Connections Between Inductive Inference and Machine Learning

Abstract

Inductive inference is a branch of computational learning theory which deals with learning in the limit. Though this topic deals with mostly theoretical work, it has provided some results which can be of use to practical machine learning. Some of these works include the work multitask or context-sensitive learning, learnability of elementary formal systems, behavioral cloning, learning to coordinate, geometrical clustering, and so on. The results in these areas also often give insights into limitations of science.