Probabilistic Methods for Big Data and Machine Learning 

Probabilistic methods are  powerful & practical tools for effectively solving many problems in search and machine learning with massive datasets. Basically, in many cases, the computation problems can  be (probabilistically) transformed into statistical estimation  problems. It is also often the case that a better solution to a particular estimation problem corresponds to a computationally more efficient algorithm. In this talk, I will go over a few successful examples (some of which have been widely used in industry) and present a few problems we are actively working on.  The content of the talk includes joint work with many collaborators such as Cun-Hui Zhang and Gennady Samorodnitsky.