Geoffrey Hinton

Geoffrey Hinton is a British-Canadian computer scientist whose neural-network research helped establish deep learning as a major approach to artificial intelligence.

David Rumelhart: An American psychologist and cognitive scientist whose work helped revive interest in neural networks. His collaboration with Hinton and Ronald Williams produced a landmark account of backpropagation.
Backpropagation: An algorithm that computes gradients through a network by propagating errors backward from its outputs. Hinton helped make this method central to training multilayer neural networks.
Deep learning: A family of machine-learning methods that uses neural networks with multiple layers. Hinton’s research helped turn deep neural networks into a dominant AI approach.
Perceptron: An early neural-network model that classifies inputs using a weighted sum and threshold. Its limitations helped frame the challenge of learning richer representations with multilayer networks.
Support-vector machine: A supervised learning method that separates classes by maximizing the margin between them. It was a strong alternative to neural networks during the period before deep learning’s resurgence.
Ronald J. Williams: An American computer scientist who developed learning algorithms for neural networks. He coauthored the influential 1986 paper on backpropagation with Hinton and Rumelhart.
Boltzmann machine: A stochastic recurrent neural network that learns probability distributions over its inputs. Hinton developed this model with Terry Sejnowski as a way to learn internal representations.
ImageNet: A large image database and visual-recognition benchmark created for computer vision research. AlexNet’s 2012 ImageNet result made deep neural networks difficult for the field to ignore.
Connectionism: An approach to cognition that models mental processes as distributed activity across networks of simple units. Hinton’s research drew on this framework to connect neural computation with learning and cognition.
Symbolic artificial intelligence: An AI approach that represents knowledge with explicit symbols and rules. Hinton’s connectionist models learn distributed representations rather than relying on hand-coded symbolic rules.

Linked from 7 pages