Thresholding RULES and iterative shrinkage/thresholding algorithm: A convergence study. Kowalski, M. In 2014 IEEE International Conference on Image Processing (ICIP), pages 4151–4155, October, 2014. ISSN: 2381-8549doi abstract bibtex Imaging inverse problems can be formulated as an optimization problem and solved thanks to algorithms such as forward-backward or ISTA (Iterative Shrinkage/Thresholding Algorithm) for which non smooth functionals with sparsity constraints can be minimized efficiently. However, the soft thresholding operator involved in this algorithm leads to a biased estimation of large coefficients. That is why a step allowing to reduce this bias is introduced in practice. Indeed, in the statistical community, a large variety of thresholding operators have been studied to avoid the biased estimation of large coefficients; for instance, the non negative Garrote or the the SCAD thresholding. One can associate a non convex penalty to these operators. We study the convergence properties of ISTA, possibly relaxed, with any thresholding rule and show that they correspond to a semi-convex penalty. The effectiveness of this approach is illustrated on image inverse problems.
@inproceedings{kowalski_thresholding_2014,
title = {Thresholding {RULES} and iterative shrinkage/thresholding algorithm: {A} convergence study},
shorttitle = {Thresholding {RULES} and iterative shrinkage/thresholding algorithm},
doi = {10.1109/ICIP.2014.7025843},
abstract = {Imaging inverse problems can be formulated as an optimization problem and solved thanks to algorithms such as forward-backward or ISTA (Iterative Shrinkage/Thresholding Algorithm) for which non smooth functionals with sparsity constraints can be minimized efficiently. However, the soft thresholding operator involved in this algorithm leads to a biased estimation of large coefficients. That is why a step allowing to reduce this bias is introduced in practice. Indeed, in the statistical community, a large variety of thresholding operators have been studied to avoid the biased estimation of large coefficients; for instance, the non negative Garrote or the the SCAD thresholding. One can associate a non convex penalty to these operators. We study the convergence properties of ISTA, possibly relaxed, with any thresholding rule and show that they correspond to a semi-convex penalty. The effectiveness of this approach is illustrated on image inverse problems.},
language = {en},
booktitle = {2014 {IEEE} {International} {Conference} on {Image} {Processing} ({ICIP})},
author = {Kowalski, Matthieu},
month = oct,
year = {2014},
note = {ISSN: 2381-8549},
keywords = {\#Analysis, \#ICIP{\textgreater}14, /unread, Sparse approximation, nonnegative garrote, relaxed ISTA, semi convex optimization},
pages = {4151--4155},
}
Downloads: 0
{"_id":"XpkQQcE8qfgacMhCB","bibbaseid":"kowalski-thresholdingrulesanditerativeshrinkagethresholdingalgorithmaconvergencestudy-2014","author_short":["Kowalski, M."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","title":"Thresholding RULES and iterative shrinkage/thresholding algorithm: A convergence study","shorttitle":"Thresholding RULES and iterative shrinkage/thresholding algorithm","doi":"10.1109/ICIP.2014.7025843","abstract":"Imaging inverse problems can be formulated as an optimization problem and solved thanks to algorithms such as forward-backward or ISTA (Iterative Shrinkage/Thresholding Algorithm) for which non smooth functionals with sparsity constraints can be minimized efficiently. However, the soft thresholding operator involved in this algorithm leads to a biased estimation of large coefficients. That is why a step allowing to reduce this bias is introduced in practice. Indeed, in the statistical community, a large variety of thresholding operators have been studied to avoid the biased estimation of large coefficients; for instance, the non negative Garrote or the the SCAD thresholding. One can associate a non convex penalty to these operators. We study the convergence properties of ISTA, possibly relaxed, with any thresholding rule and show that they correspond to a semi-convex penalty. The effectiveness of this approach is illustrated on image inverse problems.","language":"en","booktitle":"2014 IEEE International Conference on Image Processing (ICIP)","author":[{"propositions":[],"lastnames":["Kowalski"],"firstnames":["Matthieu"],"suffixes":[]}],"month":"October","year":"2014","note":"ISSN: 2381-8549","keywords":"#Analysis, #ICIP\\textgreater14, /unread, Sparse approximation, nonnegative garrote, relaxed ISTA, semi convex optimization","pages":"4151–4155","bibtex":"@inproceedings{kowalski_thresholding_2014,\n\ttitle = {Thresholding {RULES} and iterative shrinkage/thresholding algorithm: {A} convergence study},\n\tshorttitle = {Thresholding {RULES} and iterative shrinkage/thresholding algorithm},\n\tdoi = {10.1109/ICIP.2014.7025843},\n\tabstract = {Imaging inverse problems can be formulated as an optimization problem and solved thanks to algorithms such as forward-backward or ISTA (Iterative Shrinkage/Thresholding Algorithm) for which non smooth functionals with sparsity constraints can be minimized efficiently. However, the soft thresholding operator involved in this algorithm leads to a biased estimation of large coefficients. That is why a step allowing to reduce this bias is introduced in practice. Indeed, in the statistical community, a large variety of thresholding operators have been studied to avoid the biased estimation of large coefficients; for instance, the non negative Garrote or the the SCAD thresholding. One can associate a non convex penalty to these operators. We study the convergence properties of ISTA, possibly relaxed, with any thresholding rule and show that they correspond to a semi-convex penalty. The effectiveness of this approach is illustrated on image inverse problems.},\n\tlanguage = {en},\n\tbooktitle = {2014 {IEEE} {International} {Conference} on {Image} {Processing} ({ICIP})},\n\tauthor = {Kowalski, Matthieu},\n\tmonth = oct,\n\tyear = {2014},\n\tnote = {ISSN: 2381-8549},\n\tkeywords = {\\#Analysis, \\#ICIP{\\textgreater}14, /unread, Sparse approximation, nonnegative garrote, relaxed ISTA, semi convex optimization},\n\tpages = {4151--4155},\n}\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n","author_short":["Kowalski, M."],"key":"kowalski_thresholding_2014","id":"kowalski_thresholding_2014","bibbaseid":"kowalski-thresholdingrulesanditerativeshrinkagethresholdingalgorithmaconvergencestudy-2014","role":"author","urls":{},"keyword":["#Analysis","#ICIP\\textgreater14","/unread","Sparse approximation","nonnegative garrote","relaxed ISTA","semi convex optimization"],"metadata":{"authorlinks":{}},"downloads":0,"html":""},"bibtype":"inproceedings","biburl":"https://bibbase.org/zotero/zzhenry2012","dataSources":["nZHrFJKyxKKDaWYM8"],"keywords":["#analysis","#icip\\textgreater14","/unread","sparse approximation","nonnegative garrote","relaxed ista","semi convex optimization"],"search_terms":["thresholding","rules","iterative","shrinkage","thresholding","algorithm","convergence","study","kowalski"],"title":"Thresholding RULES and iterative shrinkage/thresholding algorithm: A convergence study","year":2014}