【摘要】 针对传统的基于隐Markov模型(HMM)的聚类算法在时间序列聚类的不足,提出了一种新的基于HMM的分层时间序列聚类算法HBHCTS,旨在提高聚类质量,同时对聚类结果给出类的表示·HBHCTS算法应用HMM对时间序列进行建模,并按照“最相似”的原则得到序列所对应的初始模型集,进而对这些初始模型合并更新及迭代得到聚类结果·实验中主要研究了聚类正确率与序列长度及模型距离的关系,结果表明HBHCTS算法比传统的基于HMM的聚类算法准确性高·更多还原

【Abstract】 In this paper, a novel hidden Markov model (HMM)-based hierarchical time-series clustering algorithm HBHCTS is proposed, because of the disadvantage of traditional HMM-based clustering algorithms for time-series. The main purpose is to improve clustering quality and represent the clusters easily at the same time. In HBHCTS, HMMs are built from time-series, and the initial models are obtained according to the most similarity, and then the process of merging and updating initial models is iterated until the final result is obtained. In the experiment, the relation between correctness rate and the length of a sequence, the relation between correctness rate and the model distance are researched. The results show that the HBHCTS can achieve better performance in correctness rate than the traditional HMM-based clustering algorithm.更多还原