Competitively Inhibited Optical Neural Networks Using Two-step Holographic Materials
Michael D. Lemmon, B. V. K. Vijaya Kumar · 1989
Competitively inhibited networks automatically form nonparametric representations of density functions and therefore can be used as MAP predictors on a variety of problems. The dynamics of this class of networks is briefly reviewed and an optical implementation is proposed based on two-step holographic materials.