P-ADIC DYNAMICAL SYSTEMS AND NEURAL NETWORKS

A p-adic model which describes a large class of neural networks is presented. In this model the states of neurons are described by digits in the canonical expansion of a p-adic number. Thus each p-adic number represents a configuration of firing and non firing neurons. The process of recognition of patterns is investigated in the P-adic framework. We study heteroassociative and autoassociative nets. P-adic dynamical systems are used to describe a feedback process for autoassociative nets.