AI architectures and reinforcement learning

At the University of Michigan, a group of us (including Satinder Singh, John Laird, and Thad Polk, see left panel below) is embarking on a new project to develop computational agents that operate for extended periods of time in rich and dynamic environments, and achieve mastery of many aspects of their environments without task-specific programming. To accomplish these goals, our research is exploring a space of cognitive architectures that incorporate four fundamental features of real neural circuitry: (1) reinforcing behaviors that lead to intrinsic rewards (2) executing and learning over mental, as well as, motor actions, (3) extracting regularities in mental representations, whether derived from perception or cognitive operations, and (4) continuously encoding and retrieving episodic memories of past events.

This kind of basic artificial intelligence research is critical for advancing cognitive science: we can't pretend to understand human cognition until we understand how any computational system might be organized to achieve long-lived, autonomous, adaptive behavior in complex environments. At the moment, no set of ideas in cognitive science or AI has been shown to satisfy these functional demands.

To learn more about reinforcement learning, check out Satinder Singh's website here at Michigan.

relevent publications

Attend to the copyright notice.

Singh, S., Lewis, R. L., and Barto, A. G. (2009). Where do rewards come from? In Proceedings of the Annual Conference of the Cognitive Science Society, pages 2601-2606, Amsterdam. [ DOWNLOAD PDF ]

Pearson, D., Gorski, N. A., Lewis, R. L., and Laird, J. E. (2007). Storm: A framework for biologically-inspired cognitive architecture research. In Lewis, R., Polk, T., and Laird, J., editors, The Proceedings of the 8th International Conference on Cognitive Modeling. Psychology Press/Taylor & Francis. [ DOWNLOAD PDF ]

Lewis, R. L. (2001). Cognitive theory, Soar. In International Encylopedia of the Social and Behavioral Sciences, pages 2178-2183. Pergamon (Elsevier Science), Amsterdam. [ DOWNLOAD PDF ]

These references were generated by bibtex2html 1.93.