Inductive Logic Programming: Challenges

Stephen Muggleton gave the invited talk "Meta-Interpretive Inductive Logic Programming (ILP) is a research area Learning: achievements and challenges". Meta-Interpretive formed at the intersection of Machine Learning and logicbased Learning (MIL) is an ILP technique aimed at supporting knowledge representation. ILP has originally used learning of recursive definitions, by automatically introducing logic programming as a uniform representation language sub-definitions that allow decomposition into a hierarchy for examples, background knowledge and hypotheses for of reusable parts (Muggleton et al. 2014; 2015). ILP has also explored several connections (or abducing) first-order clauses whose heads unify with with statistical learning and other probabilistic approaches, a given goal. MIL additionally fetches higher-order metarules expanding research horizons significantly. A recent survey whose heads unify with the goal and saves the resulting of ILP can be seen in (Muggleton et al. 2012).