Brian Ziebart

University of Illinois at Chicago

 

Abstract
The standard approach to supervised machine learning is to choose the form of a predictor and optimize its parameters based on training data. Approximations of the predictor's performance measure are often required to make the optimization problem tractable. Instead of approximating the performance measure and using the exact training data, this talk introduces a supervised machine learning framework that adversarially approximates the training data and uses the exact performance measure. This formulation provides flexibility for addressing sample selection bias, the fundamental problem hindering advances in active learning, and for inductively optimizing multivariate performance measures like the F-measure and the discounted cumulative gain from information retrieval and ranking tasks.

Bio:
Brian Ziebart is an Assistant Professor in the Department of Computer Science at UIC and an alum of UIUC (BS), where he conducted research under the supervision of Prof. Roy Campbell and Prof. Dan Roth. He received his PhD from Carnegie Mellon University where he was also a postdoctoral fellow. His research interests include machine learning, decision theory, game theory, robotics, and assistive technologies.