Exact Continuous Relaxations for the ℓ0-Regularized Least-Squares Criteria
E. Soubies
SIAM Conference on Imaging Science (IS'18), Bologna, Italian Republic, June 5-8, 2018, session MS37-2.
Several continuous non-convex relaxations of the ℓ0 pseudo-norm have been proposed over the past. In this talk, considering the ℓ0-regularized least-squares minimization problem (ℓ2-ℓ0), I will present theoretical results which allow to compare such relaxations from the perspective of their fidelity to the initial ℓ2-ℓ0 problem. I will exhibit necessary and sufficient conditions on separable penalties approximating the ℓ0 pseudo-norm which ensure that the associated regularized least-squares functional preserves the global minimizers of the initial one and do not add new local minimizers. From these conditions, we get a class of penalties said to be exact regarding to their properties concerning the relaxed functional.
@INPROCEEDINGS(http://bigwww.epfl.ch/publications/soubies1802.html, AUTHOR="Soubies, E.", TITLE="Exact Continuous Relaxations for the $\ell_{0}$-Regularized Least-Squares Criteria", BOOKTITLE="{SIAM} Conference on Imaging Science ({IS'18})", YEAR="2018", editor="", volume="", series="", pages="", address="Bologna, Italian Republic", month="June 5-8,", organization="", publisher="", note="session MS37-2")