半监督近邻传播聚类集成算法研究

Research on Semi Supervised Neighborhood Propagation Clustering Integration Algorithm

近邻传播聚类(Affinity Propagation,AP)算法具有良好的聚类性能,但是在聚类过程中单个聚类器的聚类性能不佳,存在聚类准确率较低的现象,如果将多个近邻传播聚类器集成起来则能够克服聚类的随机性和单个聚类器聚类准确率不高的问题。笔者利用装袋法将半监督AP算法进行集成,在标准数据集上进行测试,获得了更优的聚类精度。

The affinity propagation algorithm has good clustering performance,but during the clustering process,the clustering performance of a single clusterer is not good,and the accuracy is low.If multiple affinity propagation clusterers are integrated,it can be overcome that the randomness of clustering and the low accuracy of single clusterer.In this paper,the bagging method is used to integrate the semi-supervised AP algorithm,and the test is performed on a standard dataset to obtain better clustering accuracy.