Summary
2011 International Symposium on Nonlinear Theory and Its Applications
2011
Session Number:A1L-B
Session:
Number:A1L-B1
Analysis of Spatiotemporal Search Dynamics in The Combinatorial Optimization Algorithm Using Lebesgue Spectrum Filter
Shinnya Fukuda, Tomohiro Kato, Mikio Hasegawa,
pp.21-24
Publication Date:2011/9/4
Online ISSN:2188-5079
Summary:
In this paper, we analyze the effects of low cross-correlation spatiotemporal searching dynamics generated by the Lebesgue Spectrum Filter for asynchronously updated combinatorial optimization algorithms. We measure the relation between the performance of such heuristic algorithms and the crosscorrelation and auto-correlation of the searching dynamics. In previous researches, we have proposed an approach to improve the performances of the asynchronously updated heuristic algorithms using negative auto-correlation based on the chaotic theory. The negative auto-correlation in each dimension of the searching dynamics makes the cross-correlation among the dimensions of the searching dynamics lower. Such a lower cross-correlation dynamics make more distributive search, and the performance of the heuristic methods can be much improved. To generate such low cross-correlation searching dynamics, we introduce the Lebesgue Spectrum Filter (LSF), and apply it to mutually connected neural network, which is one of the asynchronously updated heuristic algorithms. We show that the negative autocorrelation makes low crosscorrelation, which leads high performance, by numerical simulations.