Exploratory multi-objective evolutionary algorithm: performance study and comparisons
Kay Chen Tan, Eik Fun Khor, C. M. Heng, T. H. Lee · 2001
An exploratory multi-objective evolutionary algorithm (EMOEA) has been proposed in previous publications. Its salient feature is to combine the properties of tabu search and evolutionary algorithm for effective multi-objective optimization. In addition, it also applies lateral interference, which is capable of distributing non-dominated individuals uniformly along the discovered Pareto-front at each generation without the need of any parameter setting. In this paper, the main objective is to perform extensive simulation studies to compare the performance of EMOEA against other evolutionary methods. Four benchmark test problems together with two well-known performance measures are applied. The studies have shown the competitive behavior of EMOEA to escape from local optima as well as to accurately identify the actual global optima in the noisy environment.