Learning MDL logic programs from noisy data

TL;DR

This paper presents a novel inductive logic programming method that learns minimal description length logic programs from noisy data, including recursive programs, and demonstrates improved accuracy and noise handling across various domains.

Contribution

It introduces a new approach for learning MDL logic programs from noisy data, capable of handling recursion and outperforming existing methods.

Findings

Outperforms existing approaches in predictive accuracy

Scales to moderate noise levels

Effective across domains like drug design, game playing, and program synthesis

Abstract

Many inductive logic programming approaches struggle to learn programs from noisy data. To overcome this limitation, we introduce an approach that learns minimal description length programs from noisy data, including recursive programs. Our experiments on several domains, including drug design, game playing, and program synthesis, show that our approach can outperform existing approaches in terms of predictive accuracies and scale to moderate amounts of noise.