Rank-Based Nondomination Set Identification with Preprocessing.
ADVANCES IN SWARM INTELLIGENCE, ICSI 2016, PT II(2016)
摘要
In multi-objective optimization, finding a nondomination set is a computationally expensive process and the complexity grows with the population size. In this paper, we propose a nondomination set identification method with (a) rank-based preprocessing step where the obvious dominant solutions are eliminated and (b) better order of comparison based on the average rank so that the number of comparisons can be significantly reduced. In preprocessing, the maximum rank information of the solutions that are best in each individual objectives is used. In addition, during nondomination set identification process to check if a solution is nondominant the solution is compared only with solutions that are better in terms of average rank. The experiment results demonstrate the effectiveness of the proposed method in identifying the nondominant set in less number of comparisons.
更多查看译文
关键词
Multi-objective evolutionary algorithms,Nondomination set identification,Preprocessing step,Rank-based sorting
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络