Improving the Efficiency of R2HCA-EMOA
| Autor: | Hisao Ishibuchi, Lie Meng Pang, Weiyu Chen, Ke Shang, Longcan Chen |
|---|---|
| Rok vydání: | 2021 |
| Předmět: | |
| Zdroj: | Lecture Notes in Computer Science ISBN: 9783030720612 EMO |
| Popis: | R2HCA-EMOA is a recently proposed hypervolume-based evolutionary multi-objective optimization (EMO) algorithm. It uses an R2 indicator variant to approximate the hypervolume contribution of each solution. Meanwhile, it uses a utility tensor structure to facilitate the calculation of the R2 indicator variant. This makes it very efficient for solving many-objective optimization problems. Compared with HypE, another hypervolume-based EMO algorithm designed for many-objective problems, R2HCA-EMOA runs faster and at the same time achieves better performance. Thus, R2HCA-EMOA is more attractive for practical use. In this paper, we further improve the efficiency of R2HCA-EMOA without sacrificing its performance. We propose two strategies for the efficiency improvement. One is to simplify the environmental selection, and the other is to change the number of direction vectors depending on the state of evolution. Numerical experiments clearly show that the efficiency of R2HCA-EMOA is significantly improved without deteriorating its performance. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|