改进的快速非支配排序遗传算法Ⅱ及其在投资组合中的应用  

Improved Fast Non-dominated Sorting Genetic Algorithm Ⅱ and Its Application in Portfolio Optimization

作　　者：胡博[1] 肖辉[1] 金浩 汪镭[1] HU Bo;XIAO Hui;JIN Hao;WANG Lei(School of Electronics and Information Engineering, Tongji University, Shanghai 201804, China;School of Statistics and Management, Shanghai University of Finance and Economics, Shanghai 200433, China)

机构地区：[1]同济大学,电子与信息工程学院,上海201804 [2]上海财经大学,统计与管理学院,上海200433

出　　处：《微型电脑应用》2022年第2期9-11,共3页Microcomputer Applications

基　　金：国家自然科学基金资助项目(71771176,61503287,61703279);科技部科技冬奥项目(2018YFF0300505)。

摘　　要：快速非支配排序遗传算法Ⅱ (NSGA-Ⅱ)是经典的多目标优化器,然而,其采用的拥挤度策略存在着无法有效区分较为拥挤个体的缺陷。为了弥补此不足重新设计了拥挤度评价机制。新拥挤度评价机制不仅可以弥补上述缺陷,而且可以进一步增强算法的整体性能。在ZDT测试集和投资组合优化上的分析结果说明了改进的NSGA-Ⅱ综合较优表现。Fast non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ) is a classical optimizer for multi-objective optimization. However, the crowding distance strategy in NSGA-Ⅱ has the drawback of being unable to distinguish more crowded individuals. To remedy this issue, a new metric for measuring the crowding degree is designed. The metric is not only able to remedy the inefficiency above, but also able to further enhance the overall performance of improved NSGA-Ⅱ. The experiments on ZDT test suite and portfolio optimization problems illustrate the outstanding performance of improved NSGA-Ⅱ.

关 键 词：NSGA-Ⅱ 拥挤度距离 有效性 投资组合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]