Lecture: "Abstraction: A key to perception, reasoning, and learning"

Abstraction is a pervasive activity in human perception, conceptualization and reasoning, and, as such, it enters the vocabulary of almost all disciplines, as well as the everyday life. As a result, it does not exist, as for now, any definition that is able to cover all the meanings that abstraction has acquired in the multiplicity of its utilizations. Starting from this range of meanings, we will show that all of them basically share some characteristics, which are more or less directly related to some notion of “simplicity”. Leaving then aside any general view of abstraction, we will concentrate on specific roles that this notion plays in both natural systems (typically human cognition) and artificial one (typically intelligent artefacts), and, in particular, on the relationships between abstraction and change of representation. We will then introduce some computational models, and compare abstraction with ththe notions of generalization and reformulation, using Information Theory as a basic approach. Finally, we will describe the role played by abstraction as both an help (how abstraction can speed up learning) and a target of learning (how abstraction operators can be learned), specifically in the setting of the recently proposed “deep learning” approach. Examples of application will conclude the talk.