SPOQ \ell_p-Over-\ell_q Regularization for Sparse Signal Recovery applied to Mass Spectrometry

DOI10.1109/TSP.2020.3025731arXiv2001.08496MaRDI QIDQ6333369FDOQ6333369

E. Chouzenoux, Afef Cherni, Jean-Christophe Pesquet, Laurent Duval

Publication date: 23 January 2020

Abstract: Underdetermined or ill-posed inverse problems require additional information for ldd{d} sound solutions with tractable optimization algorithms. Sparsity yields consequent heuristics to that matter, with numerous applications in signal restoration, image recovery, or machine learning. Since the

e l l_{0}

count measure is barely tractable, many statistical or learning approaches have invested in computable proxies, such as the

e l l_{1}

norm. However, the latter does not exhibit the desirable property of scale invariance for sparse data. Extending the SOOT Euclidean/Taxicab

e l l_{1}

-over-

e l l_{2}

norm-ratio initially introduced for blind deconvolution, we propose SPOQ, a family of smoothed (approximately) scale-invariant penalty functions. It consists of a Lipschitz-differentiable surrogate for

e l l_{p}

-over-

e l l_{q}

quasi-norm/norm ratios with

p i n,] 0, 2 [

and

q g e 2

. This surrogate is embedded into a novel majorize-minimize trust-region approach, generalizing the variable metric forward-backward algorithm. For naturally sparse mass-spectrometry signals, we show that SPOQ significantly outperforms

e l l_{0}

,

e l l_{1}

, Cauchy, Welsch, SCAD and Celo penalties on several performance measures. Guidelines on SPOQ hyperparameters tuning are also provided, suggesting simple data-driven choices.

Mathematics Subject Classification ID

Signal theory (characterization, reconstruction, filtering, etc.) (94A12)

This page was built for publication: SPOQ $\ell_p$-Over-$\ell_q$ Regularization for Sparse Signal Recovery applied to Mass Spectrometry

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6333369)