属性分布相似度吸引子传播聚类算法研究

A properties distribution similarity-based affinity propagation algorithm

传统吸引子传播聚类算法对数据类型敏感,文中提出一种改进的吸引子传播聚类算法,将JACCARD系数引入对象间属性分布相似度,并与吸引子传播聚类算法结合。仿真实验结果表明,该算法收敛速度快,聚类精度高,明显提高高维稀疏数据的聚类性能。

The traditional affinity propagation algorithm is sensitive to the type of data. Here we propose an improved affinity propagation algorithm which is based on property distribution similarity. JACCARD is introduced into property distribution similarity, and combined with affinity propagation clustering algorithm. Simulation results show that the method is with high precision and fast convergence, and can improve the clustering properties of high-dimensional sparse data.