Bottom-Up Stratified Probabilistic Logic Programming with Fusemate

TL;DR

This paper presents Fusemate, a probabilistic logic programming system that uses bottom-up grounding combined with relevance testing to improve inference efficiency, especially in complex, high-branching problems.

Contribution

Fusemate introduces a novel bottom-up grounding approach with query-guided pruning, enhancing probabilistic inference control and performance over existing systems.

Findings

Competitive or superior performance in high-branching problems

Effective pruning reduces grounding size and complexity

Demonstrated applicability to time-related models like Markov chains

Abstract

This paper introduces the Fusemate probabilistic logic programming system. Fusemate's inference engine comprises a grounding component and a variable elimination method for probabilistic inference. Fusemate differs from most other systems by grounding the program in a bottom-up way instead of the common top-down way. While bottom-up grounding is attractive for a number of reasons, e.g., for dynamically creating distributions of varying support sizes, it makes it harder to control the amount of ground clauses generated. We address this problem by interleaving grounding with a query-guided relevance test which prunes rules whose bodies are inconsistent with the query. % This is done We present our method in detail and demonstrate it with examples that involve "time", such as (hidden) Markov models. Our experiments demonstrate competitive or better performance compared to a state-of-the art probabilistic logic programming system, in particular for high branching problems.