Inductive logic programming (ilp) and reasoning by analogy in context of embodied robot learning

Abstract

The ability of reasoning by analogy seems to be essential for many cognitive processes from low and high level perception to categorization. Intuitively, the idea is to use old knowledge in order to explain new observations similar in some ways to what is already known. In some sense it is opposite of induction where in order to explain the observations one comes up with new hypotheses/theories. Therefore, a system capable of both ways of reasoning would be superior to either. In this paper we first present an overview of Inductive Logic Programming (ILP) systems that use reasoning by analogy. Then we present the results of applying Analogical Prediction to problem that arise in the context of physically embodied robot which tries to learn regularities in its environment.