基于原子矩阵的时间序列的相似性模型

Atomic Matrix Based Similarity Model for Time Series

时间序列是由一系列相关的点组成,对此,论文用矩阵来表达时间序列结构,提出时间序列的结构矩阵及原子矩阵,建立以原子矩阵为模板的序列模型,进行时序结构相似性研究,得到一个合理的相似度量值,而矩阵的一些变换可以提高模型的适应性。

Time series is composed of a series of time points,based on which this paper employs the matrix to represent the structure of time series in nature.The authors propose structural matrix and its atomic matrix,build up a time series model based on atomic matrix,and carry through the similarity research on time series,to find a reasonable value for similarity assessment.Furthermore,some transformation on matrix enhances the adaptability of the model.