Learning to Act under Uncertainty

Abstract

My talk will focus on problems in machine learning and AI that originate from robotics. In particular, I will present challenging problems related to decision-making under uncertainty, learning from limited demonstrations, online learning, as well as balancing exploration and exploitation. I will show how these problems can be solved by formulating them as optimization problems, derived from a higher principle such as entropy minimization. I will also show how I used the proposed solutions to solve problems in robotics.

Bio

Abdeslam Boularias received a Bachelor of Engineering degree in computer engineering in 2004 from the École Nationale Supérieure d’Informatique (Algeria). He received a Master's degree in computer science from Paris-Sud University (France), in 2005, and the Ph.D. degree from Laval University (Canada) in 2010. During his studies at Paris-Sud, he was a research assistant with the INRIA Saclay Institute, where he worked on fault tolerance in Grid Computing. In January 2006, he joined the group of Prof. B. Chaib-draa at Laval University, where he worked on decision-making in partially observable dynamical systems. From August 2010 to April 2013, he was a research scientist at the Max Planck Institute for Intelligent Systems in Tuebingen (Germany), where he worked with Prof. J. Peters in the Empirical Inference department, which was directed by Prof. B. Schoelkopf. He has been a Postdoctoral Fellow with the National Robotics Engineering Center of Carnegie Mellon University (USA) since May 2013. His main research interests include robotics, planning under uncertainty, and machine learning for decision-making.