Annealed chaotic neural network with nonlinear self-feedback and its application to clustering problem

摘要

Chaos is a revolutionary concept, which brings a novel strategy of science for researchers. In this paper, a chaotic neural network is proposed and the simulated annealing strategy also embedded to construct an annealed chaotic neural network (ACNN) and apply to the clustering problem. In addition to retain the characteristics of the conventional neural units, the ACNN displays a rich range of behavior reminiscent of that observed in neurons. Unlike the conventional neural network, the ACNN has rich range and flexible dynamics, so that it can be expected to have higher ability of searching for globally optimal or near-optimum results. However, the chaotic neural network does not stay in the global solution due to the chaotic dynamical mechanism being not clear. A chaotic mechanism with annealing strategy is introduced into the Hopfield network to construct a ACNN for expecting a better opportunity of converging to the optimal solution in this paper. In experimental results, unlike the fuzzy clustering methods getting local minima solutions, the ACNN method can always obtain the near-global optimal results. From the classification of real multispectral images, the ACNN can obtain suitable results.