Towards explainable interactive multiobjective optimization: R-XIMO

Abstract In interactive multiobjective optimization methods, the preferences of a decision maker are incorporated in solution process to find solutions interest for problems with multiple conflicting objectives. Since exist these various trade-offs, crucial identify best solution(s). However, it is not necessarily clear how lead particular and, by introducing explanations we promote novel paradigm explainable . As proof concept, introduce new method, R-XIMO , which provides reference point based methods. We utilize concepts artificial intelligence and SHAP (Shapley Additive exPlanations) values. allows learn about trade-offs underlying problem promotes confidence found. particular, supports expressing that help them improve desired objective suggesting another be impaired. This kind support has been lacking. validate numerically, an illustrative example, case study demonstrating can real maker. Our results show successfully generates sound explanations. Thus, incorporating explainability methods appears very promising exciting research area.