AI Seminar: Deep probabilistic programming and the protein folding problem


Thomas HamelryckThomas Hamelryck is associate professor at the University of Copenhagen. He has a shared position at the Department of Computer Science (DIKU) and the Department of Biology. His main interests are Bayesian machine learning, including deep probabilistic programming, and the protein folding problem.


Probabilistic Programming is an emerging technology in machine learning, currently arising in the wake of Deep Learning and Big Data Analytics. The main idea is:

1. Use a computer language augmented with statistical operators
2. formulate a suitable probabilistic model for a given data set and
3. perform automated statistical inference by executing the program.

Such an approach has been made possible by recent breakthroughs in automatic inference (including advanced sampling methods and variational inference) and numerical computing (including GPU-savvy numerical software such as PyTorch and TensorFlow). Deep probabilistic programming marries Bayesian, probabilistic reasoning with deep learning. I will outline what probabilistic programming and deep probabilistic programming are about and discuss their relevance to an important, timely paradigm challenge - the protein folding problem.