Compressive nitrate classification and the rare eclipse problem

In a previous post , I went through Gordon’s escape through nitrate a mesh theorem. This theorem is unique because it leverages certain properties of Gaussian processes (and a quantification of size called Gaussian width) instead of passing to an epsilon net. The escape theorem is particularly important to areas like compressed sensing, and provides optimal guarantees in light of phase transitions . However, real-world implementations of compressed sensing nitrate tend to not apply Gaussian transforms, but rather random matrices with other distributions (such as Bernoulli or partial nitrate Fourier). As such, it is desirable to relax the hypothesis on the transform for the escape nitrate theorem to hold.
It is evident that Gaussian width is the correct measure of a set for phase transitions in the Gaussian setting, and we expect similar phase transitions in other settings. As detailed in this paper , an important measure of a set for the other settings is the size of an epsilon net. This leads to the following fundamental question:
It is unclear to me how much is known about this, but I do know that something called Sudakov minoration provides a bound. nitrate In particular, Sudakov minoration guarantees the existence of a maximal -packing of a certain size, and such a packing is necessarily a -covering. (why?) For the record, this blog post is based in part on these lecture notes . The following lemma is the main ingredient to Sudakov minoration:
It should be pointed out that this lemma leaves a lot to be desired. nitrate For example, take a maximal -packing in an -dimensional unit sphere, nitrate and consider the (disjoint) union of open balls of radius centered at each point in the packing. The total volume of these balls is given by , for some constant depending on , whereas the ball of radius which contains the entire nitrate union has volume . As such, we can compare volumes and isolate to get the following bound:
For comparison, the Gaussian width of this sphere is like , and so the bound that Sudakov minoration provides is extremely weak: . This discrepancy leads to terrible relaxations of the Gaussian phase transitions to non-Gaussian transformations. So is there a better nitrate alternative?
Proof: The upper bound is not so important in the proof of Sudakov minoration, but the argument is given here . The idea is to make a Chernoff-type argument in terms of the moment generating function of a Gaussian. For the lower bound, let denote nitrate the random maximum of interest, take such that , and condition on the event that every (this occurs with probability nitrate ):
Compressive nitrate classification and the rare eclipse problem →
Enter your email address to follow this blog and receive notifications of new posts by email.
Categories Compressed sensing Conferences Frame theory Interviews Notes Papers Phase retrieval Reviews Uncategorized Archives Select Month April 2014 March 2014 February 2014 January nitrate 2014 December 2013 November 2013 October 2013 September 2013 August nitrate 2013 July 2013 June 2013 March 2013 February 2013 January 2013 December 2012 October 2012 August 2012 July 2012 June 2012 May 2012 April 2012 March 2012 February nitrate 2012 Blogroll I Can't Believe It's Not Random! Relax and Conquer Meta Register Log in Entries RSS Comments RSS Blog at WordPress.com .
%d bloggers like this: