一种基于两因素相结合的自适应学习三支决策阈值的算法  

Adaptive Algorithm for Learning Optimal Threshold in Three-way Decisions Based on Two Factors

作　　者：朱艳辉[1] 田海龙[1] 张永平[1] 朱道杰 

机构地区：[1]湖南工业大学计算机与通信学院,湖南株洲412008

出　　处：《小型微型计算机系统》2016年第6期1303-1307,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61170102)资助;国家社科基金项目(12BYY045)资助;湖南省教育厅重点项目(15A049)资助

摘　　要：针对三支决策自动学习阈值问题,综合考虑决策风险总损失和分类器的综合性能两因素,提出一种基于层次分析法的分类器性能综合评价模型,结合决策风险损失建立了自动学习三支决策最优化阈值模型,进而提出一种基于风险损失与评价性能两因素相结合的三支决策自适应阈值算法,实验表明,提出的算法能学习到有效的三支决策阈值,并可以灵活设置权重参数和倍率参数等相关参数,权衡决策风险损失和分类器的综合性能,使决策者在允许的决策风险损失下有效的提高分类器的综合性能.In order to automaficly achieve three-way decisions threshold which considered comprehensive performance index evaluation classifier and risk cost of decision. Therefore, a comprehensive performance index evaluation classifier model was presented based on Analytic Hierarchy Process, and then an optimal threshold model of Three-way decisions was established which combined with the risk cost of decision. After that,an algorithm for learning thresholds in Three-way decision-theoretic was proposed based on the combina- tion of the risk loss and evaluation performance. Experiment results show that the proposed algorithm can learn thresholds efficiently, set weight and rate parameters flexibly, balance the risk loss and performance evaluation of the classifier, thereby allowing the overall performance of the classifier to be improved effectively under the acceptable risk loss of decisions.

关 键 词：风险损失 分类器性能 三支决策 自适应算法 阈值 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]