Improved Helper-Objective Optimization Strategy for Job-Shop Scheduling Problem

Abstract

A single-objective optimization problem can be solved more efficiently by introducing some helper-objectives and running a multi-objective evolutionary algorithm. But what objectives should be used at each optimization stage? This paper describes a new method of adaptive helper-objectives selection in multi-objective evolutionary algorithms. The proposed method is applied to the Job-Shop scheduling problem and compared with the previously known approach, which was specially developed for the Job-Shop problem. A comparison with the previously proposed method of adaptive helper-objective selection based on reinforcement learning is performed as well.