1148 Kelley Engineering Center
Corvallis, OR 97331
Alan Fern is a Professor of Computer Science and Executive Director of A.I. Research for the College of Engineering at Oregon State University. He received his Ph.D. (2004) in Computer Engineering from Purdue University, and his B.S. (1997) in Electrical Engineering from the University of Maine. His research interests span a variety of topics in Artificial Intelligence and Robotics with a particular emphasis on building systems that can learn from experience. He co-directs the Dynamic Robotics Laboratory with Jonathan Hurst at Oregon State and is PI for a number of government funded projects including DARPA programs on Explainable Artificial Intelligence and Machine Common Sense. Most recently he is serving as the OSU lead PI for a new $20M AI Institute on Agricultural AI in collaboration with Washington State University.
CS 533 INTELLIGENT AGENTS & DECISION
My research is centered around my interest in making machines “smarter.” Machines are very far from rivaling humans in tasks such as visual understanding, planning a trip, playing real-time strategy games, language processing, tutoring, etc. Nevertheless, I believe that even small steps toward solving such tasks will yield systems that people will soon take for granted and depend on in everyday life. My general career vision is to support this progress by studying the critical computational problems that arise.
My research interests are in the artificial intelligence areas of machine learning, automated planning, and knowledge representation. I am most interested in usefully integrating these complementary areas.
One of my research thrust is to develop algorithms that can learn to interpret complex data such as videos. As an example, in the Digital Scout Project, we are studying the use of machine learning for computing interpretations of American football video. Such interpretations will allow for coaches, analysts, and fans to easily pose queries against libraries of football video for indexing and collecting statistics.
A second research thrust is to develop techniques for integrating planning and machine learning. Our work has developed new planning algorithms that can learn to solve very large problems that were beyond the scope of previous techniques. We are now extending that work to richer settings, for example, multi-agent planning, where agents must cooperate to complete a task. As a testbed we are studying planning and learning in the context of real-time strategy games, a domain where humans are currently far better than computer players.