Abstract

Though successful in well-behaved and engineered environments, current reinforcement learning methods suffer from robustness, safety, and efficiency-related issues when attempted in the real-world. In this talk, I will provide an overview of my team’s work attempting to remedying some of the above issues. Precisely, I discuss two novel methods we recently developed attaining state-of-the-art results on a variety of simulated robotic tasks.

Bio:

Haitham leads the decision-making team at Huawei technologies Research & Development UK. Prior to Huawei, Haitham led the reinforcement learning and tuneable AI team at PROWLER .io, where he contributed numerously to their technology in finance and logistics.

Prior to joining PROWLER .io, Haitham was an Assistant Professor in the Computer Science Department at the American University of Beirut (AUB). Before joining the AUB , Haitham was a postdoctoral research associate in the Department of Operational Research and Financial Engineering (ORFE) at Princeton University. Prior to Princeton, Haitham conducted research in lifelong machine learning while being employed as a postdoctoral researcher at the University of Pennsylvania. Being a former member of the General Robotics Automation Sensing and Perception (GRASP) lab, he also contributed to the application of machine learning to robotics.

Haitham acquired his PhD in Artificial Intelligence (AI) at Maastricht University in the Netherlands. He shortened a four-year study in two after publishing over 30 articles in world-leading AI and machine learning conferences and journals. He attained his Masters in Mechatronics Engineering with a summa cum-laude from the University of Applied Sciences in Ravensburg-Weingarten in Germany. Being the basis for his Master studies, Haitham acquired his Bachelors in Mechatronics Engineering from the Harriri Canadian University in Lebanon.

His primary research interests lie in the field of statistical machine learning and artificial intelligence, focusing on lifelong learning, multitask learning, knowledge transfer, and reinforcement learning. He is also interested in learning using massive amounts of data over extended time horizons – a property common to “Big-Data” problems. His research also spans different areas of control theory and nonlinear dynamical systems, as well as social networks and distributed optimization.