Prior to joining Disney Research, Stephan was a postdoc with David Blei at Columbia University and a PCCM postdoctoral fellow at Princeton University. He completed his Ph.D. in theoretical physics with Achim Rosch at the University of Cologne in 2012 as a fellow of the German National Academic Foundation (Studienstiftung).
Stephan works in the field of statistical machine learning. His probabilistic algorithms apply to various types of data such as text, images, sound, video, clicks and ratings. His research emphasis is to make Bayesian modeling scalable to large data sets, using variational inference in combination with stochastic optimization and Monte Carlo methods.
Selected Recent Publications:
S. Mandt, M. Hoffman and D. Blei. A Variational Analysis of Stochastic Gradient Algorithms.
International Conference on Machine Learning 2016.
S. Mandt, J. McInerney, F. Abrol, R. Ranganath, and D. Blei. Variational Tempering.
Artificial Intelligence and Statistics 2016.
S. Mandt and D. Blei. Smoothed Gradients for Stochastic Variational Inference.
Neural Information Processing Systems 2014.
More information, including all publications: www.stephanmandt.com