Sparsity and the Bayesian perspective

Sparsity has recently been introduced in cosmology for weak-lensing and cosmic microwave background (CMB) data analysis for different applications such as denoising, component separation, or inpainting (i.e., filling the missing data or the mask). Although it gives very nice numerical results, CMB sparse inpainting has been severely criticized by top researchers in cosmology using arguments derived from a Bayesian perspective. In an attempt to understand their point of view, we realize that interpreting a regularization penalty term as a prior in a Bayesian framework can lead to erroneous conclusions. This paper is by no means against the Bayesian approach, which has proven to be very useful for many applications, but warns against a Bayesian-only interpretation in data analysis, which can be misleading in some cases.

[1]  Axthonv G. Oettinger,et al.  IEEE Transactions on Information Theory , 1998 .

[2]  D. Donoho.  For most large underdetermined systems of equations, the minimal 𝓁1‐norm near‐solution approximates the sparsest near‐solution , 2006 .

[3]  Emmanuel J. Candès,et al.  Near-Optimal Signal Recovery From Random Projections: Universal Encoding Strategies? , 2004, IEEE Transactions on Information Theory.

[4]  Jean-Luc Starck,et al.  Morphological Component Analysis and Inpainting on the Sphere: Application in Physics and Astrophysics , 2007 .

[5]  M. Nikolova.  Model distortions in Bayesian MAP reconstruction , 2007 .

[6]  R. Trotta.  Bayes in the sky: Bayesian inference and model selection in cosmology , 2008, 0803.4089.

[7]  J. Fadili,et al.  CMB data analysis and sparsity , 2008, 0804.1295.

[8]  M. Szydłowski,et al.  Effective dynamics of the closed loop quantum cosmology , 2009, 0906.2503.

[9]  G. Efstathiou,et al.  Large-angle correlations in the cosmic microwave background , 2009, 0911.5399.

[10]  University of Oslo,et al.  BAYESIAN COMPONENT SEPARATION AND COSMIC MICROWAVE BACKGROUND ESTIMATION FOR THE FIVE-YEAR WMAP TEMPERATURE DATA , 2009, 0903.4311.

[11]  M. Kawasaki,et al.  Probing the primordial power spectra with inflationary priors , 2009, 0911.5191.

[12]  J.-L. Starck,et al.  Measuring the integrated Sachs-Wolfe effect , 2010, 1010.2192.

[13]  O. Cappé,et al.  Bayesian model comparison in cosmology with Population Monte Carlo , 2009, 0912.1614.

[14]  Volkan Cevher,et al.  Low-Dimensional Models for Dimensionality Reduction and Signal Recovery: A Geometric Perspective , 2010, Proceedings of the IEEE.

[15]  B. Wandelt,et al.  LOCAL NON-GAUSSIANITY IN THE COSMIC MICROWAVE BACKGROUND THE BAYESIAN WAY , 2010, 1010.1254.

[16]  Jean-Luc Starck,et al.  Reconstruction of the cosmic microwave background lensing for Planck , 2010 .

[17]  M. Hobson,et al.  Powellsnakes II: a fast Bayesian approach to discrete object detection in multi-frequency astronomical data sets , 2011, 1112.4886.

[18]  T. Louis,et al.  Filling in CMB map missing data using constrained Gaussian realizations , 2011, 1109.0286.

[19]  P. Berkes,et al.  Improved constraints on cosmological parameters from SNIa data , 2011, 1102.3237.

[20]  D. Herranz,et al.  A Bayesian technique for the detection of point sources in CMB maps , 2011 .

[21]  BAYESIAN NOISE ESTIMATION FOR NON-IDEAL COSMIC MICROWAVE BACKGROUND EXPERIMENTS , 2012 .

[22]  J. Starck,et al.  A hybrid approach to cosmic microwave background lensing reconstruction from all-sky intensity maps , 2012, 1201.5779.

[23]  N. Mandolesi,et al.  HARMONIC IN-PAINTING OF COSMIC MICROWAVE BACKGROUND SKY BY CONSTRAINED GAUSSIAN REALIZATION , 2012, 1202.0188.

[24]  J. Fadili,et al.  Low-ℓ CMB analysis and inpainting , 2012, 1210.6587.

[25]  J.-L. Starck,et al.  Sparse component separation for accurate cosmic microwave background estimation , 2012, 1206.1773.

[26]  D. Parkinson,et al.  Optimizing future dark energy surveys for model selection goals , 2011, 1111.1870.

[27]  Laurent Eyer,et al.  Astrostatistics and Data Mining , 2014 .