On U-Statistics and Compressed Sensing I: Non-Asymptotic Average-Case Analysis
Hoeffding's U-statistics model combinatorial-type matrix parameters (appearing in CS theory) in a natural way. This paper proposes using these statistics for analyzing random compressed sensing matrices, in the non-asymptotic regime (relevant to practice). The aim is to address certain pessimis...
Gespeichert in:
Veröffentlicht in: | arXiv.org 2012-10 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Hoeffding's U-statistics model combinatorial-type matrix parameters (appearing in CS theory) in a natural way. This paper proposes using these statistics for analyzing random compressed sensing matrices, in the non-asymptotic regime (relevant to practice). The aim is to address certain pessimisms of "worst-case" restricted isometry analyses, as observed by both Blanchard & Dossal, et. al. We show how U-statistics can obtain "average-case" analyses, by relating to statistical restricted isometry property (StRIP) type recovery guarantees. However unlike standard StRIP, random signal models are not required; the analysis here holds in the almost sure (probabilistic) sense. For Gaussian/bounded entry matrices, we show that both l1-minimization and LASSO essentially require on the order of k \cdot [\log((n-k)/u) + \sqrt{2(k/n) \log(n/k)}] measurements to respectively recover at least 1-5u fraction, and 1-4u fraction, of the signals. Noisy conditions are considered. Empirical evidence suggests our analysis to compare well to Donoho & Tanner's recent large deviation bounds for l0/l1-equivalence, in the regime of block lengths 1000-3000 with high undersampling (50-150 measurements); similar system sizes are found in recent CS implementation. In this work, it is assumed throughout that matrix columns are independently sampled. |
---|---|
ISSN: | 2331-8422 |
DOI: | 10.48550/arxiv.1210.8116 |