摘要
根据时间序列数据维度高、实值有序、数据间存在自相关性等特点,对时间序列分类过程进行研究。研究了当前比较流行的时间序列分类方法;从图像处理的角度出发,提出了一种将图片信息转化为时间序列数据的ITTS方法。shapelets作为最能够表示一条时间序列的子序列,随着时间的推移,这个特征序列可能会动态地发生变化。基于这样的思想,提出了一种基于动态发现shapelets的增量式时间序列分类算法IPST。该算法能够较好地动态发现当前最优的k个shapelets,从而提高时间序列分类的准确度。得到的shapelets集合还可以与多个传统的分类器结合,从而获得更佳的分类效果。
This paper focused on research of time series classification according to time series features of high dimensionality,ordered real-valued variables,autocorrelation and so on.From the perspective of image processing,this paper proposed a method for ITTS to transform image information to time series.As the best approach to show a subsequence of time series,shapelets would change dynamically as time goes on.Considering this thought,this paper presented time series classification algorithm based on dynamically finding shapelets which is named IPST algorithm.IPST algorithm dynamically discoveries current optimal kshapelets well,so as to improve the accuracy of time series classification.The discovered shapelets can be also used with state-of-art classification algorithms,leading to better performance.
出处
《计算机科学》
CSCD
北大核心
2016年第5期257-260,293,共5页
Computer Science
基金
国家自然科学基金项目(61563001)
国家民委科研基金项目(14BFZ008)资助
