Maximum a Posteriori Estimation by Search in Probabilistic Programs

TL;DR

The paper presents Bayesian ascent Monte Carlo (BaMC), an approximate search algorithm for fast MAP estimation in probabilistic programs, demonstrating superior speed and robustness over existing methods.

Contribution

Introduction of BaMC, a versatile anytime MAP search algorithm applicable to various probabilistic models with different variable types and dependencies.

Findings

BaMC outperforms other MAP algorithms in speed.

BaMC is more robust across diverse probabilistic models.

BaMC is applicable to models with different variable types.

Abstract

We introduce an approximate search algorithm for fast maximum a posteriori probability estimation in probabilistic programs, which we call Bayesian ascent Monte Carlo (BaMC). Probabilistic programs represent probabilistic models with varying number of mutually dependent finite, countable, and continuous random variables. BaMC is an anytime MAP search algorithm applicable to any combination of random variables and dependencies. We compare BaMC to other MAP estimation algorithms and show that BaMC is faster and more robust on a range of probabilistic models.