The Effects of Optical Thresholding in Backpropagation Neural Networks

Sigmoid-like activation functions implemented in analog hardware differ in various ways from the standard sigmoidal function as they are asymmetric, truncated, and have a non-standard gain. It is demonstrated how one can adapt the backpropagation learning rule to compensate for these non-standard sigmoids as available in hardware. This method is applied to multilayer neural networks with all-optical forward propagation and liquid crystal light valves (LCLV) as optical thresholding devices. In this paper the results of software simulations of a backpropagation neural network with five different LCLV activation functions are presented and it is shown that the adapted learning rule performs well with these LCLV curves


Editor(s):
Fogelman-Soulié, F.
Gallinari, P.
Published in:
Proceedings of the International Conference on Artificial Neural Networks (ICANN'95 and NeuroNîmes'95), 2, 339-343
Presented at:
ENNS - Proceedings of the International Conference on Artificial Neural Networks (ICANN'95 and NeuroNimes'95), Paris, France
Year:
1995
Publisher:
Paris La Défense, France, EC2 & Cie
Keywords:
Laboratories:




 Record created 2006-03-10, last modified 2018-03-17


Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)