Loading timeline…
20062000s
Simon Osindero
Co-author of the 2006 deep belief network paper that made deep architectures trainable and restarted the field.
Organizations
University of TorontoGoogle DeepMindFlickr
Major Achievements
- •Co-authored 'A Fast Learning Algorithm for Deep Belief Nets' (2006) with Geoffrey Hinton and Yee Whye Teh.
- •The greedy layer-wise pre-training it introduced showed deep networks could be trained after all, ending the second AI winter for neural networks.
- •Worked on conditional generative adversarial networks, extending GANs to controllable outputs.
- •Continued as a research scientist at DeepMind.
Key Papers
- A Fast Learning Algorithm for Deep Belief Nets
Geoffrey E. Hinton, Simon Osindero, Yee-Whye Teh · Neural Computation · 2006