基于半监督K-means的主动学习聚类算法
首发时间:2016-12-28
摘要:针对K-means算法对初始聚类中心敏感,针对不规则聚类簇效果较差的缺点,提出了一种基于半监督K-means的主动学习算法。为了针对指定的k个类别进行聚类,首先通过半监督聚类估算出每个类的中心,然后通过循环迭代过程中对特征值权重的反馈调节来适应指定聚类。对于K-means算法会对不规则类簇聚类效果差的问题,通过每次迭代时,影响多个类簇中心来实现较为精确的中心调整。在UCI中的20 Newsgroups和真实数据集实验表明,该算法在F1-measure指标上较其他几种算法,提升了分类精度。
For information in English, please click here
Active learning clustering algorithm based on semi-supervised K-means
Abstract:To solve the shortcomings of K-means algorithm, which is sensitive to initial clustering center and vulnerable to noise, a Active learning clustering algorithm based on semi-supervised K-means is proposed. In order to cluster fixed k classes, we first estimate the center of each class by semi-supervised clustering and then adjust the clustering by feedback adjustment of eigenvalue weights in the iterative process. During clustering process, adjusting the cluster centers can solve the problems which the K-means algorithm is sensitive to the irregular clusters. Experiments on 20 Newsgroups in UCI and real datasets show that the proposed algorithm improves the classification accuracy compared with other algorithms in terms of F1-measure.
Keywords: kK-means classification algorithm machine learning Noise filtering
基金:
论文图表:
引用
No.4714271117556814****
同行评议
共计0人参与
勘误表
基于半监督K-means的主动学习聚类算法
评论
全部评论