Summary
International Symposium on Nonlinear Theory and its Applications
2008
Session Number:A1L-F
Session:
Number:A1L-F1
Optimization using Higher-Order Chaotic Neural Networks
Taro Kuroda, Mikio Hasegawa,
pp.-
Publication Date:2008/9/7
Online ISSN:2188-5079
Summary:
Chaotic neural networks have been applied to various combinatorial optimization problems, and effectiveness of chaotic dynamics for solution search has been shown. However, the conventional chaotic neural networks based on the Hopfield-Tank neural network can solve the problems whose objective function is only second or lower-order function of the state of the neurons. In order to apply the chaotic optimization framework to more general problems, we introduce the higher-order neural networks which have higher-order connections and energy function. We verify effectiveness of a chaotic method on such a higher order neural network by comparing its performances with gradient dynamics and stochastic dynamics.