Focal-Plane and Multiple Chip VLSI Approaches to CNNs

Authors:
M. Anguita;F. J. Pelayo;E. Ros;D. Palomar;A. Prieto
Affiliations:
Departamento de Electronica y Tecnología de Computadores, Facultad de Ciencias, Universidad de Granada, 18071-Granada, Spain;Departamento de Electronica y Tecnología de Computadores, Facultad de Ciencias, Universidad de Granada, 18071-Granada, Spain;Departamento de Electronica y Tecnología de Computadores, Facultad de Ciencias, Universidad de Granada, 18071-Granada, Spain;Departamento de Electronica y Tecnología de Computadores, Facultad de Ciencias, Universidad de Granada, 18071-Granada, Spain;Departamento de Electronica y Tecnología de Computadores, Facultad de Ciencias, Universidad de Granada, 18071-Granada, Spain
Venue:
Analog Integrated Circuits and Signal Processing - Special issue: cellular neural networks and analog VLSI
Year:
1998

Citing 2
Cited 3

VLSI Implementation of a Neural Model Using Spikes

Analog Integrated Circuits and Signal Processing
A VLSI Approach to the Implementation of Additive and Shunting Neural Networks

IWANN '96 Proceedings of the International Workshop on Artificial Neural Networks: From Natural to Artificial Neural Computation

A QCIF Resolution Binary I/O CNN-UM Chip

Journal of VLSI Signal Processing Systems - Special issue on spatiotemporal signal processing with analog CNN visual microprocessors
Stochastic Neural Computation II: Soft Competitive Learning

IEEE Transactions on Computers
CNN based Hole filler template design using numerical integration techniques

ICANN'07 Proceedings of the 17th international conference on Artificial neural networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, three alternative VLSI analog implementationsof CNNs are described, which have been devised to perform imageprocessing and vision tasks: a programmable low-power CNN withembedded photo-sensors, a compact fixed-template CNN based onunipolar current-mode signals, and basic CMOS circuits to implementan extended CNN model using spikes. The first two VLSI approachesare intended for focal-plane image processing applications. Thethird one allows, since its dynamics is defined by process-independentlocal ratios and its input/outputs can be efficiently multiplexedin time, the construction of very large multiple chip CNNs formore complex vision tasks.