CAVIAR: A 45k Neuron, 5M Synapse, 12G Connects/s AER Hardware Sensory-Processing-Learning-Actuating System for High-Speed Visual Object Recognition and Tracking

被引：228

作者：

Serrano-Gotarredona, Rafael ^{[1
]}

Oster, Matthias ^{[2
]}

Lichtsteiner, Patrick ^{[2
]}

Linares-Barranco, Alejandro ^{[3
]}

Paz-Vicente, Rafael ^{[3
]}

Gomez-Rodriguez, Francisco ^{[3
]}

Camunas-Mesa, Luis ^{[4
]}

Berner, Raphael ^{[2
]}

Rivas-Perez, Manuel ^{[3
]}

Delbrueck, Tobi ^{[2
]}

Liu, Shih-Chii ^{[2
]}

Douglas, Rodney ^{[2
]}

Hafliger, Philipp ^{[5
]}

Jimenez-Moreno, Gabriel ^{[3
]}

Civit Ballcels, Anton ^{[3
]}

Serrano-Gotarredona, Teresa ^{[4
]}

Acosta-Jimenez, Antonio J. ^{[4
]}

Linares-Barranco, Bernabe ^{[4
]}

机构：

[1] CSIC, Seville Microelect Inst, Seville 41012, Spain

[2] Univ Zurich, Inst Neuroinformat, ETH, CH-8057 Zurich, Switzerland

[3] Univ Seville, Comp Architecture & Technol Dept, E-41012 Seville, Spain

[4] CSIC, Seville Microelect Inst, Seville 41092, Spain

[5] Univ Oslo, NO-0316 Oslo, Norway

来源：

IEEE TRANSACTIONS ON NEURAL NETWORKS | 2009年 / 20卷 / 09期

关键词：

Address-event representation (AER); neuromorphic chips; neuromorphic systems; vision; NETWORKS; COMMUNICATION; INTEGRATE; CHIP; CONTRAST; RETINA; MODEL;

D O I：

10.1109/TNN.2009.2023653

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

This paper describes CAVIAR, a massively parallel hardware implementation of a spike-based sensing-processing-learning-actuating system inspired by the physiology of the nervous system. CAVIAR uses the asychronous address-event representation (AER) communication framework and was developed in the context of a European Union funded project. It has four custom mixed-signal AER chips, five custom digital AER interface components, 45k neurons (spiking cells), up to 5M synapses, performs 12G synaptic operations per second, and achieves millisecond object recognition and tracking latencies.

引用

页码：1417 / 1438

页数：22

共 95 条

[41] Visual brain and visual perception: How does the cortex do perceptual grouping? [J].

Grossberg, S ;

Mingolla, E ;

Ross, WD .

TRENDS IN NEUROSCIENCES, 1997, 20 (03) :106-111

[42]

HAFLIGER P, 2003, P IEEE INT S CIRC SY, V1, P22

[43] Adaptive WTA with an analog VLSI neuromorphic learning chip [J].

Hafliger, Philipp .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 2007, 18 (02) :551-572

[44] A Biologically Inspired Modular VLSI System for Visual Measurement of Self-Motion [J].

Higgins, Charles M. ;

Shams, Shaikh Arif .

IEEE SENSORS JOURNAL, 2002, 2 (06) :508-528

[45] A reconfigurable neuromorphic VLSI multi-chip system applied to visual motion computation [J].

Indiveri, G ;

Whatley, AM ;

Kramer, J .

PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON MICROELECTRONICS FOR NEURAL, FUZZY AND BIO-INSPIRED SYSTEMS, MICORNEURO'99, 1999, :37-44

[46] Quantifying the contribution of low-level saliency to human eye movements in dynamic scenes [J].

Itti, L .

VISUAL COGNITION, 2005, 12 (06) :1093-1123

[47] Quantitative modelling of perceptual salience at human eye position [J].

Itti, Laurent .

VISUAL COGNITION, 2006, 14 (4-8) :959-984

[48] Address-event based platform for bio-inspired spiking systems [J].

Jimenez-Fernandez, A. ;

Lujan, C. D. ;

Linares-Barranco, A. ;

Gomez-Rodriguez, F. ;

Rivas, M. ;

Jimenez, G. ;

Civit, A. .

BIOENGINEERED AND BIOINSPIRED SYSTEMS III, 2007, 6592

[49] Asynchronous communication of 2D motion information using winner-takes-all arbitration [J].

Kalayjian, Z ;

Andreou, AG .

ANALOG INTEGRATED CIRCUITS AND SIGNAL PROCESSING, 1997, 13 (1-2) :103-109

[50]

Kempter R, 1999, ADV NEUR IN, V11, P125

← 1 2 3 4 5 6 7 8 9 10 →