[Ml-stat-talks] Fwd: PACM Colloquium 4/9/12

Gungor Polatkan polatkan at princeton.edu
Tue Apr 3 11:42:42 EDT 2012



Begin forwarded message:

> From: Valerie Marino <vmarino at math.princeton.edu>
> Date: April 3, 2012 8:49:02 AM EDT
> To: <pacm-colloquium at math.princeton.edu>
> Cc: <vmarino at math.princeton.edu>
> Subject: PACM Colloquium 4/9/12
>
> DATE: Monday, April 9, 2012
> PLACE: 214 Fine Hall
>
> TIME: 4:30 pm
>
> SPEAKER: : Prof. David Donoho, Stanford
>
> TITLE: Optimal Phase Transitions in Compressed Sensing
>
> ABSTRACT:
> "Compressed Sensing" is an active research area which touches on  
> harmonic analysis, geometric functional analysis, applied  
> mathematics, computer science electrical engineering and information  
> theory. Concrete achievements, such as speeding up pediatric MRI  
> acquistion times from several minutes to under a minute, are now  
> entering daily use. In my talk I will discuss the notion of phase  
> transitions in combinatorial geometry, describe how they precisely  
> demarcate the situations where a popular algorithm in compressed  
> sensing -- ell_1 minimization -- can succeed. Then I will discuss  
> the issue: what is the best possible phase transition of any  
> algorithm? We get different answers depending on the assumptions we  
> make. If we assume the distribution of the coefficients is known/ 
> unknown, we get different answers; in each case I describe novel  
> algorithms precisely achieving the optimal phase transition,  
> converging to the answer at an expoential rate depending only on  
> distance from phase transition. Both algorithms are applications of  
> the Approximate Message Passing algorithm introduced by Maleki,  
> Montanari and myself. This talk surveys joint work with several  
> authors, including Andrea Montanari and Jared Tanner, as well as  
> Adel Javanmard, Iain Johnstone and Arian Maleki.
>
> Coffee & Tea will be served at 3:55 p.m. in 217A Fine Hall
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/ml-stat-talks/attachments/20120403/a2c921cb/attachment.html>


More information about the Ml-stat-talks mailing list