基于主动数据选取的半监督聚类算法

摘要

利用少量标签数据获得较高聚类精度的半监督聚类技术是近年来数据挖掘和机器学习领域的研究热点.但是现有的半监督聚类算法在处理极少量标签数据和多密度不平衡数据集时的聚类精度比较低.基于主动学习技术研究标签数据选取,提出了一个新的半监督聚类算法.该算法结合最小生成树聚类和主动学习思想,选取包含信息较多的数据点作为标签数据,使用类KNN思想对类标签进行传播.通过在UCI标准数据集和模拟数据集上的测试,结果表明提出的算法比其他算法在处理多密度、不平衡数据集时有更高精度且稳定的聚类结果.%Semi-supervised clustering, which aims to significantly improve the clustering results using limited supervision, has inevitably been the research focus in data mining and machine learning in recent years. But the accuracy of existing semi-clustering algorithms is low when dealing with the datasets with little labeled data or the multi-density and unbalanced datasets. Based on the active learning,this paper studied the data selection and presented a novel semi-supervised clustering algorithm. It selected information-rich data as labeled data by combining the ideas of minimum spanning tree clustering and active learning, and then used the KNN-like technology to propagate labels. Evaluating on several UCI standard datasets and synthetic datasets,the results show that the proposed method has manifest higher accuracy and stable performance in comparison with others,even when the datasets are multi-density and unbalanced.