Performance Assessment for Preference-Based Evolutionary Multi-Objective Optimization Using Reference Points

How to quantitatively compare the performance of different algorithms is one of the most important issues in multi-objective optimization. There are a number of metrics for evaluating the performance of a multi-objective optimizer that approximates the whole efficient front. However, for evaluating the quality of a preferred subset of the efficient front, the existing metrics are inadequate. In this paper, we suggest a systematic way to adapt the existing metrics to quantitatively evaluate the performance of a preference-based evolutionary multi-objective optimization algorithm using reference points. Its basic idea is to pre-process the preferred efficient set according to a multi-criterion decision-making approach before performance assessment. Extensive experiments conducted on several artificial scenarios and benchmark problems fully demonstrate its working principle, importance and effectiveness in evaluating the quality of different preferred efficient sets with regard to various reference points supplied by a decision maker. Index Terms User-preference, evolutionary multi-objective optimization, performance metric, reference point.