A robust real-coded GA taking account of niching

摘要

Recently, some real-coded genetic algorithms (RCGAs) based on "A Functional Specialization Hypothesis for Designing Genetic Algorithms" proposed by Kita et al. have been proposed and shown good performance. However, there are some functions that the above RCGAs cannot solve, which we call deceptive functions. Ikeda et al. have proposed Innately Split Model (ISM) to solve the deceptive functions and succeeded in solving some of them. However, ISM cannot solve some functions which the above RCGAs can solve. In this paper, we propose a robust RCGA taking account of niching and show its effectiveness by applying it to functions that the above RCGAs can solve and deceptive functions that the above RCGAs cannot.