快速特征映射优化的流形密度峰聚类

Manifold density peak clustering optimized by fast feature mapping

经典的密度峰聚类不再适用于复杂的流形聚类,因此提出了快速特征映射优化的流形密度峰聚类,用快速特征映射优化的流形距离取代欧式距离,可以更好地反映不同类的点间相似性.算法首先通过寻找特征点,构造无向特征图,再通过无向特征图计算任意两个点之间的流形距离,最后按照流形距离的大小完成分配.在人工数据集和UCI数据集上的实验表明,新算法具有更高的准确率.

For complex manifold clustering,density peak clustering is no longer applicable,and fast feature mapping is proposed to optimize the manifold density clustering in this paper.By using the fast feature map to optimize the manifold distance to replace the Euclidean distance,it is better to reflect the similarity between different classes of points.The algorithm first constructs the undirected feature graph by looking for the feature points.For any two points,the manifold distance between them is calculated by the undirected feature graph,and finally the distribution is done according to the size of the manifold.Experiments on artificial datasets and UCI datasets show that the new algorithm has a higher accuracy rate.