23 Feb 2015

In the lab meeting this week, Feb 25, Mahdi is going to talk about his recent work on “Explicit feature learning in unsupervised/semi-supervised learning”. He is going to present how to extract features that represent the diversity of the data and use these new features (by themselves or alongside initial set of features) to maximize an objective function. This approach has already been used in supervised learning and here he tries to test it with EM (an unsupervised algorithm). He also will take look at different approaches to find these new features i.e. Maximum Diversity Problem. The meeting will be held in TASC1 9408 at 1:30pm.