在线分段时间序列流：一种有限自动机方法

Segmentation is one of the important tools to mine time series. Existing segmenting methods for time series mainly focus on the static data, and are infeasible under the circumstance of time series stream. A method for adaptively segmenting time series stream is proposed. The method employs finite state automata to analyze and extract the states of time series stream data, and the segmenting is performed according to the states. Extensive empirical experiments, both on synthetic and real datasets, show that the approach achieves great effectiveness on the high speed time series stream, and the quality of the segments is assured