Best Basis Representations with Prior Statistical Models

  • David Leporini
  • Jean-Christophe Pesquet
  • Hamid Krim
Part of the Lecture Notes in Statistics book series (LNS, volume 141)


Wavelet packets and local trigonometric bases provide an efficient framework and fast algorithms to obtain a “best representation” of a deterministic signal. Applying these deterministic search techniques to stochastic signals may, however, lead to statistically unreliable results. In this chapter, we revisit this problem and introduce prior models on the underlying signal in noise. We propose several techniques to derive the prior parameters and develop a Bayesian-based approach to the best basis problem. As illustrated by applications to signal denoising, this leads to reduced estimation errors while preserving the classical tree search algorithm.


Probability Density Function Reconstruction Error Wavelet Packet Prior Model Decomposition Tree 
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. [ASS98]
    Abramovich, F., Sapatinas, T. & Silverman, B.W. (1998a). Wavelet thresholding via a Bayesian approach. J. Roy. Statist Soc. B 60, 725–749.MATHCrossRefGoogle Scholar
  2. [BD95]
    J.B. Buckheit and D.L. Donoho. Wavelab and Reproducible Research. Stanford University, 1995.Google Scholar
  3. [Ber85]
    J. Berger. Statistical Decision Theory and Bayesian Analysis. Springer Verlag, New York, 1985.MATHGoogle Scholar
  4. [CCL96]
    Q. Cheng, R. Chen, and T. Li. Simultaneous Wavelet Estimation and Deconvolution of Reflection Seismic Signals. IEEE Trans. Geoscience and Rem. Sens., 34:377–384, 1996.CrossRefGoogle Scholar
  5. [CDV93]
    A. Cohen, I. Daubechies, and P. Vial. Wavelets on the interval and fast wavelet algorithms. Applied and Computational Harmonic Analysis, 1:54–81, Dec. 1993.MathSciNetMATHCrossRefGoogle Scholar
  6. [CGI96]
    F. Champagnat, Y. Goussard, and J. Idier. Unsupervised Deconvolution of Sparse Spike Trains using Stochastic Approximation. IEEE Trans. Signal Processing, 44:2988–2998, 1996.CrossRefGoogle Scholar
  7. [CI95]
    F. Champagnat and J. Idier. An alternative to standard maximum likelihood for Gaussian mixtures. In Proc. IEEE Conf. Acoust., Speech, Signal Processing, pages 2020-2023, Detroit, USA, May 9–12 1995.Google Scholar
  8. [CKM97]
    H. A. Chipman, E. D. Kolaczyck, and R. E. McCulloch. Adaptive Bayesian Wavelet Shrinkage. J. Amer. Statist. Assoc, 92:1413–1421, 1997.MATHCrossRefGoogle Scholar
  9. [CNB98]
    M. S. Crouse, R. D. Nowak, and R. G. Baraniuk. Wavelet-Based Statistical Signal Processing Using Hidden Markov Models. IEEE Trans. Signal Processing, 46:886–902, 1998.MathSciNetCrossRefGoogle Scholar
  10. [CPV98]
    M. Clyde, G. Parmigiani, and B. Vidakovic. Multiple Shrinkage and Subset Selection in Wavelets. Biometrika, 85:391–401, 1998.MathSciNetMATHCrossRefGoogle Scholar
  11. [CW92]
    R. R. Coifman and M. V. Wickerhauser. Entropy-based algorithms for best basis selection. IEEE Trans. Informat. Theory, IT-38:713–718, Mar. 1992.CrossRefGoogle Scholar
  12. [DJ94a]
    D. L. Donoho and I. M. Johnstone. Ideal denoising in an orthogonal basis chosen from a library of bases. C. R. Acad. Sci. Parts, 319:1317–1322, 1994.MathSciNetMATHGoogle Scholar
  13. [DJ94b]
    D. L. Donoho and I. M. Johnstone. Ideal spatial adaptation by wavelet shrinkage. Biometrika, 81:425–455, Sept. 1994.MathSciNetMATHCrossRefGoogle Scholar
  14. [DLR77]
    A. P. Dempster, N. M. Laird, and D. B. Rubin. Maximum Likelihood from Incomplete Data via the EM Algorithm. J. R. Statist. Soc. B, 39:1–38, 1977.MathSciNetMATHGoogle Scholar
  15. [Hub67]
    P. Huber. The behaviour of maximum-likelihood estimates under nonstandard conditions. In Proc. Berkeley Symp. on Math. Stat. and Prob., volume 1, pages 73–101, 1967.Google Scholar
  16. [HV94]
    C. Herley and M. Vetterli. Orthogonal time-varying filter banks and wavelet packets. IEEE Trans. Signal Processing, 42:2650–2663, Oct. 1994.CrossRefGoogle Scholar
  17. [KMDW95]
    H. Krim, S. Mallat, D. Donoho, and A. Willsky. Best basis algorithm for signal enhancement. In Proc. IEEE Conf. Acoust., Speech, Signal Processing, Detroit, MI, May 1995.Google Scholar
  18. [KP95]
    H. Krim and J.-C. Pesquet. On the statistics of best bases criteria. In A. Antoniadis, editor, Wavelets and statistics. Lecture Notes in Statistics, Springer Verlag, 1995.Google Scholar
  19. [Lep98]
    D. Leporini. Modélisation Statistique et Paquets d’Ondelettes: Application au Débruitage de Signaux Transitoires d’Acoustique Sous-Marine. PhD thesis, Université Paris XI, Sept. 1998.Google Scholar
  20. [Mal89]
    S. Mallat. A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans. Patt. Anal. Mach. Intell, PAMI-11:674–693, Jul. 1989.CrossRefGoogle Scholar
  21. [NH98]
    R. M. Neal and G. E. Hinton. A new view of the EM algorithm that justifies incremental and other variants. In M.I. Jordan, editor, Learning in Graphical Models. Kluwer, 1998.Google Scholar
  22. [PKLH96]
    J.-C. Pesquet, H. Krim, D. Leporini, and E. Hamman. Bayesian Approach to Best Basis Selection. In Proc. IEEE Conf. Acoust., Speech, Signal Processing, pages 2634-2638, Atlanta, USA, May 7-9 1996.Google Scholar
  23. [Sai94]
    N. Saito. Local feature extraction and its applications using a library of bases. PhD thesis, Yale University, Dec. 1994.Google Scholar
  24. [Vid98]
    B. Vidakovic. Nonlinear wavelet shrinkage with Bayes rules and Bayes factors. J. Amer. Statist. Assoc., 93:173–179, 1998.MathSciNetMATHCrossRefGoogle Scholar
  25. [Zac81]
    S. Zacks. Parametric Statistical Inference. International Series in Nonlinear Mathematics, Pergamon Press, New York, 1981.MATHGoogle Scholar

Copyright information

© Springer Science+Business Media New York 1999

Authors and Affiliations

  • David Leporini
  • Jean-Christophe Pesquet
  • Hamid Krim

There are no affiliations available

Personalised recommendations