Summary
2007 International Symposium on Nonlinear Theory and its Applications
2007
Session Number:19PM1-B
Session:
Number:19PM1-B-4
Recalling Complex Sequences of Patterns Using Neurons with Hysteretic Property
Johan Sveholm, Yoshihiro Hayakawa, Koji Nakajima,
pp.501-504
Publication Date:2007/9/16
Online ISSN:2188-5079
DOI:10.34385/proc.41.19PM1-B-4
Summary:
A network based on the Inverse Function Delayed (ID) model which can recall complex temporal sequences of patterns, is proposed. Complex pattern can be dealt with by extending the network, for each main unit, with buffer units that carries the role of a memory. Replacing the cross correlated weight matrix with a matrix computed from a linear separation problem point of view, these pattern can be stored and recalled even if they are highly correlated. It is shown that for a rather big network complex patterns of degree 3, consisting of highly correlated patterns, can be recalled.