Performance of DEMO on New Test Problems: A Comparison Study
Tea Robič · 2005
DEMO (Differential Evolution for Multiobjective Optimization) is a new algorithm for solving multiobjective problems. This paper presents a comparison between DEMO and three state-of-the-art algorithms for multiobjective optimization using a new comparison methodology and new test problems. Results show that DEMO outperforms the other algorithms used in comparison in four out of nine test problems. On only one test problem the other algorithms achieve significantly better results than DEMO, while in the remaining four test problems DEMO is the secondbest algorithm.