Model-Based Time Series Classification

  • Alexios Kotsifakos
  • Panagiotis Papapetrou
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8819)


We propose MTSC, a filter-and-refine framework for time series Nearest Neighbor (NN) classification. Training time series belonging to certain classes are first modeled through Hidden Markov Models (HMMs). Given an unlabeled query, and at the filter step, we identify the top K models that have most likely produced the query. At the refine step, a distance measure is applied between the query and all training time series of the top K models. The query is then assigned with the class of the NN. In our experiments, we first evaluated the NN classification error rate of HMMs compared to three state-of-the-art distance measures on 45 time series datasets of the UCR archive, and showed that modeling time series with HMMs achieves lower error rates in 30 datasets and equal error rates in 4. Secondly, we compared MTSC with Cross Validation defined over the three measures on 33 datasets, and we observed that MTSC is at least as good as the competitor method in 23 datasets, while achieving competitive speedups, showing its effectiveness and efficiency.


Time Series Distance Measure Hide Markov Model Cross Validation Near Neighbor 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Agrawal, R., Faloutsos, C., Swami, A.: Efficient similarity search in sequence databases. In: Lomet, D.B. (ed.) FODO 1993. LNCS, vol. 730, pp. 69–84. Springer, Heidelberg (1993)CrossRefGoogle Scholar
  2. 2.
    Assent, I., Wichterich, M., Krieger, R., Kremer, H., Seidl, T.: Anticipatory dtw for efficient similarity search in time series databases. PVLDB 2(1), 826–837 (2009)Google Scholar
  3. 3.
    Athitsos, V., Hadjieleftheriou, M., Kollios, G., Sclaroff, S.: Query-sensitive embeddings. In: SIGMOD, pp. 706–717 (2005)Google Scholar
  4. 4.
    Athitsos, V., Papapetrou, P., Potamias, M., Kollios, G., Gunopulos, D.: Approximate embedding-based subsequence matching of time series. In: SIGMOD, pp. 365–378 (2008)Google Scholar
  5. 5.
    Baum, L.E., Petrie, T., Soules, G., Weiss, N.: A maximization technique occurring in the statistical analysis of probabilistic functions of markov chains. The Annals of Mathematical Statistics 41(1), 164–171 (1970)MathSciNetCrossRefzbMATHGoogle Scholar
  6. 6.
    Bellman, R.: The theory of dynamic programming. Bull. Amer. Math. Soc. 60(6), 503–515 (1954)MathSciNetCrossRefzbMATHGoogle Scholar
  7. 7.
    Chen, H., Tang, F., Tino, P., Yao, X.: Model-based kernel for efficient time series analysis. In: SIGKDD, pp. 392–400 (2013)Google Scholar
  8. 8.
    Chen, L., Ng, R.: On the marriage of l p-norms and edit distance. In: VLDB, pp. 792–803 (2004)Google Scholar
  9. 9.
    Chen, L., Özsu, M.T.: Robust and fast similarity search for moving object trajectories. In: SIGMOD, pp. 491–502 (2005)Google Scholar
  10. 10.
    Chen, Y., Nascimento, M.A., Chin, B., Anthony, O., Tung, K.H.: Spade: On shape-based pattern detection in streaming time series. In: ICDE, pp. 786–795 (2007)Google Scholar
  11. 11.
    F. Ferraty and P. Vieu. Curves discrimination: a nonparametric functional approach. Computational Statistics and Data Analysis, 44(1-2):161–173, 2003.Google Scholar
  12. 12.
    Frentzos, E., Gratsias, K., Theodoridis, Y.: Index-based most similar trajectory search. In: ICDE, pp. 816–825 (2007)Google Scholar
  13. 13.
    Ghassempour, S., Girosi, F., Maeder, A.: Clustering multivariate time series using hidden markov models. International Journal of Environmental Research and Public Health 11(3), 2741–2763 (2014)CrossRefGoogle Scholar
  14. 14.
    González, J., Muñoz, A.: Representing functional data using support vector machines. In: Ruiz-Shulcloper, J., Kropatsch, W.G. (eds.) CIARP 2008. LNCS, vol. 5197, pp. 332–339. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  15. 15.
    Hallak, A., Di-Castro, D., Mannor, S.: Model selection in markovian processes. In: ICML (2013)Google Scholar
  16. 16.
    Keogh, E.: Exact indexing of dynamic time warping. In: VLDB, pp. 406–417 (2002)Google Scholar
  17. 17.
    Keogh, E., Zhu, Q., Hu, B., Hao, Y., Xi, X., Wei, L., Ratanamahatana, C.: The UCR time series classification/clustering homepage,
  18. 18.
    Kotsifakos, A., Athitsos, V., Papapetrou, P., Hollmén, J., Gunopulos, D.: Model-based search in large time series databases. In: PETRA (2011)Google Scholar
  19. 19.
    Kotsifakos, A., Papapetrou, P., Hollmén, J., Gunopulos, D.: A subsequence matching with gaps-range-tolerances framework: A query-by-humming application. PVLDB 4(11), 761–771 (2011)Google Scholar
  20. 20.
    Kotsifakos, A., Papapetrou, P., Hollmén, J., Gunopulos, D., Athitsos, V.: A survey of query-by-humming similarity methods. PETRA, 5:1–5:4 (2012)Google Scholar
  21. 21.
    Kotsifakos, A., Papapetrou, P., Hollmén, J., Gunopulos, D., Athitsos, V., Kollios, G.: Hum-a-song: a subsequence matching with gaps-range-tolerances query-by-humming system. PVLDB 5(12), 1930–1933 (2012)Google Scholar
  22. 22.
    Kruskall, J.B., Liberman, M.: The symmetric time warping algorithm: From continuous to discrete. In: Time Warps. Addison-Wesley (1983)Google Scholar
  23. 23.
    Lafferty, J.D., McCallum, A., Pereira, F.C.N.: Conditional random fields: Probabilistic models for segmenting and labeling sequence data. In: ICML, pp. 282–289 (2001)Google Scholar
  24. 24.
    Lemire, D.: Faster retrieval with a two-pass dynamic-time-warping lower bound. Pattern recognition 42(9), 2169–2180 (2009)CrossRefzbMATHGoogle Scholar
  25. 25.
    Lin, J., Keogh, E., Lonardi, S., Chiu, B.: A symbolic representation of time series, with implications for streaming algorithms. In: SIGMOD Workshop DMKD, pp. 2–11 (2003)Google Scholar
  26. 26.
    Marteau, P.-F.: Time warp edit distance with stiffness adjustment for time series matching. Pattern Analysis and Machine Intelligence 31(2), 306–318 (2009)CrossRefGoogle Scholar
  27. 27.
    Oates, T., Firoiu, L., Cohen, P.R.: Clustering time series with hidden markov models and dynamic time warping. In: In Proceedings of the IJCAI, pp. 17–21 (1999)Google Scholar
  28. 28.
    Pikrakis, A., Theodoridis, S., Kamarotos, D.: Classification of musical patterns using variable duration hidden Markov models. Transactions on Audio, Speech, and Language Processing 14(5), 1795–1807 (2006)CrossRefGoogle Scholar
  29. 29.
    Rabiner, L.: A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE 77(2), 257–286 (1989)CrossRefGoogle Scholar
  30. 30.
    Sakoe, H., Chiba, S.: Dynamic programming algorithm optimization for spoken word recognition. Transactions on Acoustics, Speech and Signal Processing 26, 43–49 (1978)CrossRefzbMATHGoogle Scholar
  31. 31.
    A. Stefan, V. Athitsos, and G. Das. The move-split-merge metric for time series. Transactions on Knowledge and Data Engineering (2012)Google Scholar
  32. 32.
    Vlachos, M., Kollios, G., Gunopulos, D.: Discovering similar multidimensional trajectories. In: ICDE, pp. 673–684 (2002)Google Scholar
  33. 33.
    Wang, S.B., Quattoni, A., Morency, L.-P., Demirdjian, D., Darrell, T.: Hidden conditional random fields for gesture recognition. In: CVPR, pp. 1521–1527 (2006)Google Scholar
  34. 34.
    Wang, X., Mueen, A., Ding, H., Trajcevski, G., Scheuermann, P., Keogh, E.J.: Experimental comparison of representation methods and distance measures for time series data. Data Minining and Knowledge Discovery 26(2), 275–309 (2013)MathSciNetCrossRefGoogle Scholar
  35. 35.
    Ye, L., Keogh, E.: Time series shapelets: a novel technique that allows accurate, interpretable and fast classification. Data Mining and Knowledge Discovery 22(1-2), 149–182 (2011)MathSciNetCrossRefzbMATHGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Alexios Kotsifakos
    • 1
  • Panagiotis Papapetrou
    • 2
  1. 1.Department of Computer Science and EngineeringUniversity of Texas at ArlingtonUSA
  2. 2.Department of Computer and Systems SciencesStockholm UniversitySweden

Personalised recommendations