Videolibrary
RUS  ENG    JOURNALS   PEOPLE   ORGANISATIONS   CONFERENCES   SEMINARS   VIDEO LIBRARY   PACKAGE AMSBIB  
Video Library
Archive
Most viewed videos

Search
RSS
New in collection






International Workshop on Statistical Learning
June 28, 2013 12:00–12:30
 


Disentangling mixtures of Gaussians

A. Moitra

Institute for Advanced Study, School of Mathematics
Supplementary materials:
Adobe PDF 721.8 Kb

Number of views:
This page:184
Materials:42
Youtube:

A. Moitra



Abstract: Given data drawn from a mixture of multivariate Gaussians, a basic problem is to accurately estimate the mixture parameters. We provide a polynomial-time algorithm for this problem for any fixed number $k$ of Gaussians in $n$ dimensions (even if they overlap), with provably minimal assumptions on the Gaussians and polynomial data requirements.
In statistical terms, our estimator converges at an inverse polynomial rate, and no such estimator (even exponential time) was known for this problem (even in one dimension, restricted to two Gaussians). Our algorithm reduces the $n$-dimensional problem to the one dimensional problem, where the method of moments is applied. Additionally, in order to prove correctness for our univariate learning algorithm, we develop a novel explanation for why the method of moments (due to Pearson in 1894) works based on connections to algebraic geometry. [Joint work with Adam Tauman Kalai and Gregory Valiant. See also independent work of Mikhail Belkin and Kaushik Sinha]

Supplementary materials: moitra.pdf (721.8 Kb)

Language: English
 
  Contact us:
 Terms of Use  Registration to the website  Logotypes © Steklov Mathematical Institute RAS, 2024