Last updated December 2015.
If you're looking for the ICML'15 Deep Learning Workshop video recordings, click here.
2015-12-12: Invited Talk, Black Box Inference and Learning Workshop, NIPS'15, Montreal, Canada
Variational Auto-Encoders and Extensions
Presentation: [PDF]
2015-10-14: Invited Talk, University of Cambridge, U.K.
2015-10-21: Invited Talk, Columbia University, U.S.A.
Efficient Inference and Learning with Intractable Posteriors? Yes, Please.
Presentation: [PDF]
2014-12-9: Spotlight Talk (Deep Spotlights), NIPS'14, Montreal, Canada
Semi-Supervised Learning with Deep Generative Models
Presentation: [PDF]
Visualisation of latent space of deep generative model of SVHN digits:
2014-12-13: Invited Talk, NIPS'14 Workshop on Advances in Variational Inference, Montreal, Canada
Stochastic Backpropagation, Variational Inference, and Semi-Supervised Learning
Presentation: [PDF]
2014-09: Deep Probabilistic Models Workshop, Sheffield, U.K.
Deep Generative Models
Presentation: [PDF]
2014-06-27: Invited Talk, Tsinghua University, Beijing, China
Auto-Encoding Variational Bayes
Presentation: [PDF]
2014-06: International Conference on Machine Learning, Beijing, China (ICML 2014)
Efficient Gradient-Based Inference through Transformations between Bayes Nets and Neural Nets
Presentation: [PDF] [ODP]
2014-04: International Conference on Learning Representations, Banff, Canada (ICLR 2014)
Auto-Encoding Variational Bayes
Presentation: [PDF] [ODP]
2014-03-26: Invited Talk, Google Deepmind, London, U.K.
Stochastic Gradient VB and the Variational Auto-Encoder
Paper: [PDF]
2014-01: IAS Talk, Univ. of Amsterdam, Netherlands
Stochastic Gradient VB. Intractable posterior distributions? Gradients to the rescue!
Presentation: [PDF] [ODP]
2013-07: CIFAR NCAP Summer School, Univ. of Toronto, Canada, hosted by Geoff Hinton
Speeding up Gradient-Based Inference and Learning in deep/recurrent Bayes Nets with Continuous Latent Variables
Presentation: [PDF] [ODP]
As the ODP files have been stripped from the embedded video's, here are some of the video's hosted on Youtube.
With L-BFGS:
With Monte Carlo Expectation Maximization (MCEM):
With Auto-Encoding Variational Bayes (AEVB):
3D manifold of MNIST (Learned with AEVB):