Optimizing reserve size in genetic algorithms with reserve selection using reinforcement learning

Abstract

Recently, an improved genetic algorithm with a reserve selection mechanism (GARS) has been proposed to prevent premature convergence, where a parameter called reserve size plays an important role in optimization performance. In this paper, we propose an approach to the learning of an optimal reserve size in GARS based on the technique of reinforcement learning, where the learning model and algorithm are presented respectively. The experimental results demonstrate the effectiveness of learning algorithm in discovering the optimal reserve size accurately and efficiently.