Geoffrey Hinton, a respected Computer Science/AI Prof at the University of Toronto, has been the subject of many popular sci-tech articles, especially after Google bought his startup DNNresearch Inc. in 2012. (DNN= Deep Neural Networks). I would like to point out that nowadays what is called Deep Learning Neural Nets is really a hybrid of what I call in this blog Bayesian Networks and what was referred to as artificial Neural Nets by Minsky in the 1970’s (although he was by no means the originator of the subject of artificial NNs. The Wikipedia article on artificial neural networks traces back their history as early as 1943, and that leaves out the ancient Greece philosophers, who no doubt said something about them).
You can see the deep integration between Neural and Bayesian networks in Hinton’s work if you look at the slides of the following introductory talk he gave at UCLA in 2012 (he calls them Belief networks instead of Bayesian networks but it’s the same thing). Of course, quantum computers have the potential to do Bayesian Networks and AI calculations much faster than classical computers.
Graduate Summer School: Deep Learning, Feature Learning,
July 9 – 27, 2012, IPAM/UCLA
Geoffrey Hinton (University of Toronto)
PART 1: Introduction to Deep Learning & Deep Belief Nets (PDF Parts 1 & 2)
PART 2: Using backpropagation for fine-tuning a generative model to be better at discrimination
(My thanks to jesuslopez for alerting me to this link)