Friday, February 10, 2017

Phase Transitions in Oblivious Compressed Sensing

 Ori just sent me the following:
 


Hi Igor,

How are you doing?

Please find attached a paper studying the thresholds of “Oblivious CS”. We believe you and the blog’s savvy readers will find it of interest.

The abstract reads:

“In this paper we study the phase transition thresholds for oblivious compressed sensing, namely a noisy compressed sensing problem solved, unawarely or reluctantly, as a noiseless system. Modeling the unaccounted for noise as Gaussian and given a certain measurement rate, the maximal input sparsity allowing for stable decoding is derived, in the large-system limit, for a general Lp-norm decoder. The obliviousness penalty, with respect to the noise-aware threshold, is explicitly quantified.“

On a personal note: Nuit Blanche is certainly my #1 source for up-to-date information on CS (and recently also on ML).

So at this opportunity I would like to thank you for that!

Best regards,

Ori

---

Ori Shental

Communications Theory Department
Mobile Radio Research
Nokia Bell Laboratories
791 Holmdel Rd, Room R-213
Holmdel, NJ 07733
https://www.bell-labs.com/usr/ori.shental 
In this paper we study the phase transition thresholds for oblivious compressed sensing, namely a noisy compressed sensing problem solved, unawarely or reluctantly, as a noiseless system. Modeling the unaccounted for noise as Gaussian and given a certain measurement rate, the maximal input sparsity allowing for stable decoding is derived, in the large-system limit, for a generic lp-norm decoder. The obliviousness penalty, with respect to the noise-aware threshold, is explicitly quantified.

 
 
 
 
 
Join the CompressiveSensing subreddit or the Google+ Community or the Facebook page and post there !
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.

No comments:

Printfriendly