CHAOTIC BROWNIAN NETWORK

A system of many Brownian particles interacting via the chaotic force is proposed as a new model of neural network, which is called the chao tic Brownian network. The network is applied to the associative memory problem. The coupling constants are determined by the Hebb rule. It i s shown that some neuron pairs, which are specified by the stored patt erns, exhibit the synchronized motions with the same sign or with the opposite sign. The relation between the synchronized motion and the ch aotic wandering motion is discussed. The network is shown to retrieve the stored patterns or the reverse ones systematically and very quickl y. It is shown that the network can store and retrieve the number of p atterns comparable to the number of neurons. (C) 1998 Elsevier Science B.V. All rights reserved.