From Reinforcement Learning to Artificial Intelligence

Satinder Singh

Abstract:
The field of reinforcement learning (RL) has made considerable
progress in applications from control and operations research. I will
briefly discuss the mathematical basis for these successes and then
spend the bulk of my talk laying out some of the research challenges
and questions that need to be addressed to allow RL to achieve
artificial intelligence. For some of these questions I will present
preliminary answers from research from my group.  I will conclude with
some ideas for future research.