each picture in terms of viewpoint, orientation, time, and acquired elements. This information permits ordering all the images in a prediction tree and associates to each of them a reference picture. From this structure, it is possible to build a predictive coding strategy that exploits the redundant elements between the image to be coded and its reference. Experimental results show an average bit rate reduction up to 75% with respect to HEVC Intra low complexity coding. (C) 2016 Elsevier B.V. All rights reserved.

关键词： Image coding Photo collections Geotagging Synchronization SIFT predictive coding

来源：评论

学校读者我要写书评

暂无评论

Dynamic causal modeling of touch-evoked potentials in the rubber hand illusion

引用

NEUROIMAGE 2016年 138卷 266-273页

作者： Zeller, Daniel Friston, Karl J. Classen, Joseph Univ Wurzburg Dept Neurol Josef Schneider Str 11 D-97080 Wurzburg Germany UCL Inst Neurol Wellcome Trust Ctr Neuroimaging London WC1N 3BG England Univ Leipzig Dept Neurol D-04103 Leipzig Germany

The neural substrate of bodily ownership can be disclosed by the rubber hand illusion (RHI);namely, the illusory self-attribution of an artificial hand that is induced by synchronous tactile stimulation of the subject's hand that is hidden from view. Previous studies have pointed to the premotor cortex (PMC) as a pivotal area in such illusions. To investigate the effective connectivity between - and within - sensory and premotor areas involved in bodily perceptions, we used dynamic causal modeling of touch-evoked responses in 13 healthy subjects. Each subject's right hand was stroked while viewing their own hand ("REAL"), or an artificial hand presented in an anatomically plausible ("CONGRUENT") or implausible ("INCONGRUENT") position. Bayesian model comparison revealed strong evidence for a differential involvement of the PMC in the generation of touch-evoked responses under the three conditions, confirming a crucial role of PMC in bodily self-attribution. In brief, the extrinsic (forward) connection from left occipital cortex to left PMC was stronger for CONGRUENT and INCONGRUENT as compared to REAL, reflecting the augmentation of bottom-up visual input when multisensory integration is challenged. Crucially, intrinsic connectivity in the primary somatosensory cortex (S1) was attenuated in the CONGRUENT condition, during the illusory percept. These findings support predictive coding models of the functional architecture of multisensory integration (and attenuation) in bodily perceptual experience. (C) 2016 Elsevier Inc. All rights reserved.

关键词： Bodily ownership Multisensory integration predictive coding Premotor cortex

来源：评论

学校读者我要写书评

暂无评论

The dysconnection hypothesis (2016)

引用

SCHIZOPHRENIA RESEARCH 2016年第2-3期176卷 83-94页

作者： Friston, Karl Brown, Harriet R. Siemerkus, Jakob Stephan, Klaas E. UCL Wellcome Trust Ctr Neuroimaging Inst Neurol London WC1N 3BG England Univ Oxford Oxford Ctr Human Brain Act Oxford OX1 2JD England Univ Zurich Translat Neuromodeling Unit Inst Biomed Engn CH-8006 Zurich Switzerland Swiss Fed Inst Technol Zurich Switzerland Dept Psychiat Psychotherapy & Psychosomat Zurich Switzerland

Twenty years have passed since the dysconnection hypothesis was first proposed (Friston and Frith, 1995;Weinberger, 1993). In that time, neuroscience has witnessed tremendous advances: we now live in a world of non-invasive neuroanatomy, computational neuroimaging and the Bayesian brain. The genomics era has come and gone. Connectomics and large-scale neuroinformatics initiatives are emerging everywhere. So where is the dysconnection hypothesis now? This article considers how the notion of schizophrenia as a dysconnection syndrome has developed - and how it has been enriched by recent advances in clinical neuroscience. In particular, we examine the dysconnection hypothesis in the context of (i) theoretical neurobiology and computational psychiatry;(ii) the empirical insights afforded by neuroimaging and associated connectomics - and (iii) how bottom-up (molecular biology and genetics) and top-down (systems biology) perspectives are converging on the mechanisms and nature of dysconnections in schizophrenia. (C) 2016 The Authors. Published by Elsevier B.V.

关键词： Schizophrenia Dysconnection Neuromodulation Bayesian predictive coding Neurogenetics

来源：评论

学校读者我要写书评

暂无评论

Cardio-visual integration modulates the subjective perception of affectively neutral stimuli

引用

INTERNATIONAL JOURNAL OF PSYCHOPHYSIOLOGY 2016年第0期99卷 10-17页

作者： Azevedo, Ruben T. Ainley, Vivien Tsakiris, Manos Royal Holloway Univ London Dept Psychol Lab Act & Body London England

Interoception, which refers to the perception of internal body signals, has been consistently associated with emotional processing and with the sense of self. However, its influence on the subjective appraisal of affectively neutral and body -unrelated stimuli is still largely unknown. Across two experiments we sought to investigate this issue by asking participants to detect changes in the flashing rhythm of a simple stimulus (a circle) that could either be pulsing synchronously with their own heartbeats or following the pattern of another person's heart. While overall task performance did not vary as a function of cardio-visual synchrony, participants were better at identifying trials in which no change occurred when the flashes were synchronous with their own heartbeats. This study adds to the growing body of research indicating that we use our body as a reference point when perceiving the world;and extends this view by focusing on the role that signals coming from inside the body, such as heartbeats, may play in this referencing process. Specifically we show that private interoceptive sensations can be combined with affectively neutral information unrelated to the self to influence the processing of a multisensory percept. Results are discussed in terms of both standard multisensory integration processes and predictive coding theories. (C) 2015 Elsevier B.V. All rights reserved.

关键词： Interoception Multisensory integration Heartbeat Neutral stimuli predictive coding

来源：评论

学校读者我要写书评

暂无评论

predictive compression of animated 3D models by optimized weighted blending of key-frames

引用

COMPUTER ANIMATION AND VIRTUAL WORLDS 2016年第6期27卷 556-576页

作者： Hajizadeh, Mohammadali Ebrahimnezhad, Hossein Sahand Univ Technol Dept Elect Engn Comp Vis Res Lab Tabriz Iran

Efficient compression techniques are required for animated mesh sequences with fixed connectivity and time-varying geometry. In this paper, we propose a key-frame-based technique for three-dimensional dynamic mesh compression. First, key-frames are extracted from the animated sequence. Extracted key-frames are then linearly combined using blending weights to predict the vertex locations of the other frames. These blending weights play a key role in the proposed algorithm because the prediction performance and the required number of key-frames greatly depend on these weights. We present a novel method in order to compute the optimum blending weight that makes it possible to predict location of the vertices of the non-key frames with the minimum number of key-frames. The residual prediction errors are finally quantized and encoded using Huffman coding and another heuristic method. Experimental results on different test sequences with various sizes, topologies, and geometries demonstrate the privileged performance of the proposed method compared with the previous techniques. Copyright (c) 2015 John Wiley & Sons, Ltd.

关键词： animated mesh compression key-frame (key-pose) extraction predictive coding optimized blending weights dynamic geometry 3D dynamic mesh

来源：评论

学校读者我要写书评

暂无评论

Encoding of Stimulus Probability in Macaque Inferior Temporal Cortex

引用

CURRENT BIOLOGY 2016年第17期26卷 2280-2290页

作者： Bell, Andrew H. Summerfield, Christopher Morin, Elyse L. Malecek, Nicholas J. Ungerleider, Leslie G. NIMH Lab Brain & Cognit 9000 Rockville Pike Bethesda MD 20892 USA Univ Oxford Dept Expt Psychol S Parks Rd Oxford OX1 3UD England MRC Cognit & Brain Sci Unit 15 Chaucer Rd Cambridge CB2 7EF England

Optimal perceptual decisions require sensory signals to be combined with prior information about stimulus probability. Although several theories propose that probabilistic information about stimulus occurrence is encoded in sensory cortex, evidence from neuronal recordings has not yet fully supported this view. We recorded activity from single neurons in inferior temporal cortex (IT) while monkeys performed a task that involved discriminating degraded images of faces and fruit. The relative probability of the cue being a face versus a fruit was manipulated by a latent variable that was not revealed to the monkeys and that changed unpredictably over the course of each recording session. In addition to responding to stimulus identity (face or fruit), population responses in IT encoded the long-term stimulus probability of whether a face or a fruit stimulus was more likely to occur. Face-responsive neurons showed reduced firing rates to expected faces, an effect consistent with "expectation suppression," but expected stimuli were decoded from multivariate population signals with greater accuracy. These findings support "predictive coding" theories, whereby neural signals in the mammalian visual system actively encode and update predictions about the local sensory environment.

关键词： decision making expectation suppression monkey neuron predictive coding

来源：评论

学校读者我要写书评

暂无评论

Reduced frontal theta oscillations indicate altered crossmodal prediction error processing in schizophrenia

引用

JOURNAL OF NEUROPHYSIOLOGY 2016年第3期116卷 1396-1407页

作者： Romero, Yadira Roa Keil, Julian Balz, Johanna Gallinat, Juergen Senkowski, Daniel St Hedwig Hosp Charite Univ Med Berlin Hosp Dept Psychiat & Psychotherapy Berlin Germany Univ Med Ctr Hamburg Eppendorf Dept Psychiat & Psychotherapy Hamburg Germany

Our brain generates predictions about forthcoming stimuli and compares predicted with incoming input. Failures in predicting events might contribute to hallucinations and delusions in schizophrenia (SZ). When a stimulus violates prediction, neural activity that reflects prediction error (PE) processing is found. While PE processing deficits have been reported in unisensory paradigms, it is unknown whether SZ patients (SZP) show altered crossmodal PE processing. We measured high-density electroencephalography and applied source estimation approaches to investigate crossmodal PE processing generated by audiovisual speech. In SZP and healthy control participants (HC), we used an established paradigm in which high- and low-predictive visual syllables were paired with congruent or incongruent auditory syllables. We examined crossmodal PE processing in SZP and HC by comparing differences in event-related potentials and neural oscillations between incongruent and congruent high-and low-predictive audiovisual syllables. In both groups event-related potentials between 206 and 250 ms were larger in high-compared with low-predictive syllables, suggesting intact audiovisual incongruence detection in the auditory cortex of SZP. The analysis of oscillatory responses revealed theta-band (4-7 Hz) power enhancement in high-compared with low-predictive syllables between 230 and 370 ms in the frontal cortex of HC but not SZP. Thus aberrant frontal theta-band oscillations reflect crossmodal PE processing deficits in SZ. The present study suggests a top-down multisensory processing deficit and highlights the role of dysfunctional frontal oscillations for the SZ psychopathology.

关键词： predictive coding audiovisual speech neural synchrony multisensory processing oscillatory activity

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：