An Enhanced Multi-point Interactive Method for Multi-objective Evolutionary Algorithms

Abstract

Translate Article

In the real world, it often has multiple objectives (the objectives are often conflicting) in optimization problems. In many cases, a single solution is not being optimized with regards to all objectives. Dealing the problems, Multi-objective Evolutionary Algorithms (MOEAs) is known as a great potential. It is a hot trend in getting suitable solutions and making up the convergence of MOEAs, when the Decision-Maker’s (DM) consideration during the search (the interacting with a DM) to check, analyze the results, and give the preference. Recently, there are many researchers who focused on interactive methods for MOEAs, in [9], the authors proposed a multi-point methods to interactive with MOEAs, and MOEA/D is selected to build up the proposal. In [10], and its updated version in [11] based on DMEA-II [8], interactive ways with concept of rays were introduced. We found out some issues in these proposals and it raised to use a buffer instead of rays to improve the algorithm. The new method was confirmed on some experiments with popular benchmark sets.