检索结果-内蒙古大学图书馆

Interspeech Conference

作者： Springenberg, Sebastian Lakomkin, Egor Weber, Cornelius Wermter, Stefan Univ Hamburg Dept Informat Knowledge Technol Hamburg Germany

Unsupervised learning represents an important opportunity for obtaining useful speech representations. Recently, variational autoencoders (VAEs) have been shown to extract useful representations in an unsupervised manner. These models are usually not designed to explicitly disentangle specific sources of information. When processing data of sequential nature which involves multi-timescale information, disentanglement can however be beneficial. In this paper we address this issue by developing a predictive auxiliary variational autoencoder to obtain speech representations at different timescales. We will present an auxiliary lower bound which is used to develop a model that we call the predictive Aux-VAE. The model is designed to disentangle global from local information into a dedicated auxiliary variable. Learned representations are analysed with respect to their ability to capture global speech characteristics. We observe that representations of individual speakers are separated well in the latent space and can successfully be used in a subsequent speaker identification task where they achieve high classification accuracy, comparable to a fully supervised model. Moreover, manipulating the global variable allows to change global characteristics while retaining the local content during generation which demonstrates the success of our model to disentangle global from local information.

关键词： unsupervised learning speech representation learning predictive coding predictive auxiliary VAE speaker identification

来源：评论

学校读者我要写书评

暂无评论

Event-Based Re-configurable Hierarchical Processors for Smart Image Sensors 30

Event-Based Re-configurable Hierarchical Processors for Smar...

引用

30th IEEE International Conference on Application-Specific Systems, Architectures and Processors (ASAP)

作者： Bhowmik, Pankaj Pantho, Jubaer Hossain Bobda, Christophe Univ Florida Dept Elect & Comp Engn Gainesville FL 32611 USA

ISBN: (纸本)9781728116013

This paper presents a reconfigurable hierarchical hardware architecture at the pixel and region level for smart image sensors to accelerate machine vision applications. This architecture maintains hierarchical processing that begins at the pixel level. It reduces the computational burden on the sequential processor and accelerates the overall processing. There are three hierarchical layers, and each layer passes only the relevant information to the next layer by removing redundant information. This method facilitates the final processor to respond if there is an event in the scene. Identifying relevant information is complex, and we adapted different bio-inspired algorithms in the hierarchical layers to meet this purpose. Besides, these processors are made runtime reconfigurable to different applications, and it adds flexibility after fabrication. This hierarchical processing breaks the traditional sequential image processing and introduces parallelism for the machine vision applications. We evaluate the design in FPGA and achieve the GDSII file in ASIC platform at 800MHz. Simulation results show that the area overhead and power penalty for adding reconfiguration feature stay in an acceptable range. Besides, removing redundant information 84.01% and 96.91% dynamic power can be saved at the pixel-level and region-level, respectively.

关键词： Pixel Level Processing Reconfigurability predictive coding Attention Module Smart Image Sensor FPGA ASIC

来源：评论

学校读者我要写书评

暂无评论

An Investigation Of Factual And Counterfactual Feedback Information In Early Visual Cortex

An Investigation Of Factual And Counterfactual Feedback Info...

引用

作者： Angus T. Paton University of Glasgow

学位级别：博士

Primary visual cortex receives approximately 90% of the input to the retina, however this only accounts for around 5% of the input to V1 (Muckli, 2010). The majority of the input to V1 is in fact from other cortical and sub-cortical parts of the brain that arrive there via lateral and feedback pathways. It is therefore critical to our knowledge of visual perception to understand how these feedback responses influence visual processing. The aim of this thesis is to investigate different sources of non-visual feedback to early visual cortex. To do this we use a combination of an occlusion paradigm, derived from F. W. Smith and Muckli (2010), and functional magnetic resonance imagining. Occlusion offers us a method to inhibit the feedforward flow of information to the retina from a specific part of the visual field. By inhibiting the feedforward information we exploit the highly precise retinotopic organisation of visual cortex by rendering a corresponding patch of cortex free of feedforward input. From this isolated patch of cortex we can ask questions about the information content of purely feedback information. In Chapter 3 we investigated whether or not information about valance was present in non-stimulated early visual cortex. We constructed a 900 image set that contained an equal number of images for neutral, positive and negative valance across animal, food and plant categories. We used an m-sequence design to allow us to present image set within a standard period of time for fMRI. We were concerned about low-level image properties being a potential confound, so a large image set would allow us to average out these low-level properties. We occluded the lower-right quadrant of each image and presented each image only once to our subjects. The image set was rated for valance and arousal after fMRI so that individual subjectivity could be accounted for. We used multivariate pattern analysis (MVPA) to decode pairs of neutral, positive and negative valance. We found

关键词： predictive coding fMRI vision neuroscience V1 V2 V3 striate extrastriate cortex feedback affect valance audition virtual reality

来源：评论

学校读者我要写书评

暂无评论

Enhancing the predictive coding efficiency with control technologies for lossless compression of images

引用

IET IMAGE PROCESSING 2012年第3期6卷 251-263页

作者： Lee, C. -H. Kau, L. -J. Natl Chung Hsin Univ Dept Mech Engn Taichung 402 Taiwan Natl Taipei Univ Technol Dept Elect Engn Taipei 10608 Taiwan Natl Taipei Univ Technol Grad Inst Comp & Commun Engn Taipei 10608 Taiwan

This study applies techniques commonly used in control systems to enhance the efficiency of predictive coding in lossless compression of images for pixels around boundaries. Actually, the predictive coding system behaves just like a multi-input single-output system with the predictor itself can be regarded as the system model. Besides, the prediction error is usually feedback for the adaptation of predictor coefficients so that the prediction error of consecutive pixels can be minimised. When compared with a control system, which is to follow the system command as precisely as possible, the authors find the objective of both systems are the same. Moreover, a boundary among image pixels can be considered a step command in control systems. These observations lead to the idea of using control technologies to improve the prediction result around boundaries. To realise this idea, an adaptive Takagi-Sugeno fuzzy neural network and a proportional controller in control systems are applied as the predictor and the error compensator, respectively. To accelerate the run-time performance of the proposed system under limited resources, the online training area is even not used for network adaptation, but the performance is still comparable with state-of-the-art predictors and coders as the authors will see in the experiment.

关键词： adaptive Takagi-Sugeno fuzzy neural network predictor Neural computing techniques error compensator control technology Computer vision and image processing techniques Image and video coding proportional control image coding multiinput single-output system predictive coding lossless image compression proportional controller neural nets

来源：评论

学校读者我要写书评

暂无评论

Neuromorphic Image Sensor Design with Region-Aware Processing 18

Neuromorphic Image Sensor Design with Region-Aware Processin...

引用

18th IEEE-Computer-Society Annual Symposium on VLSI (ISVLSI)

作者： Pantho, Md Jubaer Hossain Bhowmik, Pankaj Bobda, Christophe Univ Florida Gainesville FL 32611 USA

ISBN: (纸本)9781728133911

This paper presents a pixel parallel architecture of a neuromorphic image sensor, designed as a 3D bottom-up architecture composing of several computational planes where each plane performs different image processing algorithms. The model emulates the hierarchical process in biological vision by providing feedforward and feedback information flow between different planes. The on-chip attention module dynamically detects regions with relevant information and produces a feedback path to sample those regions with a higher clock frequency, whereas regions with low spatial and temporal information receive less attention. The results suggest that by sampling non-relevant regions with a lower frequency, the sensor can reduce redundancy and enable high-performance computing at low power. Furthermore, by deploying high-level reasoning only on the selected regions instead of the entire image the model can decrease computational expenses.

关键词： Image Sensor Visual Attention Neuromorphic FPGA predictive coding

来源：评论

学校读者我要写书评

暂无评论

Investigating Sensory Prediction in Autism Spectrum Conditions

Investigating Sensory Prediction in Autism Spectrum Conditio...

引用

作者： Finnemann, Johanna J. S. University of Cambridge

学位级别：博士

Autism is the umbrella term for a family of neurodevelopmental disorders characterised by heterogeneous clinical presentations affecting social interaction, communication, sensory atypicalities and restricted and repetitive behaviours (DSM 5). The idea of a unifying explanation that can account for the range of cognitive, behavioural and neurological features associated with autism is attractive, but as of yet no such cognitive or physiological underpinnings have been identified. However, the last years have seen a growing interest in using approaches within the nascent field of the predictive processing framework to explore the potential causal role of aberrant prediction for the autistic phenotype. While hypothesised differences in predictive abilities have demonstrated some explanatory power for symptoms of psychotic spectrum disorders, empirical investigations into autism are still sparse. In this thesis I follow up on the theoretical work about difficulties with expectation generation in autism with three empirical studies on prediction in perception and sensory processing (Chapters 2-4). My results did not support the idea of autism as a generalised disorder of prediction; however better phenotyping in future work might help to tease apart some of the variability observed in the autism group. Furthermore I also examined the psychometric properties of two widely-used self-report questionnaires assessing autistic traits and schizotypy (Chapter 5). If latent traits are not measured equivalently across clinical and non-clinical populations, this could have implications for studies using high self-reported traits in healthy participants as proxies for the clinical condition as well as for correlational studies.

关键词： autism ASD Asperger's sensory motor prediction predictive coding predictive processing Bayesian Brain

来源：评论

学校读者我要写书评

暂无评论

AN AFFINE-LINEAR INTRA PREDICTION WITH COMPLEXITY CONSTRAINTS 26

AN AFFINE-LINEAR INTRA PREDICTION WITH COMPLEXITY CONSTRAINT...

引用

26th IEEE International Conference on Image Processing (ICIP)

作者： Schaefer, Michael Stallenberger, Bjoern Pfaff, Jonathan Helle, Philipp Schwarz, Heiko Marpe, Detlev Wiegand, Thomas Heinrich Hertz Inst Nachrichtentech Berlin GmbH Fraunhofer Inst Telecommun Video Coding & Analyt Dept Einsteinufer 37 D-10587 Berlin Germany

ISBN: (纸本)9781538662496

This paper presents a novel method for a data-driven training of affine-linear predictors which perform intra prediction in state-of-the-art video coding. The main aspect of our training design is the use of subband decomposition of both the input and the output of the prediction. Due to this architecture, the same set of predictors can be shared across different block shapes leading to a very limited memory requirement. Also, the computational complexity of the resulting predictors can be limited such that it does not exceed the complexity of the conventional angular intra prediction. In the training itself, a loss function modelling the bit-rate of the DCT-transformed residuals is used. The obtained predictors are incorporated into the Versatile Video coding Test Model 3 in addition to the conventional intra prediction modes. All-Intra bit-rate savings ranging from 0.8% to 1.4% across different resolutions have been measured in terms of the Bjontegaard-Delta bit rate (BD-rate).

关键词： High Efficiency Video coding (HEVC) Versatile Video coding (VVC) Intra Prediction predictive coding Machine Learning

来源：评论

学校读者我要写书评

暂无评论

Evidence from auditory and visual event-related potential (ERP) studies of deviance detection (MMN and vMMN) linking predictive coding theories and perceptual object representations

引用

INTERNATIONAL JOURNAL OF PSYCHOPHYSIOLOGY 2012年第2期83卷 132-143页

作者： Winkler, Istvan Czigler, Istvan Hungarian Acad Sci Inst Psychol H-1394 Budapest Hungary Univ Szeged Inst Psychol Szeged Hungary Eotvos Lorand Univ Inst Psychol Budapest Hungary

predictive coding theories posit that the perceptual system is structured as a hierarchically organized set of generative models with increasingly general models at higher levels. The difference between model predictions and the actual input (prediction error) drives model selection and adaptation processes minimizing the prediction error. Event-related brain potentials elicited by sensory deviance are thought to reflect the processing of prediction error at an intermediate level in the hierarchy. We review evidence from auditory and visual studies of deviance detection suggesting that the memory representations inferred from these studies meet the criteria set for perceptual object representations. Based on this evidence we then argue that these perceptual object representations are closely related to the generative models assumed by predictive coding theories. (C) 2011 Elsevier B.V. All rights reserved.

关键词： Perception predictive coding Deviance detection Object representation Event-related brain potentials (ERP) Mismatch negativity (MMN) Visual mismatch negativity (vMMN)

来源：评论

学校读者我要写书评

暂无评论

Autoregressive Image Generative Models with Normal and t-distributed Noise and the Bayes Codes for Them

Autoregressive Image Generative Models with Normal and t-dis...

引用

International Symposium on Information Theory and Its Applications, ISITA

作者： Yuta Nakahara Toshiyasu Matsushima Center for Data Science Waseda University Tokyo Japan Waseda University Tokyo Japan

ISBN: (数字)9784885523304

ISBN: (纸本)9781728128559

In this paper, we propose an autoregressive stochastic generative model for images. This model should be one of the most basic models for the new type of lossless image compression which explicitly assume the stochastic generative model. We can easily expand it and theoretically interpret the implicitly assumed stochastic generative models in the various previous predictive coding methods as the expanded versions of our model. Moreover, we can utilize the achievements in the related fields where the linear regression analysis and its expansion are studied to construct the Bayes codes for these generative models. As an example, we expand our generative model from the one with normal noise to the one with the t-distributed noise. Then, we construct the sub-optimal Bayes codes for this generative model by utilizing the variational Bayesian method.

关键词： Image coding Linear regression Stochastic processes predictive models predictive coding Bayes methods Information theory

来源：评论

学校读者我要写书评

暂无评论

The thing that should not be: predictive coding and the uncanny valley in perceiving human and humanoid robot actions

引用

SOCIAL COGNITIVE AND AFFECTIVE NEUROSCIENCE 2012年第4期7卷 413-422页

作者： Saygin, Ayse Pinar Chaminade, Thierry Ishiguro, Hiroshi Driver, Jon Frith, Chris Univ Calif San Diego Dept Cognit Sci La Jolla CA 92093 USA Univ Calif San Diego Neurosci Program La Jolla CA 92093 USA UCL Wellcome Trust Ctr Neuroimaging London England Aix Marseille Univ CNRS Mediterranean Inst Cognit Neurosci INCM Marseille France Osaka Univ Dept Syst Innovat Osaka Japan ATR Intelligent Robot & Commun Labs Keihanna Sci City Kyoto Japan Univ Aarhus DK-8000 Aarhus C Denmark

Using functional magnetic resonance imaging (fMRI) repetition suppression, we explored the selectivity of the human action perception system (APS), which consists of temporal, parietal and frontal areas, for the appearance and/or motion of the perceived agent. Participants watched body movements of a human (biological appearance and movement), a robot (mechanical appearance and movement) or an android (biological appearance, mechanical movement). With the exception of extrastriate body area, which showed more suppression for human like appearance, the APS was not selective for appearance or motion per se. Instead, distinctive responses were found to the mismatch between appearance and motion: whereas suppression effects for the human and robot were similar to each other, they were stronger for the android, notably in bilateral anterior intraparietal sulcus, a key node in the APS. These results could reflect increased prediction error as the brain negotiates an agent that appears human, but does not move biologically, and help explain the 'uncanny valley' phenomenon.

关键词： functional magnetic resonance imaging (fMRI) repetition suppression action perception predictive coding temporal cortex anterior intraparietal sulcus mirror neuron system extrastriate body area

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：