INTERACTING BROWNIAN PARTICLES AS A MODEL OF NEURAL-NETWORK

A model of neural network is proposed, in which the dynamics of the in ternal state of neurons is based on the Brownian motion with the chaot ic force, and the bifurcation parameter of the chaotic force is modula ted by the position of the Brownian particles. This model is applied t o the associative memory problem. The parameter dependence of the dyna mics is studied. It is shown that the model can retrieve all embedded patterns and reverse patterns successively. It is found that all neuro ns exhibit synchronized behavior while a pattern is retrieved.