Addressing Current Challenges in Evolutionary Multi-Objective Optimization: Indicator-based Selection, Convergence and Applicability

Basic data for this project

Description

The problem of simultaneously optimizing multiple and at least partly contradicting objectives cannot be solved by searching for only a single solution but by determining an optimal set ofcompromises. For about two decades, the area of Evolutionary Multi-Objective Optimization deals with this problem in an approximate, algorithmic way. Still, there are many unsolved challenges in this domain: It is still a challenge to determine the quality of gained solutions and to interpret the meaning of existing indicators. Also their application in state-of-the-art algorithms is still a hot research topic. Further, it has to be clarified how to describe the convergence behavior of evolutionary multi-objective optimizers and how to use this insight for termination of algorithms, how many objectives (> 3) can be handled efficiently, and how all the proposed algorithms can be transferred to real-world application. This project aims to initiate and intensify bi-lateral collaboration of researchers from Brazil and Germany under the umbrella of the before mentioned research questions by personal exchange.