Blind Calibration for Sparse Regression: A State Evolution Analysis

Marylou Gabrie, Jean Barbier, Florent Krzakala, Lenka Zdeborova

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Sparse regression, such as the one used in compressed sensing, allows to acquire compressible signals with a small number of measurements. As such, a correct calibration of a potential hardware problem is a central issue. Blind calibration, that is performing at the same time calibration and compressed sensing when the training signals are sparse but unknown, is thus particularly appealing. A potential approach was suggested by Schülke et al, using an approximate message passing (AMP) for blind calibration (cal-AMP). Here, we show that the asymptotic performances of this algorithm can be analysed by an exact state evolution equation. It allows to confirm that cal-AMP requires a smaller number of measurements and/or signals in order to perform with respect to standard convex approaches, and opens the way to more complex message passing techniques.

Original languageEnglish
Title of host publication2019 IEEE 8th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2019 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages649-653
Number of pages5
ISBN (Electronic)9781728155494
DOIs
Publication statusPublished - 1 Dec 2019
Externally publishedYes
Event8th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2019 - Le Gosier, Guadeloupe
Duration: 15 Dec 201918 Dec 2019

Publication series

Name2019 IEEE 8th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2019 - Proceedings

Conference

Conference8th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2019
Country/TerritoryGuadeloupe
CityLe Gosier
Period15/12/1918/12/19

Keywords

  • Belief propagation
  • Compressed sensing
  • approximate message passing
  • blind calibration

Fingerprint

Dive into the research topics of 'Blind Calibration for Sparse Regression: A State Evolution Analysis'. Together they form a unique fingerprint.

Cite this