Note: Some of the projects proposed on this page can be considered for either Master semester project or Master thesis project, the difficulty of the project being then adapted accordingly.
-
Markov Chain monte Carlo based on Message Passing (SMILS)
In this project we want to investigate new types of Markov chain Monte Carlo algorithms for large probabilistic graphical models, where the update rules of the Markov chain are based on a message passing calculation in a limited neighborhood of the nodes…
-
Message passing for Quantum Models (SMILS)
In this project we want to develop message passing techniques for simple quantum spin models on graphs. …
-
Universal prediction and alpha-mutual information
In this project we wish to study — through some experimental work (with MATLAB or Python) — the performance of a newly-devised class of predictors that estimate data sequences when the information about the distribution generating it is limited. The predictors are based on a generalized definition of mutual information.