According to both domain expert knowledge and empirical evidence, wavelet coefficients of real signals tend to exhibit clustering patterns, in that they contain connected regions of coefficients of similar magnitude (large or small). A wavelet de-noising approach that takes into account such a feature of the signal may in practice outperform other, more vanilla methods, both in terms of the estimation error and visual appearance of the estimates. Motivated by this observation, we present a Bayesian approach to wavelet de-noising, where dependencies between neighbouring wavelet coefficients are a priori modelled via a Markov chain-based prior, that we term the caravan prior. Posterior computations in our method are performed via the Gibbs sampler. Using representative synthetic and real data examples, we conduct a detailed comparison of our approach with a benchmark empirical Bayes de-noising method (due to Johnstone and Silverman). We show that the caravan prior fares well and is therefore a useful addition to the wavelet de-noising toolbox.
Mots-clés : Caravan prior, discrete wavelet transform, Gamma markov chain, Gibbs sampler, regression, wavelet de-noising
@article{PS_2019__23__947_0, author = {Gugushvili, Shota and van der Meulen, Frank and Schauer, Moritz and Spreij, Peter}, title = {Bayesian wavelet de-noising with the caravan prior}, journal = {ESAIM: Probability and Statistics}, pages = {947--978}, publisher = {EDP-Sciences}, volume = {23}, year = {2019}, doi = {10.1051/ps/2019019}, mrnumber = {4046860}, zbl = {1507.62231}, language = {en}, url = {http://www.numdam.org/articles/10.1051/ps/2019019/} }
TY - JOUR AU - Gugushvili, Shota AU - van der Meulen, Frank AU - Schauer, Moritz AU - Spreij, Peter TI - Bayesian wavelet de-noising with the caravan prior JO - ESAIM: Probability and Statistics PY - 2019 SP - 947 EP - 978 VL - 23 PB - EDP-Sciences UR - http://www.numdam.org/articles/10.1051/ps/2019019/ DO - 10.1051/ps/2019019 LA - en ID - PS_2019__23__947_0 ER -
%0 Journal Article %A Gugushvili, Shota %A van der Meulen, Frank %A Schauer, Moritz %A Spreij, Peter %T Bayesian wavelet de-noising with the caravan prior %J ESAIM: Probability and Statistics %D 2019 %P 947-978 %V 23 %I EDP-Sciences %U http://www.numdam.org/articles/10.1051/ps/2019019/ %R 10.1051/ps/2019019 %G en %F PS_2019__23__947_0
Gugushvili, Shota; van der Meulen, Frank; Schauer, Moritz; Spreij, Peter. Bayesian wavelet de-noising with the caravan prior. ESAIM: Probability and Statistics, Tome 23 (2019), pp. 947-978. doi : 10.1051/ps/2019019. http://www.numdam.org/articles/10.1051/ps/2019019/
[1] Wavelets: A package of functions for computing wavelet filters, wavelet transforms and multiresolution analyses. R package version 0.3-0. https://CRAN.R-project.org/package=wavelets.
,[2] Julia: a fresh approach to numerical computing. SIAM Rev. 59 (2017) 65–98. | DOI | MR | Zbl
, , and ,[3] Incorporating information on neighbouring coefficients into wavelet estimation. Sankhyā Ser. B 63 (2001) 127–148. | MR | Zbl
and ,[4] The horseshoe estimator for sparse signals. Biometrika 97 (2010) 465–480. | DOI | MR | Zbl
, and ,[5] Conjugate gamma Markov random fields for modelling nonstationary sources. In , , , (Eds.) Independent Component Analysis and Signal Separation: 7th International Conference, ICA 2007, London, UK, September 9–12, 2007. Proceedings. Springer, Berlin, Heidelberg (2007) pp. 697–705. | DOI | Zbl
and ,[6] Variational and stochastic inference for Bayesian source separation. Digit. Signal Process. 17 (2007) 891–913. | DOI
, and ,[7] Translation-invariant de-noising. In , (Eds.) Wavelets and Statistics. Springer, New York (1995), pp. 125–150. | DOI | Zbl
and ,[8] Wavelet-based statistical signal processing using hidden Markov models. IEEE Trans. Signal Process. 46 (1998) 886–902. | DOI | MR
, and ,[9] Gamma Markov random fields for audio source modeling. IEEE Trans. Audio, Speech, Language Process. 18 (2010) 589–601. | DOI
and ,[10] De-noising by soft-thresholding. IEEE Trans. Inf. Theory 41 (1995) 613–627. | DOI | MR | Zbl
,[11] Ideal spatial adaptation by wavelet shrinkage. Biometrika 81 (1994) 425–455. | DOI | MR | Zbl
and ,[12] Adapting to unknown smoothness via wavelet shrinkage. J. Am. Statist. Assoc. 90 (1995) 1200–1224. | DOI | MR | Zbl
and ,[13] ggmcmc: Analysis of MCMC samples and Bayesian inference. J. Stat. Softw. 70 (2016) 1–20. | DOI
,[14] Sampling-based approaches to calculating marginal densities. J. Am. Stat. Assoc. 85 (1990) 398–409. | DOI | MR | Zbl
and ,[15] Fundamentals of nonparametric Bayesian inference, vol. 44 of Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press, Cambridge (2017). | MR | Zbl
and ,[16] Code accompanying the paper “Bayesian wavelet de-noising with the caravan prior”. Zenodo (2018). | DOI | Numdam | MR | Zbl
, , and ,[17] Fast and scalable non-parametric Bayesian inference for Poisson point processes. Preprint (2018). | arXiv
, , and ,[18] Nonparametric Bayesian volatility learning under microstructure noise. Preprint (2018). | arXiv | MR
, , and ,[19] Nonparametric Bayesian volatility estimation. Edited by , , . 2017 MATRIX Annals. Springer International Publishing, Cham (2019) 279–302. | DOI | MR
, , and ,[20] R programs for empirical Bayes thresholding. J. Stat. Softw. 12 (2005) 1–38. | DOI
and , :[21] Needles and straw in haystacks: empirical Bayes estimates of possibly sparse sequences. Ann. Statist. 32 (2004) 1594–1649. | DOI | MR | Zbl
and ,[22] Empirical Bayes selection of wavelet thresholds. Ann. Statist. 33 (2005) 1700–1752. | DOI | MR | Zbl
and ,[23] Bayesian variable selection in linear regression. With comments by James Berger and C.L. Mallows and with a reply by the authors. J. Amer. Statist. Assoc. 83 (1988) 1023–1036. | MR | Zbl
and ,[24] Discussion on the meeting on ‘Statistical approaches to inverse problems’. J. R. Statist. Soc. B 66 (2004) 628–630.
,[25] Wavelet methods for time series analysis, vol. 4 of Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press, Cambridge (2000). | MR | Zbl
and ,[26] Group sparsity with overlapping partition functions. In 2011 19th European Signal Processing Conference. IEEE (2011) 303–307.
and ,[27] Shrink globally, act locally: sparse Bayesian regularization and prediction. With discussions by , , , , , , , , and . In Vol. 9 of Bayesian statistics. Oxford Univ. Press, Oxford (2011) 501–538. | DOI | MR
and ,[28] Sparse Bayesian learning and the Relevance Vector Machine. J. Mach. Learn. Res. 1 (2001) 211–244. | MR | Zbl
,[29] quantification for the horseshoe (with discussion). Bayesian Anal. 12 (2017) 1221–1274. With a rejoinder by the authors. | MR | Zbl
and[30] The horseshoe estimator: posterior concentration around nearly black vectors. Electron. J. Stat. 8 (2014) 2585–2618. | DOI | MR | Zbl
, and ,[31] waveslim: Basic wavelet routines for one-, two- and three-dimensional signal processing. R package version 1.7.5. Available at: https://CRAN.R-project.org/package=waveslim (2015).
,[32] ggplot2: Elegant graphics for data analysis. Springer-Verlag, New York (2009). | DOI | Zbl
,[33] Bayesian variable selection and regularization for time-frequency surface estimation. J. R. Stat. Soc. Ser. B Stat. Methodol. 66 (2004) 575–589. | DOI | MR | Zbl
, and ,Cité par Sources :