Abstract: Every day we are told about information overflow; how so much data is coming at us through so many channels that it is difficult to weed out the gems that are important to us from all the noise. We believe that the solution to this problem is personalization. In this talk I will describe some recent work at Microsoft Research which approaches personalization from two different angles: on one hand, we use computational methods to find statistical correlations in large amounts of data. On the other hand, we augment the algorithmic approach with human intelligence. Through social networks we can leverage the trust relation between friends and followers to improve personalization beyond the computational approach.
Bio: Jurgen is an applied research at Microsoft Research FUSE Labs where I work on incorporating the latest research in machine learning and probabilistic modelling into new applications and services. Before joining FUSE Labs, Jurgen was a graduate student at the University of Cambridge working on Bayesian nonparametric models with Prof. Zoubin Ghahramani. Jurgen earned his MSc. from the University of Wisconsin at Madison where he worked on statistical natural language processing with Prof. Jerry Zhu. Jurgen's undergraduate degree is from the University of Leuven.
Project Emporia link
ReplyDeletehttp://www.projectemporia.com/