基于流形距离的密度峰值快速搜索聚类算法

Clustering by Fast Search and find of Density Peaks Based on Manifold Distance

针对2014年6月发表在Science上的基于密度峰和快速搜索的聚类算法容易忽略无密度极值的簇的缺陷,提出了一种基于流形距离的密度峰值快速搜索聚类算法。算法利用流形距离弥补了传统欧式距离对于复杂数据无法反应聚类的全局一致性(即位于同一个类中的样本点之间有较高的相似度)的缺陷,通过近邻点充分挖掘复杂数据的流形结构信息,使处于同一个流形中的样本点之间相似性较高,从而正确找到密度极值点作为聚类中心点,完成聚类。本文算法能够发现较复杂的流形结构,在公开数据集上能取得较好的实验结果。

The clustering algorithm based on density peak and fast search,which was published on Science in June 2014,is easy to ignore the cluster which has no density extreme value.So We propose an algorithm based on manifold distance to solve this problem.Instead of Euclidean distance,the algorithm uses manifold distance to reflect the global consistency of samples,which means the samples in the same cluster have high similarity.We find manifold structure information of complex data by neighbor points,so that samples in the same manifold have high similarity and the cluster center is easy to find. In this paper,we can find manifold structure of complex data,and obtain better results in the open data sets.