Iterative incremental clustering of time series

摘要: We present a novel anytime version of partitional clustering algorithm, such as k-Means and EM, for time series. The algorithm works by leveraging off the multi-resolution property wavelets. dilemma choosing initial centers is mitigated initializing at each approximation level, using final returned coarser representations. In addition to casting algorithms algorithms, this approach has two other very desirable properties. By working lower dimensionalities we can efficiently avoid local minima. Therefore, quality usually better than batch algorithm. addition, even if run completion, our much faster its counterpart. explain, empirically demonstrate these surprising properties with comprehensive experiments on several publicly available real data sets. further that be generalized framework broader range or mining problems.