Curriculum Vitae (ps)

Research

Robert Levinson's research crosses the broad spectrum of topics in Artificial Intelligence research - specifically in the areas of machine learning, heuristic search, knowledge representation, associative pattern retrieval and machine creativity. His major project, unifying his research, has been an adaptive pattern-oriented chess system named Morph, a system that learns to play respectable chess from its experience only.

Levinson's primary research goal is to give computers the benefit of experience. He believes that the fact that most computers do not make use of previous experience is a tremendous waste of computation. To serve his research goal he has developed a domain-independent learning architecture APS by which systems can improve their decision making and predictive abilities over time with little human assistance. To further his goal he has developed sophisticated algorithms for the content-based retrieval of structural patterns. These algorithms are being used in the Knowledge Representation community and commercially in Gerard Ellis's Santiago system. Levinson's methods have been or are currently being applied in the domains of organic chemistry, signal recognition, image alignment, medical diagnosis, and manufacturing systems design in addition to chess and other games.

Levinson is co-authoring a book called Scientific Thinking: A Systems Approach with Andrew Goodwin. The draft version of it is currently out for peer review.