Inductive Logic Programming: Towards machine assisted theory formation?

by Dr. Steve Moyle (Oxford University Computing Laboratory)

Wednesday 11 Sept 2002, 14:00 → 15:00 Europe/Zurich

500/1-001 - Main Auditorium (CERN)

How can a machine assistant help in the design of new drugs? Why do proteins fold one way and not another? Who causes accidents on the London orbital motorway? What strategies are hackers using to break into the system? How can organisations recognise 'High Potential' managers? What rules did Rachmaninoff use to play his piano? Inductive Logic Programming (ILP) -- a research area formed at the intersection of Machine Learning and Logic Programming -- is being applied to construct answers to these, and many other questions. ILP systems construct predicate descriptions from examples and background knowledge. The examples, background knowledge and final descriptions are all described as logic programs. A unifying theory of Inductive Logic Programming is being constructed from lattice-based concepts such as refinement, least general generalisation, inverse resolution and most specific corrections. Presently successful applications areas for ILP systems include problems in science, engineering, language and music. Some results from ILP systems are already published scientific discoveries. This talk will provide a short introduction to the curious world of ILP. It is aimed towards the application of ILP, and only presents basic theoretical ILP concepts.