检索结果-内蒙古大学图书馆

Proceedings of the 4th International Conference on neural Information Processing systems

作者： Tim Horiuchi John Lazzaro Andrew Moore Christof Koch Computation and Neural Systems Program California Institute of Technology Pasadena CA Department of Computer Science California Institute of Technology Pasadena CA

ISBN: (纸本)9781558601840

Inspired by a visual motion detection model for the rabbit retina and by a computational architecture used for early audition in the barn owl, we have designed a chip that employs a correlation model to report the one-dimensional field motion of a scene in real time. Using subthreshold analog VLSI techniques, we have fabricated and successfully tested a 8000 transistor chip using a standard MOSIS process.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Modeling structural plasticity in the barn owl auditory localization system with a spike-time dependent Hebbian learning rule

Modeling structural plasticity in the barn owl auditory loca...

引用

International Joint Conference on neural Networks (IJCNN)

作者： S.P. Mysore S.R. Quartz Control and Dynamical Systems Program California Institute of Technology Pasadena CA USA Division of the Humanities and Social Sciences Computation and Neural Systems Program California Institute of Technology Pasadena CA USA

Auditory localization behavior in barn owls is mediated by the integration of topographically encoded visual and auditory space maps. In juvenile owls, disruption of the audio visual map alignment by exposure to spectacles that laterally shift the visual input results in behavioral adaptation over the course of several weeks. It has been reported in literature that this adaptation is produced by architectural plasticity in the neural circuits encoding the space maps. It is known that this plasticity is guided by visual input in a topographic manner, and that the error signal is embedded in the firing dynamics of neurons in the inferior colliculus. In this work, we use leaky integrate-and-fire neurons to model the key elements in the auditory localization circuit of barn owls. We demonstrate that a Hebbian spike time dependent learning rule, coupled with an activity-dependent mechanism that promotes growth, can account for the essentials of circuit level plasticity associated with prism experience. We point out the importance of inhibition in both the normal functioning of this circuit, and prism induced plasticity, and comment on potential mechanisms for activity induced growth.

关键词： OWL Hebbian theory Neurons Circuits Optical sensors Space technology Electronic mail Biology computing Biological system modeling Predictive models

来源：评论

学校读者我要写书评

暂无评论

A CMOS imager with focal-plane computation for feature detection

A CMOS imager with focal-plane computation for feature detec...

引用

IEEE International Symposium on Circuits and systems (ISCAS)

作者： A. Pesavento C. Koch Department of Electrical Engineering California Institute of Technology Pasadena CA Computation and Neural Systems Program California Institute of Technology Pasadena CA USA

ISBN: (纸本)0780366859

Feature detection and tracking is a fundamental problem in computer vision research. By detecting and tracking features in an image sequence it is possible to recover information about both the motion of the viewer and the structure of the environment. We designed, fabricated and tested a CMOS imager with analog VLSI focal-plane computation for feature detection. The chip implements a feature detection algorithm that is suitable for integration in a compact analog VLSI chip. We review the algorithm, its analog VLSI implementation and experimental results from the chip.

关键词： Computer vision Cameras Very large scale integration Eigenvalues and eigenfunctions CMOS technology Image sequences Streaming media Detection algorithms Brightness Motion detection

来源：评论

学校读者我要写书评

暂无评论

Automated event detection in underwater video

Automated event detection in underwater video

引用

OCEANS

作者： D.R. Edgington K.A. Salamy M. Risi R.E. Sherlock D. Walther Christof Koch Monterey Bay Aquarium Research Institute Moss Landing CA USA Computation and Neural Systems Program California Institute of Technology Pasadena CA USA

We present an attentional selection system for processing video streams from remotely operated underwater vehicles (ROVs). The system identifies potentially interesting visual events spanning multiple frames based on low-level spatial properties of salient tokens, which are associated with those events and tracked over time. If video frames contain interesting events, they are labeled "interesting", otherwise they are labeled "boring". By marking the interesting events and omitting "boring" frames in the output stream, we augment the productivity of human video annotators, or, alternatively, provide input for a subsequent object classification algorithm.

关键词： Event detection Remotely operated vehicles Animals Cameras Layout Streaming media Underwater tracking Productivity Humans Object detection

来源：评论

学校读者我要写书评

暂无评论

Feature detection in analog VLSI

Feature detection in analog VLSI

引用

Asilomar Conference on Signals, systems & Computers

作者： A. Pesavento C. Koch Department of Electrical Engineering California Institute of Technology Pasadena CA USA Computation and Neural Systems Program California Institute of Technology Pasadena CA USA

Feature detection and tracking is a fundamental problem in computer vision research. By detecting and tracking features in an image sequence it is possible to recover information on both the motion of the viewer and the structure of the environment. We designed and tested a CMOS imager with analog VLSI focal plane computation for feature detection. The chip implements a feature detection algorithm that is suitable for integration in a compact analog VLSI chip. We review the algorithm, its analog VLSI implementation and results from the chip.

关键词： Computer vision Very large scale integration Eigenvalues and eigenfunctions Brightness Artificial intelligence Motion detection Analog computers Detection algorithms Frequency Ear

来源：评论

学校读者我要写书评

暂无评论

Methods and circuits for focal-plane computation of features in CMOS visual sensors

Methods and circuits for focal-plane computation of features...

引用

Conference on Advanced Research in VLSI

ISBN: (纸本)0769510388

Feature detection, and tracking is a fundamental problem in computer vision research. By detecting and tracking features in an image sequence it is possible to recover information about both the motion of the viewer and the structure of the environment. The selection of features is a computationally intensive task. We derive two low-complexity algorithms that are suitable for integration in a CMOS sensor with focal-plane processing. We review the two algorithms and the circuits that implement them. We presents results from accurate simulations and experimental results from fabricated CMOS sensors.

关键词： Sensor phenomena and characterization Computer vision CMOS image sensors CMOS technology Cameras Sensor systems CMOS process Circuit simulation Image sequences Streaming media

来源：评论

学校读者我要写书评

暂无评论

Dual-coding representations for robot vision programming in Tekkotsu

Dual-coding representations for robot vision programming in ...

引用

作者： Touretzky, David S. Halelamien, Neil S. Tira-Thompson, Ethan J. Wales, Jordan J. Usui, Kei Computer Science Department Carnegie Mellon University Pittsburgh PA 15213 United States Computation and Neural Systems Program California Institute of Technology California United States Robotics Institute Carnegie Mellon University Pittsburgh PA 15213 United States P. O. Box 821650 Vancouver WA 98682 Canada

We describe complementary iconic and symbolic representations for parsing the visual world. The iconic pixmap representation is operated on by an extensible set of "visual routines" (Ullman, 1984;Forbus et al., 2001). A symbolic representation, in terms of lines, ellipses, blobs, etc., is extracted from the iconic encoding, manipulated algebraically, and re-rendered iconically. The two representations are therefore duals, and iconic operations can be freely intermixed with symbolic ones. The dual-coding approach offers robot programmers a versatile collection of primitives from which to construct application-specific vision software. We describe some sample applications implemented on the Sony AIBO. © Springer Science+Business Media, LLC 2007.

关键词： Computer vision

来源：评论

学校读者我要写书评

暂无评论

Generalized smoothing networks in early vision

Generalized smoothing networks in early vision

引用

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： S.-C. Liu J.G. Harris Rockwell Science Center Inc. Thousand Oaks CA USA California Institute of Technology 216-76 Computation and Neural Systems Program Pasadena CA USA

Generalized smoothing networks have been developed which enforce smoothness constraints for any arbitrary level of derivative of the input data. Furthermore, discontinuities of any order of derivative can be detected by providing for continuous line processes, which selectively inhibit smoothing. Second- and higher-order networks are required for many problems in early vision; first-order networks are often unsatisfactory. Examples in surface interpolation, edge detection, and image segmentation are shown. Solution of these types of problems typically takes a prohibitive amount of time, even on supercomputers. A significant advantage of these proposed networks is that they can be mapped directly to analog VLSI hardware.< >

关键词： Smoothing methods Intelligent networks Image edge detection Image segmentation Very large scale integration Interpolation Voltage Supercomputers neural network hardware Computer vision

来源：评论

学校读者我要写书评

暂无评论

Fast learning with predictive forward models 91

Fast learning with predictive forward models

引用

Proceedings of the 5th International Conference on neural Information Processing systems

作者： Carlos Brody Computation and Neural Systems Program California Institute of Technology Pasadena CA and Dept. of Computer Science IIMAS UNAM México D.F. México

ISBN: (纸本)9781558602229

A method for transforming performance evaluation signals distal both in space and time into proximal signals usable by supervised learning algorithms, presented in [Jordan & Jacobs 90], is examined. A simple observation concerning differentiation through models trained with redundant inputs (as one of their networks is) explains a weakness in the original architecture and suggests a modification: an internal world model that encodes action-space exploration and, crucially, cancels input redundancy to the forward model is added. Learning time on an example task, cartpole balancing, is thereby reduced about 50 to 100 times.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Feedforward architectures driven by inhibitory interactions

arXiv

引用

arXiv 2017年

作者： Billeh, Yazan N. Schaub, Michael T. Computation and Neural Systems Program California Institute of Technology Université Catholique de Louvain Massachusetts Institute of Technology Institute for Data Systems and Society Massachusetts Institute of Technology Department of Engineering Science University of Oxford

Directed information transmission is paramount for many social, physical, and biological sys- tems. For neural systems, scientists have studied this problem under the paradigm of feedforward networks for decades. In most models of feedforward networks, activity is exclusively driven by excitatory neurons and the wiring patterns between them, while inhibitory neu- rons play only a stabilizing role for the network dy- namics. Motivated by recent experimental discoveries of hippocampal circuitry, cortical circuitry, and the di- versity of inhibitory neurons throughout the brain, here we illustrate that one can construct such networks even if the connectivity between the excitatory units in the system remains random. This is achieved by endowing inhibitory nodes with a more active role in the network. Our findings demonstrate that apparent feedforward activity can be caused by a much broader network- architectural basis than often assumed. Copyright © 2017, The Authors. All rights reserved.

关键词： Brain

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：