Using self-adaptable probes for dynamic parameter control of parallel evolutionary algorithms

Abstract

Controlling parameters during execution of parallel evolutionary algorithms is an open research area. Some recent research have already shown good results applying self-calibrating strategies. The motivation of this work is to improve the search of parallel genetic algorithms using monitoring techniques. Monitoring results guides the algorithm to take some actions based on both the search state and the values of its parameters. In this paper, we propose a parameter control architecture for parallel evolutionary algorithms, ba sed on self-adaptable monitoring techniques. Our approach provides an efficient and low cost monitoring technique to design parameters control strategies. Moreover, it is completely independant of the implementation of the evolutionary algorithm. © Springer-Verlag Berlin Heidelberg 2005.