The Architectures of Geoffrey Hinton

Abstract

Geoffrey Everest Hinton is a pioneer of deep learning, an approach to machine learning which allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction, whose numerous theoretical and empirical contributions have earned him the title the Godfather of deep learning. This chapter offers a brief outline of his education, early influences and prolific scientific career that started in the midst of AI winter when neural networks were regarded with deep suspicion. With a single goal fueling his ambitions—understanding how the mind works by building machine learning models inspired by it—he surrounded himself with like-minded collaborators and worked on inventing and improving many of the deep learning building blocks such as distributed representations, Boltzmann machines, backpropagation, variational learning, contrastive divergence, deep belief networks, dropout, and rectified linear units. The current deep learning renaissance is the result of that. His work is far from finished; a revolutionary at heart, he is still questioning the basics and currently developing a new approach to deep learning in the form of capsule networks.