DISCRETE ALL-POSITIVE MULTILAYER PERCEPTRONS FOR OPTICAL IMPLEMENTATION

All-optical multilayer perceptrons differ in various ways from the ide al neural network model. Examples are the use of nonideal activation f unctions, which are truncated, asymmetric, and have a nonstandard gain ; restriction of the network parameters to non-negative values, and th e limited accuracy of the weights. A backpropagation-based learning ru le is presented that compensates for these nonidealities and enables t he implementation of all-optical multilayer perceptrons where learning occurs under computer control. The good performance of this learning rule, even when using a small number of weight levels, is illustrated by a series of computer simulations incorporating the nonidealities. ( C) 1998 Society of Photo-Optical Instrumentation Engineers.