RockIt: Exploiting Parallelism and Symmetry for MAP Inference in Statistical Relational Models

TL;DR

RockIt is a MAP inference engine for statistical relational models that leverages symmetry and parallelism, significantly improving efficiency and solution quality over existing systems.

Contribution

Introduces cutting plane aggregation (CPA) to exploit symmetries and parallelizes MAP inference, advancing the state-of-the-art in efficiency and scalability.

Findings

Outperforms Alchemy, Markov TheBeast, and Tuffy in benchmarks.

Uses CPA to create more compact ILPs with explicit symmetry.

Parallelizes MAP inference to utilize multi-core architectures.

Abstract

RockIt is a maximum a-posteriori (MAP) query engine for statistical relational models. MAP inference in graphical models is an optimization problem which can be compiled to integer linear programs (ILPs). We describe several advances in translating MAP queries to ILP instances and present the novel meta-algorithm cutting plane aggregation (CPA). CPA exploits local context-specific symmetries and bundles up sets of linear constraints. The resulting counting constraints lead to more compact ILPs and make the symmetry of the ground model more explicit to state-of-the-art ILP solvers. Moreover, RockIt parallelizes most parts of the MAP inference pipeline taking advantage of ubiquitous shared-memory multi-core architectures. We report on extensive experiments with Markov logic network (MLN) benchmarks showing that RockIt outperforms the state-of-the-art systems Alchemy, Markov TheBeast, and Tuffy both in terms of efficiency and quality of results.