Advertisement

Classification of Time Series with Optimized Time-Frequency Representations

  • Christoph Heitz
Conference paper
Part of the Studies in Classification, Data Analysis, and Knowledge Organization book series (STUDIES CLASS)

Summary

We address the problem of classifying time series with finite length. In contrast to the usual feature-based classification schemes we use time-frequency representations (TFRs) as non-parametric representations of signals. Since there are infinitely many different TFRs, each yielding a different representation of the same signal, it is possible to adapt the representation to the structure of the signals under consideration. It is shown how, for the problem of classification, the optimum TFR can be found, if the signal classes are given by a set of realizations. Two examples show the advantage of using the optimum representation in the time-frequency domain compared with the original time representation.

Keywords

Time Series Kernel Function Wigner Distribution Classify Time Series Optimum Kernel Parameter 
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.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Baraniuk, R., and Jones, D.L. (1993): A signal-dependent time-frequency representation: Optimal kernel design. IEEE Trans. Accoustics, Speech and Signal Processing, 41 (4), 1589–1602.CrossRefGoogle Scholar
  2. Choi, H., and Williams, W.J. (1989): Improved time-frequency representation of multicomponent signals using exponential kernels. IEEE Trans. Accoustics, Speech and Signal Processing, 37(6), 86, 2–871.Google Scholar
  3. Cohen, L. (1989): Time-frequency-distributions - a review. Proc. IEEE, 77 (7), 941–981.CrossRefGoogle Scholar
  4. Duda, R. and Hart, P. (1973): Pattern Classification and Scene Analysis. Wiley, New York.Google Scholar
  5. Dulfer, B.U. (1993): Klassifikation and Merkmalsextraktion. PhD-Thesis, Department of Physics, University of Freiburg, Germany.Google Scholar
  6. Flandrin, P. (1988): A time-frequency formulation of optimum detection. IEEE Trans. Accoustics, Speech and Signal Processing, 36 (9), 1377–1384.CrossRefGoogle Scholar
  7. Flandrin, P. (1993): Temps-Frequence. Edition Hermes, Paris.Google Scholar
  8. Fukunaga, I. (1990): Introduction to Statistical Pattern Recognition. Academic Press, New York, 2nd edition.Google Scholar
  9. Hartung, J., and Elpelt, B. (1989): Multivariate Statistik. Oldenbourg, Miinchen.Google Scholar
  10. Heitz, C. (1994): Optimized time-frequency representations for the classification and detection of signals. Preprint, submitted for publication.Google Scholar
  11. Heitz, C. (1995): Using optimized time-frequency representations for acoustic quality control of motors. Preprint, submitted for publicationGoogle Scholar
  12. Honerkamp, J. (1994): Stochastic dynamical systems. VCII Publishers, New York.Google Scholar
  13. Jones, D.L. and Parks, T. (1990): A high resolution data-adaptive time- frequency representation. IEEE Trans. Accoustics, Speech and Signal Processing, 38 (12), 2127–2135.CrossRefGoogle Scholar
  14. Kay, S. and Boudreaux-Bartels, G.F. (1985): On the optimality of the Wigner distribution for detection. Proc. of the IEEE Int. Conf. on Acoust., Speech and Signal Processing ICASSP-85. pp. 27.2.1–27.2. 4, Tampa, FL.Google Scholar

Copyright information

© Springer-Verlag Berlin · Heidelberg 1996

Authors and Affiliations

  • Christoph Heitz
    • 1
  1. 1.Center for Data Analysis and Model BuildingUniversity of FreiburgFreiburgGermany

Personalised recommendations