Skip to main content

Distributional prior in Inductive Logic Programming

Supervisor

Suitable for

MSc in Advanced Computer Science

Abstract

Inductive Logic Programming is a form (ILP) of Machine Learning based on Computational Logic. Given examples and some background knowledge, the goal of an ILP learner is to search for a hypothesis that generalises the examples. However, the search space in ILP is a function of the size of the background knowledge. All predicates are treated as equally likely, and current ILP systems cannot make use of distributional assumptions to improve the search. This project focuses on making use of an assumed given prior probability over the background predicates to learn more efficiently. The idea is to order subsets of the background predicates in increasing order of probability of appearance. This project is a mix of theory, implementation, and experimentation.

Prerequisites: logic programming, probability theory, or interest to learn about it