Efficient rule induction by ignoring pointless rules

2025-02-03

AI Summary

Inductive Logic Programming (ILP) suffers from low learning efficiency and limited generalization due to an exponentially large hypothesis space rife with semantically redundant rules. Method: This paper formally defines “meaningless rules”—those either semantically equivalent to existing rules or incapable of distinguishing negative examples—and introduces a dual pruning criterion integrating semantic equivalence and discriminative power, enabling theoretically sound, lossless hypothesis space reduction. The approach leverages predicate logic inference and rule discriminability analysis to support efficient, semantics-preserving pruning. Contribution/Results: Evaluated across vision reasoning and game-playing tasks, the framework achieves a 99% speedup in training while maintaining zero accuracy degradation in prediction. To our knowledge, this is the first ILP pruning framework that simultaneously guarantees formal correctness and practical efficiency.

Technology Category

• Knowledge Representation and Reasoning: Logic Programming
• Machine Learning: Statistical Relational/Logic Learning
• Reasoning under Uncertainty: Probabilistic Programming

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

The goal of inductive logic programming (ILP) is to find a set of logical rules that generalises training examples and background knowledge. We introduce an ILP approach that identifies pointless rules. A rule is pointless if it contains a redundant literal or cannot discriminate against negative examples. We show that ignoring pointless rules allows an ILP system to soundly prune the hypothesis space. Our experiments on multiple domains, including visual reasoning and game playing, show that our approach can reduce learning times by 99% whilst maintaining predictive accuracies.

Problem

• Inductive Logic Programming
• Rule Redundancy
• Learning Efficiency

Innovation

• Inductive Logic Programming
• Rule Pruning
• Efficiency Enhancement