Approximating Profile Maximum Likelihood Efficiently - New Bounds on the Bethe Permanent by Moses Charikar
Offered By: International Centre for Theoretical Sciences via YouTube
Course Description
Overview
Explore a seminar on approximating profile maximum likelihood efficiently and new bounds on the Bethe permanent. Delve into symmetric properties of distributions and their estimators, focusing on the universal plug-in estimator that maximizes profile maximum likelihood (PML). Learn about the first polynomial-time computable PML estimator with a 2^{n^{1-delta}}-approximation guarantee, computable in nearly linear time. Discover the connection between the introduced convex relaxation and the Bethe free energy approximation, leading to new bounds on the Bethe permanent of non-negative matrices. Gain insights from speaker Moses Charikar of Stanford University as he covers topics including setup, sequence maximum likelihood, fixed distribution P, universal estimators, open questions, and new results.
Syllabus
Introduction
Setup
Profile Maximum Likelihood
Sequence Maximum Likelihood
Fixed Distribution P
How the two connected
Universal estimator
Two questions
Open question
Recap
New Result
Theorem
Conclusion
Taught by
International Centre for Theoretical Sciences
Related Courses
Физика как глобальный проектNational Research Nuclear University MEPhI via Coursera Advanced statistical physics
École Polytechnique Fédérale de Lausanne via edX Insights on Gradient-Based Algorithms in High-Dimensional Learning
Simons Institute via YouTube Statistical Physics and Computation in High Dimension
Simons Institute via YouTube Computing Partition Functions, Part I
Simons Institute via YouTube