ARTICLE Moerland-96.4/IDIAP Incorporation of Liquid-Crystal Light Valve Non-Linearities in Optical Multilayer Neural Networks Moerland, Perry Fiesler, Emile Saxena, Indu (artificial) neural network activation function curve fit gain hardware implementation liquid crystal light valve (LCLV) optical multilayer neural network EXTERNAL https://publications.idiap.ch/attachments/reports/1996/LCLV.pdf PUBLIC Applied Optics 35 26 5301-5307 0003-6935 1996 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 non-standard 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. In this paper, 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.