An IGD+ Indicator-Based Metric for Interactive Evolutionary Multi-Objective Optimization Algorithms
Hao Tian, Fei Li, Baiyu Zhou, Yujie Yang, Xun Cheng Huang · 2024
Interactive evolutionary multi-objective optimization algorithms have received widespread research. The goal of these algorithms is to iteratively involve decision makers(DM) in the solution process, by providing preference information, the solution is guided to regions of interest. However, most indicators are designed to assess how well algorithms approximate the entire Pareto optimal front. In order to assess the quality of the preferred solution sets obtained by interactive algorithms, this paper introduces a novel performance indicator based on the modified inverted generational distance (IGD+). The indicator is examined using interactive evolutionary methods and demonstrating its capability to evaluate the performance of interactive algorithms.