Gerald F. DeJong
- Title: Professor
- Group: Intelligence, Learning, and Plasticity
- Status: Beckman Affiliate Faculty
- Home: Computer Science
Gerald DeJong received his Ph.D. from Yale University in 1979. He is a professor in the Department of Computer Science at the U of I and a full-time Beckman Institute faculty member in the Artificial Intelligence group. His fields of professional interest are machine learning, planning, robotics, plausible inference, artificial intelligence, and cognitive modeling.
Achievement award from the National Aeronautics and Space Administration for contribution to "Advanced Development for the Deep Space Mission System" (1999); Fellow, American Association for Artificial Intelligence (1994); Alcoa Foundation Faculty Recognition Grant (1989); Arnold O. Beckman Research Award (1984); and Exxon Faculty Assistance Grant (1982).
Gerald DeJong's research concerns machine learning--how computers, by interacting with their environments, come to behave more effectively. In particular, he is one of the founders of Explanation-Based Learning in which declarative background knowledge is employed to guide learning. The resulting model is capable of acquiring a new concept from relatively few examples. The approach has been applied in diverse areas including problems in robotics, natural language processing, planning, and scheduling. Experiments have demonstrated the psychological plausibility of the model in both adults and as a developmental account of infant learning.
From the point of view of the Human-Computer Intelligent Interaction theme, the research investigates how robots can be taught, rather than programmed, to perform complex tasks. When training a human to perform a new task, an expert explains concepts having to do with goals and high level techniques, relying on the human's ability to interpret and instantiate these into successful plans. The human apprentice must often watch the expert perform the task a number of times, and he continues to improve with practice. Artificial Intelligence may allow computers to shoulder a similar responsibility for learning.
The research also fits within the Biological Intelligence area. Rational decision making is one application of DeJong's learning research. Humans are magnificent at arriving at appropriate conclusions and behavior most of the time. Indeed this is a hallmark, if not a potential definition of intelligence. However, it is well established that humans do not naturally reason by logical syllogisms. They employ some other inferential system. While their lack of logical sophistication can result in poor behavior on some tasks (like taking symbolic logic tests), more often is it formal logic that cannot be applied to complex human situations. The research has lead to some new developmental psychology experiments, as well as reinterpretation of previously established psychological results.
DeJong's sources of research support include the NSF, ONR, U.S. Army, and Yamaha.
DeJong, G. (1999), "AI Can Rival Control Theory for Goal Achievement in a Challenging Dynamical System," Computational Intelligence, 15/4, pp. 333-366.
Brodie, M. and DeJong, G. (1999), "Learning to Ride a Bicycle Using Iterated Phantom Induction," Proceedings of the International Conference on Machine Learning (Bled, Slovenia: Morgan Kaufmann), pp. 57-66.
Brodie, M. and DeJong, G. (1998), "Iterated Phantom Induction: A Little Knowledge Can Go a Long Way," Proceedings of the National Conference on Artificial Intelligence, Madison, WI, July, pp. 665-670.
Fijalkiewicz, P. and DeJong. G. (1998), "CHESHIRE: An Intelligent Adaptive User Interface," Second Annual Symposium on Advanced Displays (University of Maryland), pp. 15-19.
DeJong, G. F. and Bennett, S. W. (1997), "Permissive Planning: Extending Classical Planning to Uncertain Task Domains," Artificial Intelligence, 89/1-2, pp. 173-217.
DeJong, G. (1997), "Explanation-based Learning," in A. Tucker, ed., Encyclopedia of Computer Science (Boca Raton: CRC Press), pp. 499-520.
The Communications Office maintains the information included in Beckman Institute's online directory listings. In order to update your directory listing, please submit the following information to email@example.com:
- a short bio including information on your educational background and your field
- any honors and awards you may have received
- a description of your research (approximately 200-400 words)
- a list of recent representative publications
- a photo of yourself (you can submit one or we can take one for you)