基于信息共享策略的加权逐点预测动态多目标优化
WEIGHTED POINT-WISE PREDICTION DYNAMIC MULTIPLE OBJECTIVE OPTIMIZATION BASED ON INFORMATION SHARING STRATEGY
包全磊 1陈红星2
作者信息
- 1. 太原学院 山西太原 030032
- 2. 山西大学计算机与信息技术学院 山西太原 030006
- 折叠
摘要
为了有效实现Pareto最优解,并且提升对于输入误差的鲁棒性,提出一种基于信息共享策略的加权逐点预测动态多目标优化算法.引入一种信息共享策略,该策略允许每个点利用其相邻解决方案中的信息有效提升模型的鲁棒性.引入一个相似性度量,并对其与一些常用的相似性度量进行对比分析,进一步提出一种加权逐点预测方法,逐点特性极大增强了捕捉各种模式的能力.在EC2018 DMO测试套件上的实验结果验证了该方法的有效性.
Abstract
In order to achieve the Pareto optimal solution effectively and improve the robustness to the input error,a weighted point-wise prediction dynamic multiple objective optimization algorithm based on information sharing strategy is proposed.An information sharing strategy was introduced,which allowed each point to make use of the information in its adjacent solutions to improve the robustness of the model.A similarity measure was introduced,and by comparing with some common similarity measures,a weighted point-wise prediction method was proposed,which greatly enhanced the ability to capture various patterns.Experimental results on EC2018 DMO test suite show the effectiveness of the proposed method.
关键词
信息共享/鲁棒性/多目标优化/加权逐点预测Key words
Information sharing/Robustness/Multiple objective optimization/Weighted point-wise prediction引用本文复制引用
基金项目
山西省专利推广项目计划(201005-201105)
出版年
2024
国家科技期刊平台
NSTL
万方数据