Fall 2014 Seminar Series

DATE: 9/30/2014 (Tuesday)
LOCATION: E25-111
TIME: 12:00pm
Reception immediately following

SPEAKER:
Michael I. Jordan

TITLE
Optimistic Concurrency Control for Distributed Machine Learning

ABSTRACT
Research on distributed machine learning algorithms has focused primarily on one of two extremes---algorithms that obey strict concurrency constraints or algorithms that obey few or no such constraints. Taking a page from the database literature, we consider an intermediate alternative in which algorithms optimistically assume that conflicts are unlikely and if conflicts do arise a conflict-resolution protocol is invoked. We view this "optimistic concurrency control" (OCC) approach as particularly appropriate for learning problems which include discrete structural variables and which are combinatorial in nature. We explore the OCC paradigm in two rather different problem domains---Bayesian inference under combinatorial stochastic process priors and the maximization of non-monotone submodular functions.

Joint work with Xinghao Pan, Joseph Gonzalez, Stefanie Jegelka, Tamara Broderick, and Joseph Bradley.