A Gaussian synapse circuit for analog VLSI neural networks

Authors:
Joongho Choi;Bing J. Sheu;Josephine C.-F. Chang
Affiliations:
IBM T. J. Watson Research Center, Yorktown Heights, NY and University of Southern California, Los Angeles, CA;Department of Electrical Engineering, University of Southern California, Los Angeles, CA;Department of Electrical Engineering, University of Southern California, Los Angeles, CA
Venue:
IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Year:
1994

Citing 4
Cited 2

Analog VLSI and neural systems

Analog VLSI and neural systems
Neural network architectures: an introduction

Neural network architectures: an introduction
A resource-allocating network for function interpolation

Neural Computation
Analysis and Design of Analog Integrated Circuits

Analysis and Design of Analog Integrated Circuits

Optimised PWL recursive approximation and its application to neuro-fuzzy systems

Mathematical and Computer Modelling: An International Journal
Low-voltage improved accuracy Gaussian function generator with fourth-order approximation

Microelectronics Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

Back-propagation neural networks with Gaussian funetion synapses have better convergence property over those with linear-multiplying synapses. In digital simulation, more computing time is spent on Gaussian hetion evaluation. We present a compact analog synapse cell which is not biased in the subthreshold region for fully-parallel operation. This cell can approximate a Gaussian function with accuracy around 98% in the ideal case. Device mismatcb induced by fabrication process will cause some degradation to this approximation. The Gaussian synapse cell can alsa he used in unsupervised learning. Programmability of the proposed Gaussian synapse cell is achieved by changing the stored synapse weight Wji, the reference eumnt and the sizes of transistors in the differential pair.