Nonlinear Associative Dynamics and Pattern Representations in Chaotic Neural Networks(Structure and Bifurcation of Dynamical Systems)
Tohru Ikeguchi, Kazuki Hatamoto, Kazuyuki Aihara · Kyoto University Research Information Repository (Kyoto University) · 1992
We breifly review our model of chaotic neural networks and apply its dynamics to associative memory.In particular, we examine the influence of different encoding schemes of pattern vectors; namely orthogonal, nonorthogonal and sparse coding.The results obtained in this paper show that the output patterns of chaotic neural networks can be not only periodic but non-periodic (chaotic) and recall the stored patterns intermittently and succesively.With our model of chaotic neural networks, dynamical associative memory can be realized.