Sequential parameter optimization for multi-objective problems

Optimizing an algorithm´s parameter set for evolutionary multi-objective optimization (EMO) algorithms is not performed regularly until now. However, it could have been learned from single-objective optimization that doing so yields remarkable improvements in algorithm´s performance. Here, the sequential parameter optimization (SPO) framework is exemplarily applied to one EMO algorithm (EMOA) with different questions handled in different experiments. The main goal is to show the wide application area of such methods with a second, minor focus on the achievable improvements.