Dimension free PAC-Bayes bounds and unsupervised classification

In this talk, I will present some dimension free PAC Bayes bounds and explain how they can be used to perform Principle Component Analysis and unsupervised clustering in high dimension.