一种处理部分标记数据的粗糙集属性约简算法  

Rough Set Attribute Reduction Algorithm for Partially Labeled Data

作　　者：张维[1,2,3] 苗夺谦[1,3] 高灿[4,5] 李峰[1,3] ZHANG Wei MIAO Duo-qian GAO Can LI Feng(Department of Computer Science and Technology, School of Electronics and Information Engineering, Tongii University, Shanghai 201804, China Department of Computer Science and Technology, Shanghai University of Electric Power, Shanghai 200090, China The Key Laboratory of Embedded System and Service Computing, Ministry of Education, Tongji University, Shanghai 201804, China School of Computer and Software, Shenzhen University, Guangdong 518060, China Institute of Textiles ＆ Clothing,The Hong Kong Polytechnic University, Hong Kong,China)

机构地区：[1]同济大学电子与信息工程学院计算机科学与技术系,上海201804 [2]上海电力学院计算机科学与技术学院,上海200090 [3]同济大学嵌入式系统与服务计算教育部重点实验室,上海201804 [4]深圳大学计算机与软件学院,广东518060 [5]香港理工大学应用科学与纺织学院

出　　处：《计算机科学》2017年第1期25-31,共7页Computer Science

基　　金：国家自然科学基金项目(61273304);2013年度高等学校博士学科点专项科研基金(20130072130004)资助

摘　　要：属性约简是粗糙集理论中重要的研究内容之一,是数据挖掘中知识获取的关键步骤。Pawlak粗糙集约简的对象一般是有标记的决策表或者是无标记的信息表。而在很多现实问题中有标记数据很有限,更多的是无标记数据,即半监督数据。为此,结合半监督协同学习理论,提出了处理半监督数据的属性约简算法。该算法首先在有标记数据上构造两个差异性较大的约简来构造基分类器;然后在无标记数据上交互协同学习,扩大有标记数据集,获得质量更好的约简,构造性能更好的分类器,该过程迭代进行,从而实现利用无标记数据提高有标记数据的约简质量,最终获得质量较好的属性约简。UCI数据集上的实验分析表明,该算法是有效且可行的。Attribute reduction, as an important preprocessing step for knowledge acquiring in data mining, is one of the key issues in rough set theory. Rough set theory is an effective supervised learning model for labeled data. However, at- tribute reduction for partially labeled data is outside the realm of traditional rough set theory. In this paper,a rough set attribute reduction algorithm for partially labeled data was proposed based on co-training which capitalizes on the unla- beled data to improve the quality of attribute reducts from few labeled data. It gets two diverse reducts of the labeled da- ta,employs them to train its base classifiers, and then co-trains the two base classifiers iteratively. In every round, the base classifiers learn from each other on the unlabeled data and enlarge the labeled data, so better quality reducts could be computed from the enlarged labeled data and employed to construct base classifiers of higher performance. The theo- retical analysis and experimental results with UCI data sets show that the proposed algorithm can select a few attributes but keep classification power.

关 键 词：粗糙集 增量式属性约简 协同学习 部分标记数据 半监督学习 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]