BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//project/author//NONSGML v1.0//EN
CALSCALE:GREGORIAN
BEGIN:VEVENT
DTEND:20191227T120000Z
UID:0da1de660786bdde6affa65845583f56-32
DTSTAMP:19700101T120016Z
DESCRIPTION:The Bethe and Sinkhorn Permanents of Structured Matrices and its Implications for Profile Maximum Likelihood
URL;VALUE=URI:http://demo.onlypixels.com/iisc-csa/event/32/the-bethe-and-sinkhorn-permanents-of-structured-matrices-and-its-implications-for-profile-maximum-likelihood/
SUMMARY:Here we consider the problem of computing the likelihood of the profile of a discrete distribution, i.e. the probability of observing the multiset of element frequencies, and computing a profile maximum likelihood (PML) distribution, i.e. a distribution with the maximum profile likelihood. For each problem we provide polynomial time algorithms that given $n$ i.i.d samples from a discrete distribution, achieve an approximation factor of $expleft(-O(sqrt{n} log n) right)$, improving upon the previous best-known bound achievable in polynomial time of $exp(-O(n^{2/3} log n))$ (Charikar, Shiragur and Sidford, 2019). Through the work of Acharya, Das, Orlitsky and Suresh (2016), this implies a polynomial time universal estimator for symmetric properties of discrete distributions in a broader range of error parameter.
We achieve these results by providing new bounds on the quality of approximation of the Bethe and Sinkhorn permanents (Vontobel, 2012 and 2014).
We show that each of these are $exp(O(k log(N/k)))$ approximations to the permanent of $N times N$ non-negative matrices with at most $k$ distinct columns, improving upon the previous known bounds of $exp(O(N))$.
To obtain our results on PML,
we exploit the fact that the PML objective is proportional to the permanent of a certain Vandermonde matrix with $sqrt{n}$ distinct columns.
As a by-product of our work we establish a surprising connection between the convex relaxation in prior work (CSS19) and the well-studied Bethe and Sinkhorn approximations.
This is in joint work with Moses Charikar and Aaron Sidford.
DTSTART:20191227T120000Z
END:VEVENT
END:VCALENDAR