基于形状相似距离的K-means聚类算法

K-means clustering algorithm based on the shape similarity distance

把向量作为空间中的物体展开相似度的评估,分析了向量间各维差值与形状差异的间的近似关系,提出了基于形状相似距离的K-means算法。在三个UCI(University of California,Irvine)标准数据集上的聚类结果表明,对于有关形状信息的数据,基于形状相似距离的K-means算法比采用传统距离的K-means算法,聚类准确度显著提高。

In this paper, we represent vectors as objects of the feature space, and present the relationship between the difference of vectors and shape similarity. With this idea, the K-means algorithm based on the shape similarity distance （SSD-K-means） is proposed. These approaches have been tested on three well - known datasets from the UCI repository. Experiment results show that, in the data processing contains shape message, SSD-K-means can achieve higher accuracy than K-means algorithm with the classical distances.