Summary

International Symposium on Nonlinear Theory and its Applications

2010

Session Number:B2L-A

Session:

Number:B2L-A1

Effects of Chaotic Fluctuation on the Hopfield Model for the Traveling Salesman Problem

Ayano Hirahara,  Osamu Araki,  

pp.285-288

Publication Date:2010/9/5

Online ISSN:2188-5079

DOI:10.34385/proc.44.B2L-A1

PDF download (825.5KB)

Summary:
In previous studies, it has been shown that the chaotic or stochastic fluctuation which has a negative and declining autocorrelation contributes to closer reach to the optimum solution in the Hopfield net. In this study, we examine this property from another viewpoint of evaluation.We evaluate the performance by reaching speed to the neighborhood of the optimum solution for a constant learning period. We made the Hopfield net solve the traveling salesman problem with chaotic fluctuations, which are generated from the logistic map and Bernoulli map. The results show that negative autocorrelations of fluctuations may contribute to better performance in both maps, which corresponds to the results of previous studies. In addition, they suggest that even periodic signals can contribute to the performance, but intermittency never does.