International Symposium on Nonlinear Theory and its Applications
Retrieval Characteristics of a Dynamical Associative Memory Using Nonlinear Support Vector Machines
Akinori KATO, Masaharu ADACHI,
PDF download (208.5KB)
In this paper, we introduce chaotic model neurons as the constituents of the associative memory model with nonlinear SVMs. We examine the retrieval characteristics of the dynamical associative memory model through numerical experiments. As for the stored patterns used in our previous paper, we have demonstrated that a dynamic associative memory can be realized by the proposed model. In this paper, we investigate retrieval characteristics of recall and transition frequencies of the stored patterns with considering correlation between them.