Tutorial Abstract

Logical and Relational Learning: A New Synthesis

Luc De Raedt

Monday, September 15, afternoon
Location: R008

I use the term ``logical and relational learning'' (LRL) to refer to the subfield of artificial intelligence,machine learning and data mining that is concerned with learning in expressive logical or relational representations. It is the union of inductive logic programming, (statistical) relational learning and multi-relational data mining. The topic of logical and relational learning has been around in the artificial intelligence and machine learning communities for at least 40 years now. Starting with early work by researchers such as Plotkin and Michalski, it has become a subdiscipline of machine learning and data mining since the advent of inductive logic programming in the early 90s. Today it is receiving a lot of attention thanks to the popularity of statistical relational learning and the mining of structured data (including graphs, trees and sequences).
This tutorial is intended to provide a new synthesis of the field by summarizing some of the key lessons learned during the past 40 years. These lessons will not only be concerned with the what, why and how of logical and relational learning, but will mainly show which principles of logical and relational learning are relevant to other approaches in machine learning and data mining, including those that do not use logical or relational representations (such as mining and learning from graphs, trees and sequences).