Advertisement

Model Selection in the Sparsity Context for Inverse Problems in Bayesian Framework

  • Mircea Dumitru
  • Li Wang
  • Ali Mohammad-Djafari
  • Nicolas Gac
Conference paper
Part of the Springer Proceedings in Mathematics & Statistics book series (PROMS, volume 239)

Abstract

The Bayesian approach is considered for inverse problems with a typical forward model accounting for errors and a priori sparse solutions. Solutions with sparse structure are enforced using heavy-tailed prior distributions. The particular case of such prior expressed via normal variance mixtures with conjugate laws for the mixing distribution is the main interest of this paper. Such a prior is considered in this paper, namely, the Student-t distribution. Iterative algorithms are derived via posterior mean estimation. The mixing distribution parameters appear in updating equations and are also used for the initialization. For the choice of mixing distribution parameters, three model selection strategies are considered: (i) parameters approximating the mixing distribution with Jeffrey law, i.e., keeping the mixing distribution well defined but as close as possible to the Jeffreys priors, (ii) based on the prior distribution form, fixing the parameters corresponding to the form inducing the most sparse solution and (iii) based on the sparsity mechanism, fixing the hyperparameters using the statistical measures of the mixing and prior distribution. For each strategy of model selection, the theoretical advantages and drawbacks are discussed and the corresponding simulations are reported for a 1D direct sparsity application in a biomedical context. We show that the third strategy seems to provide the best parameter selection strategy for this context.

Keywords

Inverse problems Gaussian scale mixtures Sparsity enforcing Parameter selection 

References

  1. 1.
    Mohammad-Djafari, A.: EURASIP J. Adv. Signal Process. 52 (2012).  https://doi.org/10.1186/1687-6180-2012-52
  2. 2.
    Babacan, S.D., Nakajima, S., Do, M.N.: IEEE Trans. Signal Process. 14, 2906 (2014).  https://doi.org/10.1109/TSP.2014.2319775CrossRefGoogle Scholar
  3. 3.
    Dobigeon, N., Hero, A.O., Tourneret, J.Y.: IEEE Trans. Image Process. 18, 2059 (2009).  https://doi.org/10.1109/TIP.2009.2024067MathSciNetCrossRefGoogle Scholar
  4. 4.
    Smídl, V., Quinn, A.: The Variational Bayes Method in Signal Processing. Signals and communication technology, 1st edn. Springer, Heidelberg (2006).  https://doi.org/10.1007/3-540-28820-1
  5. 5.
    Beal, M.J.: Variational Algorithms for Approximate Bayesian Inference. Ph.D. thesis, University College London (2003)Google Scholar
  6. 6.
    Dumitru, M.: A Bayesian Approach for Periodic Components Estimation for Chronobiological Signals. Ph.D. thesis, Université Paris-Saclay (2016)Google Scholar
  7. 7.
    Mohammad-Djafari, A., Dumitru, M.: Digit. Signal Process. 47, 128 (2015).  https://doi.org/10.1016/j.dsp.2015.08.0CrossRefGoogle Scholar
  8. 8.
    Orieux, F., Giovannelli, J.F., Rodet, T.: J. Opt. Soc. Am. A 27(7), 1593 (2010).  https://doi.org/10.1364/JOSAA.27.001593CrossRefGoogle Scholar
  9. 9.
    Dumitru, M., Mohammad-Djafari, A., B.S, S.: EURASIP J. Bioinf. Syst. Biol. 3 (2016).  https://doi.org/10.1186/s13637-015-0033-6

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  • Mircea Dumitru
    • 1
  • Li Wang
    • 1
  • Ali Mohammad-Djafari
    • 1
  • Nicolas Gac
    • 1
  1. 1.Laboratoire des signaux et systèmesGif sur YvetteFrance

Personalised recommendations