I'll be talking about a paper in preparation, which is a support paper for the recent paper by E. Abbe and A. Singer . I'll focus on explaining Sanov's Theorem, how it can be useful in hypothesis testing and estimation, and present its noisy version and application to alignment problems.
 E. Abbe, J. M. Pereira and A. Singer, Sample complexity of the Boolean multireference alignment problem. Submitted to ISIT, arXiv preprint arXiv:1701.07540, 2017.