IMPROVING TIME SERIES CLASSIFICATION ACCURACY: COMBINING GLOBAL AND LOCAL INFORMATION IN THE SIMILARITY CRITERION
Keywords:Time series, Classification, Complexity, Shape information, Similarity measure
Given the widespread use of time series classification in many domains, how to improve the accuracy of classification has attracted considerable focus. In this paper, a new similarity measure (SIMscl) based on the global and local information has been proposed for improving the precision rate of one nearest neighbor (1NN) classifier. Specifically, the global information records the intrinsic properties of time series, and is reflected by two indicators: the shape information and the complexity; the local information pays attention to the exact match of value, and is realized by LB_keogh. Simultaneously, a method based on multi-scale discrete haar wavelet transform, key point extraction, and symbolization has been put forward to extract the shape information. To test the efficacy of the proposed shape similarity SIMshape and hybrid similarity SIMscl, the experiments are conducted on two data sets: star light curve and beef. Experimental evaluations show that SIMshape can deal with some time series misclassified by Euclidean Distance (ED), LB_keogh, and Complexity Invariant Distance (CID), and SIMscl has higher precision than ED, LB_keogh, and CID in time series 1NN classification.
Once a paper is accepted for publication, the author is assumed to have transferred its copyright to the Publisher. The Publisher will not, however, put any limitation on the personal freedom of the author to use material from the paper in other publications. From September 2019 it is required that authors explicitly sign a copyright release form before their paper gets published. The Author Copyright Release form can be found here