Many real world tasks performed by humans, such as driving an automobile or playing a complex game, require complex real-time behaviors, which pose significant challenges to current machine learning and AI techniques. In this talk I will summarize our recent work on two complementary approaches for defining such complex behaviors: learning from demonstration and game tree search. Concerning learning from demonstration, I will focus on our work on modeling human automobile driving behavior, and illustrate the key theoretical differences between learning from demonstration and supervised machine learning. Concerning game tree search, I will focus on game playing behavior, and on the challenge of playing games with very large action spaces, beyond what state-of-the-art search algorithms can handle. I will conclude with a discussion of open challenges and future research directions in these research areas.

Santiago Ontañón is an Assistant Professor in the Computer Science Department at Drexel University. His main research interests are machine learning, case-based reasoning and game AI. He has published more than 130 peer-reviewed papers and articles. He obtained his PhD form the Autonomous University of Barcelona (UAB), Spain. Before joining Drexel University, he held postdoctoral research positions at the Artificial Intelligence Research Institute (IIIA) in Spain and at the Georgia Institute of Technology. He also lectured at the University of Barcelona (UB), Spain.