• DocumentCode
    2189555
  • Title

    Learning the solution sparsity of an ill-posed linear inverse problem with the Variational Garrote

  • Author

    Andersen, Michael Riis ; Hansen, Sofie Therese ; Hansen, Lars Kai

  • Author_Institution
    Dept. of Appl. Math. & Comput. Sci., Tech. Univ. of Denmark, Lyngby, Denmark
  • fYear
    2013
  • fDate
    22-25 Sept. 2013
  • Firstpage
    1
  • Lastpage
    6
  • Abstract
    The Variational Garrote is a promising new approach for sparse solutions of ill-posed linear inverse problems (Kappen and Gomez, 2012). We reformulate the prior of the Variational Garrote to follow a simple Binomial law and assign a Beta hyper-prior on the parameter. With the new prior the Variational Garrote, we show, has a wide range of parameter values for which it at the same time provides low test error and high retrieval of the true feature locations. Furthermore, the new form of the prior and associated hyper-prior leads to a simple update rule in a Bayesian variational inference scheme for its hyperparameter. As a second contribution we provide evidence that the new procedure can improve on cross-validation of the parameters and we find that the new formulation of the prior outperforms the original formulation when both are cross-validated to determine hyperparameters.
  • Keywords
    Bayes methods; binomial distribution; inverse problems; variational techniques; Bayesian variational inference scheme; Beta hyper-prior; hyperparameter; ill-posed linear inverse problem; parameter cross-validation; parameter values; simple binomial law; solution sparsity learning; true feature locations; update rule; variational Garrote; Bayes methods; Data models; Equations; Mathematical model; Mean square error methods; Noise; Vectors; Ill-posed inverse problem; Variational Garrote; linear regression;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Machine Learning for Signal Processing (MLSP), 2013 IEEE International Workshop on
  • Conference_Location
    Southampton
  • ISSN
    1551-2541
  • Type

    conf

  • DOI
    10.1109/MLSP.2013.6661919
  • Filename
    6661919