GPU-based Implementations of MM Algorithms. Application to Spectroscopy Signal Restoration

被引:0
作者
Gharbi, Mouna [1 ]
Chouzenoux, Emilie [1 ]
Pesquet, Jean-Christophe [1 ]
Duval, Laurent [2 ]
机构
[1] Univ Paris Saclay, INRIA, Cent Supelec, CVN, Gif Sur Yvette, France
[2] IFP Energies Nouvelles, Rueil Malmaison, France
来源
29TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO 2021) | 2021年
基金
欧盟地平线“2020”;
关键词
Majorization-Minimization; subspace acceleration; unfolding; regularization parameter; GPU; mass spectrometry; MAXIMUM-ENTROPY; REGULARIZATION;
D O I
10.23919/EUSIPCO54536.2021.9616274
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
Restoration of analytical chemistry data from degraded physical acquisitions is an important task for chemists to obtain accurate component analysis and sound interpretation. The high-dimensional nature of these signals and the large amount of data to be processed call for fast and efficient reconstruction methods. Existing works have primarily relied on optimization algorithms to solve a penalized formulation. Although very powerful, such methods can be computationally heavy, and hyperparameter tuning can be a tedious task for non-experts. Another family of approaches explored recently consists in adopting deep learning to perform the signal recovery task in a supervised fashion. Although fast, thanks to their formulations amenable to GPU implementations, these methods usually need large annotated databases and are not explainable. In this work, we propose to combine the best of both worlds, by proposing unfolded Majorization-Minimization (MM) algorithms with the aim to reach fast and accurate methods for sparse spectroscopy signal restoration. Two state-of-the-art iterative MM algorithms are unfolded onto deep network architectures. This allows both the deployment of GPU-friendly tools for accelerated implementation, as well as the introduction of a supervised learning strategy for tuning automatically the regularization parameter. The effectiveness of our approach is demonstrated on the restoration of a large dataset of realistic mass spectrometry data.
引用
收藏
页码:2094 / 2098
页数:5
相关论文
共 32 条
[1]   ACCELERATED LINE-SEARCH AND TRUST-REGION METHODS [J].
Absil, P. -A. ;
Gallivan, K. A. .
SIAM JOURNAL ON NUMERICAL ANALYSIS, 2009, 47 (02) :997-1018
[2]   On instabilities of deep learning in image reconstruction and the potential costs of AI [J].
Antun, Vegard ;
Renna, Francesco ;
Poon, Clarice ;
Adcock, Ben ;
Hansen, Anders C. .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2020, 117 (48) :30088-30095
[3]   Deep unfolding of a proximal interior point method for image restoration [J].
Bertocchi, C. ;
Chouzenoux, E. ;
Corbineau, M-C ;
Pesquet, J-C ;
Prato, M. .
INVERSE PROBLEMS, 2020, 36 (03)
[4]  
Cadoni S, 2016, IEEE IMAGE PROC, P3194, DOI 10.1109/ICIP.2016.7532949
[5]  
Chalvidal M, 2020, P 27 IEEE INT C IM P
[6]   Deterministic edge-preserving regularization in computed imaging [J].
Charbonnier, P ;
BlancFeraud, L ;
Aubert, G ;
Barlaud, M .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 1997, 6 (02) :298-311
[7]   SPOQ lp-Over-lq Regularization for Sparse Signal Recovery Applied to Mass Spectrometry [J].
Cherni, Afef ;
Chouzenoux, Emilie ;
Duval, Laurent ;
Pesquet, Jean-Christophe .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2020, 68 :6070-6084
[8]   PALMA, an improved algorithm for DOSY signal processing [J].
Cherni, Afef ;
Chouzenoux, Emilie ;
Delsuc, Marc-Andre .
ANALYST, 2017, 142 (05) :772-779
[9]   Convergence Rate Analysis of the Majorize-Minimize Subspace Algorithm [J].
Chouzenoux, Emilie ;
Pesquet, Jean-Christophe .
IEEE SIGNAL PROCESSING LETTERS, 2016, 23 (09) :1284-1288
[10]   A Majorize-Minimize Subspace Approach for l2-l0 Image Regularization [J].
Chouzenoux, Emilie ;
Jezierska, Anna ;
Pesquet, Jean-Christophe ;
Talbot, Hugues .
SIAM JOURNAL ON IMAGING SCIENCES, 2013, 6 (01) :563-591