Adaptive Control of the Number of Crossed Genes in Many-Objective Evolutionary Optimization

Abstract

To realize effective genetic operation in evolutionary many-objective optimization, crossover controlling the number of crossed genes (CCG) has been proposed. CCG controls the number of crossed genes by using an user-defined parameter α. CCG with small α significantly improves the search performance of multi-objective evolutionary algorithm in many-objective optimization by keeping small the number of crossed genes. However, to achieve high search performance by using CCG, we have to find out an appropriate parameter α by conducting many experiments. To avoid parameter tuning and automatically find out an appropriate α in a single run of the algorithm, in this work we propose an adaptive CCG which adopts the parameter α during the solutions search. Simulation results show that the values of α controlled by the proposed method converges to an appropriate value even when the adaptation is started from any initial values. Also we show the adaptive CCG achieves more than 80% with a single run of the algorithm for the maximum search performance of the static CCG using an optimal α^*.