Consistent Signal and Matrix Estimates in Quantized Compressed Sensing by Amirafshar Moshtaghpour, Laurent Jacques, Valerio Cambareri, Kevin Degraux, Christophe De Vleeschouwer
This paper focuses on the estimation of low-complexity signals when they are observed through
Muniformly quantized compressive observations. Among such signals, we consider 1-D sparse vectors, low-rank matrices, or compressible signals that are well approximated by one of these two models. In this context, we prove the estimation efficiency of a variant of Basis Pursuit Denoise, called Consistent Basis Pursuit (CoBP), enforcing consistency between the observations and the re-observed estimate, while promoting its low-complexity nature. We show that the reconstruction error of CoBP decays like M−1/4when all parameters but Mare fixed. Our proof is connected to recent bounds on the proximity of vectors or matrices when (i) those belong to a set of small intrinsic "dimension", as measured by the Gaussian mean width, and (ii) they share the same quantized (dithered) random projections. By solving CoBP with a proximal algorithm, we provide some extensive numerical observations that confirm the theoretical bound as Mis increased, displaying even faster error decay than predicted. The same phenomenon is observed in the special, yet important case of 1-bit CS.
Implementation is here: http://sites.uclouvain.be/ispgroup/index.php/Softwares/HomePage
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.