Incorporation of liquid-crystal light valve nonlinearities in optical multilayer neural networks
Authors:
- P. D. MoerlandPMP. D. Moerland
- E. FieslerEFE. Fiesler
- I. SaxenaISI. Saxena
- 10 September 1996
- journal article
- Published by Optica Publishing Group in Applied Optics
- Vol. 35 (26) , 5301-5307
Abstract
Sigmoidlike activation functions, as available in analog hardware, differ in various ways from the standard sigmoidal function because they are usually asymmetric, truncated, and have a nonstandard gain. We present an adaptation of the backpropagation learning rule to compensate for these nonstandard sigmoids. This method is applied to multilayer neural networks with all-optical forward propagation and liquid-crystal light valves (LCLV) as optical thresholding devices. The results of simulations of a backpropagation neural network with five different LCLV response curves as activation functions are presented. Although LCLV’s perform poorly with the standard backpropagation algorithm, it is shown that our adapted learning rule performs well with these LCLV curves.Keywords
This publication has 8 references indexed in Scilit:
- The Interchangeability of Learning Rate and Gain in Backpropagation Neural NetworksNeural Computation, 1996
- Liquid-crystal light valves as thresholding elements in neural networks: basic device requirementsApplied Optics, 1994
- Dynamic optical interconnections using holographic lenslet arrays for adaptive neural networksOptical Engineering, 1993
- Noise analysis of polarization-based optoelectronic connectionist machinesApplied Optics, 1992
- Optical neural network with pocket-sized liquid-crystal televisionsOptics Letters, 1990
- All-optical neural network with inhibitory neuronsOptics Letters, 1989
- Analysis of hidden units in a layered network trained to classify sonar targetsNeural Networks, 1988
- Application Of The Liquid Crystal Light Valve To Real-Time Optical Data ProcessingOptical Engineering, 1978