Using genetic algorithms for an optical thin-film learning model

Abstract

A novel connectionist architecture based on an optical thin-film multilayer model (OTFM) is described. The architecture is explored as an alternative to the widely used neuron-inspired models, with the thin-film thicknesses serving as adjustable ‘weights’ for the computation. The use of genetic algorithms for training the thin-film model, along with experimental results on the parity problem and the iris data classification are presented.