| 000 | 00000cam u2200205 a 4500 | |
| 001 | 000045975744 | |
| 005 | 20190315103755 | |
| 008 | 190314s1999 maua b 000 0 eng d | |
| 010 | ▼a 98038511 | |
| 020 | ▼a 0262133504 (hc. : alk. paper) | |
| 035 | ▼a (KERIS)REF000005066178 | |
| 040 | ▼a DLC ▼c DLC ▼d DLC ▼d 211009 | |
| 050 | 0 0 | ▼a QA76.87 ▼b .P85 1999 |
| 082 | 0 0 | ▼a 006.3/2 ▼2 23 |
| 084 | ▼a 006.32 ▼2 DDCK | |
| 090 | ▼a 006.32 ▼b P9822 | |
| 245 | 0 0 | ▼a Pulsed neural networks / ▼c edited by Wolfgang Maass, Christopher M. Bishop. |
| 260 | ▼a Cambridge, Mass. : ▼b MIT Press, ▼c c1999. | |
| 300 | ▼a xxix, 377 p. : ▼b ill. ; ▼c 26 cm. | |
| 500 | ▼a "A Bradford book." | |
| 504 | ▼a Includes bibliographical references. | |
| 650 | 0 | ▼a Neural networks (Computer science). |
| 700 | 1 | ▼a Maass, Wolfgang, ▼d 1949 August 21-. |
| 700 | 1 | ▼a Bishop, Christopher M. |
| 945 | ▼a KLPA |
소장정보
| No. | 소장처 | 청구기호 | 등록번호 | 도서상태 | 반납예정일 | 예약 | 서비스 |
|---|---|---|---|---|---|---|---|
| No. 1 | 소장처 과학도서관/Sci-Info(2층서고)/ | 청구기호 006.32 P9822 | 등록번호 121248276 (3회 대출) | 도서상태 대출가능 | 반납예정일 | 예약 | 서비스 |
컨텐츠정보
책소개
Most practical applications of artificial neural networks are based on a computational model involving the propagation of continuous variables from one processing unit to the next. In recent years, data from neurobiological experiments have made it increasingly clear that biological neural networks, which communicate through pulses, use the timing of the pulses to transmit information and perform computation. This realization has stimulated significant research on pulsed neural networks, including theoretical analyses and model development, neurobiological modeling, and hardware implementation.
This book presents the complete spectrum of current research in pulsed neural networks and includes the most important work from many of the key scientists in the field. Terrence J. Sejnowski's foreword, "Neural Pulse Coding," presents an overview of the topic. The first half of the book consists of longer tutorial articles spanning neurobiology, theory, algorithms, and hardware. The second half contains a larger number of shorter research chapters that present more advanced concepts. The contributors use consistent notation and terminology throughout the book.
Contributors:
Peter S. Burge, Stephen R. Deiss, Rodney J. Douglas, John G. Elias, Wulfram Gerstner, Alister Hamilton, David Horn, Axel Jahnke, Richard Kempter, Wolfgang Maass, Alessandro Mortara, Alan F. Murray, David P. M. Northmore, Irit Opher, Kostas A. Papathanasiou, Michael Recce, Barry J. P. Rising, Ulrich Roth, Tim Schonauer, Terrence J. Sejnowski, John Shawe-Taylor, Max R. van Daalen, J. Leo van Hemmen, Philippe Venier, Hermann Wagner, Adrian M. Whatley, Anthony M. Zador.
정보제공 :
저자소개
크리스토퍼 비숍(엮은이)
마이크로소프트 리서치 케임브리지의 부 디렉터이자 에든버러 대학교 컴퓨터 공학과의 학과장을 맡고 있다. 또한, 케임브리지 다윈 칼리지와 왕립 공학회의 펠로우이기도 하다. 크리스는 양자론에 관한 논문으로 세인트 캐서린 대학과 옥스퍼드 대학교에서 물리학 학사, 에든버러 대학교에서 이론 물리학 박사 학위를 취득했다.
Wolfgang Maass(엮은이)
목차
Section Section Description Page Number
Foreword
Neural Pulse Coding
Spike Timing
Population Codes
Hippocampal Place Field
Hardware Models
References
Preface
The Isaac Newton Institute
Overview of the Book
Acknowledgments
Contributors
Part I Basic Concepts and Models
1 Spiking Neurons
1.1 The Problem of Neural Coding
1.1.1 Motivation
1.1.2 Rate Codes
1.1.2.1 Rate as a Spike Count (Average over Time)
1.1.2.2 Rate as a Spike Density (Average over Several Runs)
1.1.2.3 Rate as Population Activity (Average over Several Neurons)
1.1.3 Candidate Pulse Codes
1.1.3.1 Time-to-First-Spike
1.1.3.2 Phase
1.1.3.3 Correlations and Synchrony
1.1.3.4 Stimulus Reconstruction and Reverse Correlation
1.1.4 Discussion: Spikes or Rates?
1.2 Neuron Models
1.2.1 Simple Spiking Neuron Model
1.2.2 First Steps towards Coding by Spikes
1.2.3 Threshold-Fire Models
1.2.3.1 Spike Response Model -- Further Details
1.2.3.2 Integrate-and-Fire Model
1.2.3.3 Models of Noise
1.2.4 Conductance-Based Models
1.2.4.1 Hodgkin-Huxley Model
1.2.4.2 Relation to the Spike Response Model
1.2.4.3 Compartmental Models
1.2.5 Rate Models
1.3 Conclusions
References
2 Computing with Spiking Neurons
2.1 Introduction
2.2 A Formal Computational Model for a Network of Spiking Neurons
2.3 McCulloch-Pitts Neurons versus Spiking Neurons
2.4 Computing with Temporal Patterns
2.4.1 Conincidence Detection
2.4.2 RBF-Units in the Temporal Domain
2.4.3 Computing a Weighted Sum in Temporal Coding
2.4.4 Universal Approximation of Continuous Functions with Spiking Neurons Remarks:
2.4.5 Other Computations with Temporal Patterns in Networks of Spiking Neurons
2.5 Computing with a Space-Rate Code
2.6 Computing with Firing Rates
2.7 Computing with Firing Rates and Temporal Correlations
2.8 Networks of Spiking Neurons for Storing and Retrieving Information
2.9 Computing on Spike Trains
2.10 Conclusions
References
3 Pulse-Based Computation in VLSI Neural Networks
3.1 Background
3.2 Pulsed Coding: A VLSI Perspective
3.2.1 Pulse Amplitude Modulation
3.2.2 Pulse Width Modulation
3.2.3 Pulse Frequency Modulation
3.2.4 Phase or Delay Modulation
3.2.5 Noise, Robustness, Accuracy and Speed
3.3 A MOSFET Introduction
3.3.1 Subthreshold Circuits for Neural Networks
3.4 Pulse Generation in VLSI
3.4.1 Pulse Intercommunication
3.5 Pulsed Arithmetic in VLSI
3.5.1 Addition of Pulse Stream Signals
3.5.2 Multiplication of Pulse Stream Signals
3.5.3 MOS Transconductance Multiplier
3.5.4 MOSFET Analog Multiplier
3.6 Learning in Pulsed Systems
3.7 Summary and Issues Raised
References
4 Encoding Information in Neuronal Activity
4.1 Introduction
4.2 Synchronization and Oscillations
4.3 Temporal Binding
4.4 Phase Coding
4.5 Dynamic Range and Firing Rate Codes
4.6 Interspike Interval Variability
4.7 Synapses and Rate Coding
4.8 Summary and Implications
References
Part II Implementations
5 Building Silicon Nervous Systems with Dendritic Tree Neuromorphs
5.1 Introduction
5.1.1 Why Spikes?
5.1.2 Dendritic Processing of Spikes
5.1.3 Tunability
5.2 Implementation in VLSI
5.2.1 Artificial Dendrites
5.2.2 Synapses
5.2.3 Dendritic Non-Linearities
5.2.4 Spike-Generating Soma
5.2.5 Excitability Control
5.2.6 Spike Distribution -- Virtual Wires
5.3 Neuromorphs in Action
5.3.1 Feedback to Threshold-Setting Synapses
5.3.2 Discrimination of Complex Spatio-Temporal Patterns
5.3.3 Processing of Temporally Encoded Information
5.4 Conclusions
Acknowledgments
References
6 A Pulse-Coded Communications Infrastructure for Neuromorphic Systems
6.1 Introduction
6.2 Neuromorphic Computational Nodes
6.3 Neuromorphic aVLSI Neurons
6.4 Address Event Representation (AER)
6.5 Implementations of AER
6.6 Silicon Cortex
6.6.1 Basic Layout
6.7 Functional Tests of Silicon Cortex
6.7.1 An Example Neuronal Network
6.7.2 An Example of Sensory Input to SCX
6.8 Future Research on AER Neuromorphic Systems
Acknowledgements
References
7 Analog VLSI Pulsed Networks for Perceptive Processing
7.1 Introduction
7.2 Analog Perceptive Nets Communication Requirements
7.2.1 Coding Information with Pulses
7.2.2 Multiplexing of the Signals Issued by Each Neuron
7.2.3 Non-Arbitered PFM Communication
7.3 Analysis of the NAPFM Communication Systems
7.3.1 Statistical Assumptions
7.3.2 Detection
7.3.2.1 Detection by Time-Windowing
7.3.2.2 Direct Interpulse Time Measurement
7.3.3 Performance
7.3.3.1 Detection by Time-Windowing
7.3.3.2 Direct Interpulse Time Measurement
7.3.4 Data Dependency of System Performance
7.3.5 Discussion
7.3.5.1 Detection by Time-Windowing
7.3.5.2 Detection by Direct Interpulse Time Measurement
7.4 Address Coding
7.5 Silicon Retina Equipped with the NAPFM Communication System
7.5.1 Circuit Description
7.5.2 Noise Measurement Results
7.6 Projective Field Generation
7.6.1 Overview
7.6.2 Anisotropic Current Pulse Spreading in a Nonlinear Network
7.6.3 Analysis of the Spatial Response of the Nonlinear Network
7.6.4 Analysis of the Size and Shape of the Bubbles Generable by the Nonlinear Network
7.7 Description of the Integrated Circuit for Orientation Enhancement
7.7.1 Overview
7.7.2 Circuit Description
7.7.3 System Measurement Results
7.7.4 Other Applications
7.7.4.1 Weighted Projective Field Generation
7.7.4.2 Complex Projective Field Generation
7.8 Display Interface
7.9 Conclusion
References
8 Preprocessing for Pulsed Neural VLSI Syste
8.1 Introduction
8.2 A Sound Segmentation System
8.3 Signal Processing in Analog VLSI
8.3.1 Continuous Time Active Filters
8.3.2 Sampled Data Active Switched Capacitor (SC) Filters
8.3.3 Sampled Data Active Switched Current (SI) Filters
8.3.4 Discussion
8.4 Palmo -- Pulse Based Signal Processing
8.4.1 Basic Palmo Concepts
8.4.1.1 The Palmo Signal Representation
8.4.1.2 The Analog Palmo Cell
8.4.1.3 A Palmo Signal Processing System
8.4.1.4 Sources of Harmonic Distortion in a Palmo System
8.4.2 A CMOS Analog Palmo Cell Implementation
8.4.2.1 The Analog Palmo Cell: Details of Circuit Operation
8.4.3 Interconnecting Analog Palmo Cells
8.4.4 Results from a Palmo VLSI Device
8.4.5 Digital Processing of Palmo Signals
8.4.6 CMOS Analog Palmo Cell: Performance
8.5 Conclusions
8.6 Further Work
8.7 Acknowledgements
References
9 Digital Simulation of Spiking Neural Networks
9.1 Introduction
9.2 Implementation Issues of Pulse-Coded Neural Networks
9.2.1 Discrete-Time Simulation
9.2.2 Requisite Arithmetic Precision
9.2.3 Basic Procedures of Network Computation
9.3 Programming Environment
9.4 Concepts of Efficient Simulation
9.5 Mapping Neural Networks on Parallel Computers
9.5.1 Neuron-Parallelism
9.5.2 Synapse-Parallelism
9.5.3 Pattern-Parallelism
9.5.4 Partitioning of the Network
9.6 Performance Study
9.6.1 Single PE Workstations
9.6.2 Neurocomputer
9.6.3 Parallel Computers
9.6.4 Results of the Performance Study
9.6.5 Conclusions
References
Part III Design and Analysis of Pulsed Neural Systems
10 Populations of Spiking Neurons
10.1 Introduction
10.2 Model
10.3 Population Activity Equation
10.3.1 Integral Equation for the Dynamics
10.3.2 Normalization
10.4 Noise-Free Population Dynamics
10.5 Locking
10.5.1 Locking Condition
10.5.2 Graphical Interpretation
10.6 Transients
10.7 Incoherent Firing
10.7.1 Determination of the Activity
10.7.2 Stability of Asynchronous Firing
10.8 Conclusions
References
11 Collective Excitation Phenomena and Their Applications
11.1 Introduction
11.1.1 Two Variable Formulation of IAF Neurons
11.2 Synchronization of Pulse Coupled Oscillators
11.3 Clustering via Temporal Segmentation
11.4 Limits on Temporal Segmentation
11.5 Image Analysis
11.5.1 Image Segmentation
11.5.2 Edge Detection
11.6 Solitary Waves
11.7 The Importance of Noise
11.8 Conclusions
Acknowledgment
References
12 Computing and Learning with Dynamic Synapses
12.1 Introduction
12.2 Biological Data on Dynamic Synapses
12.3 Quantitative Models
12.4 On the Computational Role of Dynamic Synapses
12.5 Implications for Learning in Pulsed Neural Nets
12.6 Conclusions
References
13 Stochastic Bit-Stream Neural Networks
13.1 Introduction
13.2 Basic Neural Modelling
13.3 Feedforward Networks and Learning
13.3.1 Probability Level Learning
13.3.2 Bit-Stream Level Learning
13.4 Generalization Analysis
13.5 Recurrent Networks
13.6 Applications to Graph Colouring
13.7 Hardware Implementation
13.7.1 The Stochastic Neuron
13.7.2 Calculating Output Derivatives
13.7.3 Generating Stochastic Bit-Streams
13.7.4 Recurrent Networks
13.8 Conclusions
References
14 Hebbian Learning of Pulse Timing in the Barn Owl Auditory System
14.1 Introduction
14.2 Hebbian Learning
14.2.1 Review of Standard Formulations
14.2.2 Spike-Based Learning
14.2.3 Example
14.2.4 Learning Window
14.3 Barn Owl Auditory System
14.3.1 The Localization Task
14.3.2 Auditory Localization Pathway
14.4 Phase Locking
14.4.1 Neuron Model
14.4.2 Phase Locking -- Schematic
14.4.3 Simulation Results
14.5 Delay Tuning by Hebbian Learning
14.5.1 Motivation
14.5.2 Selection of Delays
14.6 Conclusions
References