Abstract:
Maximum a posteriori (MAP) inference is a fundamental problem in machine learning that involves identifying the most likely configuration of a discrete-valued Markov random field. Due to the difficulty of this combinatorial problem, linear programming (LP) relaxations are commonly used to derive specialized message passing algorithms that are often interpreted as coordinate descent on the dual LP. To achieve more desirable computational properties, a number of methods regularize the LP with an entropy term, leading to a class of smooth message passing algorithms with convergence guarantees. In this paper, we present randomized methods for accelerating these algorithms by leveraging techniques that underlie classical accelerated gradient methods. Crucially, the proposed algorithms incorporate the familiar steps of standard smooth message passing algorithms, which can be viewed as coordinate minimization steps. We show that the accelerated variants achieve faster rates for finding $\epsilon$-optimal points of the unregularized problem. When the LP is tight, we prove that the proposed algorithms recover the true MAP solution in fewer iterations than the best-known results.
