基于决策空间变换最近邻方法的Pareto支配性预测  

Prediction of Pareto Dominance Using Nearest Neighbor Method Based on Decision Space Transformation

作　　者：李文彬[1,2] 贺建军[1] 冯彩英[2] 郭观七[2] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]湖南理工学院信息与通信工程学院,岳阳414006

出　　处：《自动化学报》2017年第2期294-301,共8页Acta Automatica Sinica

基　　金：国家自然科学基金(60975049;61174132);湖南省教育厅科学研究重点项目(15A079);湖南省高校科技创新团队支持计划资助~~

摘　　要：为提高在决策空间运用最近邻方法预测多目标优化Pareto支配性的精度,提出一种基于决策空间变换的最近邻预测方法.在分析目标函数与决策分量相关性的基础上,提出属性变化趋势模型的构造方法,建立低计算成本的属性趋势代理模型.通过属性趋势模型引入决策空间到目标空间的映射知识,对多目标问题的决策空间进行变换,使决策空间的最近邻更有效反映目标空间的最近邻.选取具有不同相关系数特征的典型多目标优化问题,进行Pareto支配性预测的可对比实验,结果表明在新空间中运用最近邻方法可显著提高分类准确性.In this paper, nearest neighbor prediction is used to decide Pareto dominance of candidate solutions in expensive multi-objective optimization. For improving the accuracy of predicting Pareto dominance in decision space, a transformation method of decision space is proposed. Based on correlation analysis of objective functions and decision attributes, computationally efficient attribute tendency models of objective functions are set up. These models are used to re-construct the decision space such that the knowledge of objective space is introduced and the neighborhood relation in decision space can more effectively reflect that in the original objective space. Experiments of Pareto dominance prediction are carried out on a group of typical multi-objective optimization problems. The results indicate that the prediction of Pareto dominance using the proposed method is more accurate and efficient than the existing methods.

关 键 词：趋势模型 空间变换 最近邻方法 Pareto支配性 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]