• Home
• Staff & Students
• Vacancies

Max Welling

 

Wednesday 15th April 2015

Time: 4.00pm

 

Basement Seminar Room

Alexandra House, 17 Queen Square, London, WC1N 3AR

 

Bayesian Inference in Complex Generative Models

 

In a time when deep learning and big data claim the center 
stage in machine learning, generative models and Bayesian inference have 
moved somewhat out of the limelight. I argue that generative modeling 
and Bayesian learning will remain key for many exciting applications. I 
present three advances in this direction developed in my group: 1) a new 
variational learning algorithm for the Helmholtz machine (the 
variational auto-encoder) applied to semi-supervised learning, 2) a new 
large-scale distributed posterior MCMC sampling procedure applied to 
Matrix Factorization and 3) an efficient posterior MCMC sampling 
algorithms for complex, likelihood free simulator models. I will 
conclude with a reflection on what is still missing to build truly large 
scale, distributed and privacy preserving learning systems.

Joint work with Ted Meeds, Durk Kingma and Sungjin Ahn.