Improved binary additive epsilon indicator for obtaining uniformly distributed solutions in multi-objective optimization
Improved binary additive epsilon indicator for obtaining uniformly distributed solutions in multi-objective optimization
复制标题
改进的二元加性 epsilon 指标，用于在多目标优化中获得均匀分布的解
DOI：
10.1145/3319619.3322025
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
H. Nishida
中科院分区：
文献类型：
--
作者：
Tatsumasa Ishikawa;Hiroaki Fukumoto;A. Oyama;H. Nishida
一些研究报告指出，基于指标的进化算法（IBEA）与二进制加性指标的解决方案是有偏的目标空间中的特定区域。本文揭示了产生偏差的原因是由于没有考虑显性关系而导致的非对称性。本文还提出了一种改进的随机指标，以获得均匀分布的目标空间中的解决方案。当两个解之间不存在优势关系时，该指标使用基于切比雪夫距离的指标值。实验结果表明，该指标在WFG5问题上提高了IBEA的多样性。
Several studies report that the solutions obtained by indicator-based evolutionary algorithm (IBEA) with the binary additive epsilon indicator are biased to specific regions in the objective space. This paper reveals the reason of the bias is due to the asymmetricity of the epsilon indicator without considering the dominance relationship. This paper also proposes a modified epsilon indicator to obtain uniformly distributed solutions in the objective space. The proposed indicator uses indicator value based on Chebyshev distance when the dominance relationship does not exist between two solutions. The results of the experiments show that the diversity of IBEA is enhanced by the proposed indicator on WFG5 problem.
DOI：
--
发表时间：
2016
期刊：
--
影响因子：
--
作者：
Bingdong Li;K. Tang;Jinlong Li;X. Yao
通讯作者：
Bingdong Li;K. Tang;Jinlong Li;X. Yao