基于约束的模糊分类算法改进研究  

An improved restraint-based fuzzy classification algorithm

作　　者：令狐大智[1] 武新丽[2] 王东红[3] 

机构地区：[1]广西大学商学院,广西南宁530004 [2]广西大学行健文理学院,广西南宁530005 [3]广西财经学院工商管理系,广西南宁530003

出　　处：《辽宁工程技术大学学报（自然科学版）》2010年第5期970-973,共4页Journal of Liaoning Technical University (Natural Science)

基　　金：广西自然科学基金资助项目(0640026);广西青年科技基金资助项目(桂科青0991097);广西教育厅基金资助项目(200705LX300)

摘　　要：针对基于约束的模糊数据归类算法的不足,构建模式自学习调节模型,并提出基于约束规则的模式微调算法CIP、基于自学习的分类规则优化更新算法OCRS。新算法基于模式间相关性、距离和支持度等因素,建立模式自主更新标准和算法协调机制。实验研究表明:新算法在准确度相同的情况下,增加了算法的识别率、自学习能力和鲁棒性。In order to overcome the shortcoings in restraint-based fuzzy data classification algorithm,a model of pattern self-learning is developed in this study. Also,a constraint-based inching pattern algorithm （CIP） and optimized classification rule algorithm based on self-learning （OCRS） are proposaed. The new algorithms incorporate the standard of pattern self-update and the coordination mechanism which are based on various factors,such as models correlation,distance and support etc. An experimental study shows that the new algorithms enhance the ability of self-learning and the robustness.

关 键 词：数据挖掘 自学习 模糊集 聚类分析 模式优化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]