Integrating Data-Driven Segmentation, Local Feature Extraction and Fisher Kernel Encoding to Improve Time Series Classification
- 211 Downloads
The uniform sampling strategy is widely used in time series segmentation, but unable to handle time warping problem or preserve the latent patterns in time series. To solve these shortcomings, a brand new data-driven segmentation method is proposed, which could segment time series into subsequences with different lengths adaptively. Then a time series classification method under the bag-of-word framework is proposed. Two kinds of mutually complementary features, i.e., interval feature and normal cloud model feature, are extracted from subsequences. And then time series are encoded into Fisher Vectors. Finally, a linear support vector machine is used as the classifier. Experiments on 43 UCR datasets show that the newly proposed method has promising classification accuracies comparing with state of the art baselines. Moreover, due to the data-driven segmentation and timesaving local feature extraction, the method has low time complexity, which is also demonstrated in the experiments.
KeywordsTime series classification Data-driven segmentation Fisher vector Normal cloud model
This study was supported by the National Natural Science Foundation of China (U1664264, U1509203).
- 1.Berndt DJ, Clifford J (1994) Using dynamic time warping to find patterns in time series. In: Working notes of the knowledge discovery in databases workshopGoogle Scholar
- 2.Shen J et al (2016) A novel similarity measure model for multivariate time series based on LMNN and DTW. Neural Process Lett 2016:1–13Google Scholar
- 4.Vlachos M, Kollios G, Gunopulos D (2002) Discovering similar multidimensional trajectories. In: International conference on data engineeringGoogle Scholar
- 8.Gudmundsson S, Runarsson T.P, Sigurdsson S (2008) Support vector machines and dynamic time warping for time series. In: IEEE international joint conference on neural networksGoogle Scholar
- 13.Lei Y, He Z, Zi Y (2010) Application of an intelligent classification method to mechanical fault diagnosis. Radiat Effects Defects Solids 2010(36):9941–9948Google Scholar
- 20.Keogh E.J et al (2001) An online algorithm for segmenting time series. In: IEEE international conference on data miningGoogle Scholar
- 24.Jaakkola TS, Haussler D (1998) Exploiting generative models in discriminative classifiers. Adv Neural Inf Process Syst 1998(11):487–493Google Scholar
- 25.Perronnin F, Nchez J, Mensink T (2010) Improving the fisher kernel for large-scale image classification. In: European conference on computer visionGoogle Scholar
- 26.Perronnin F, Dance C (2007) Fisher Kernels on visual vocabularies for image categorization. In: IEEE conference on computer vision and pattern recognitionGoogle Scholar
- 27.Chen Y, Keogh E, Hu B, Begum N, Bagnall A, Mueen A, Batista G (2015) The UCR time series classification archive. www.cs.ucr.edu/~eamonn/time_series_data/
- 28.Rakthanmanon T, Keogh E (2013) Fast shapelets: a scalable algorithm for discovering time series shapelets. In: Proceedings of the 13th SIAM international conference on data miningGoogle Scholar