Teodor C. Przymusinski

Professor of Computer Science and Engineering

University of California at Riverside

 


Education

M.S. in Mathematics (1970), University of Warsaw Poland

Ph.D. in Mathematics (1974), Institute of Mathematics, Polish Academy of Sciences

Habilitated Ph.D. in Mathematics (1979), Institute of Mathematics, Polish Academy of Sciences


Professional Career

Professor Teodor Przymusinski joined the Department of Computer Science and Engineering at the University of California at Riverside in July of 1991. Since 1992 he is also a cooperating faculty member in the Department of Electrical Engineering. Prior to that he served as a professor at several US and foreign universities and research institutions, including the Institute of Mathematics of the Polish Academy of Sciences, Steklov's Institute of Mathematics in Moscow, University of Toronto, University of Pittsburgh and University of Texas.

He is the recipient of two Scientific Awards from the Polish Academy of Sciences and the Outstanding Contribution Award at the First International Conference on Principles of Knowledge Representation and Reasoning in Toronto. He is also the recipient of several major grants funded by the National Science Foundation, Army Research Office, NATO, Swedish Technical Research Board and other agencies. He presented over 100 talks at various universities and conferences throughout the world, over 60 of which were invited, and is the author of over 100 research papers.

See Curriculum Vitae for more information (in PDF format).


Research Interests

Professor Przymusinski's research work belongs to the broad area of Artificial Intelligence in Computer Science. The main objective of this fascinating research field is to provide computers with the necessary skills and intelligence needed to perform tasks that currently require human involvement. As it was vividly demonstrated by the recent chess match between Anatoly Kasparov and IBM's Big Blue computer, which ended in the defeat of the world chess champion, the field of Artificial Intelligence (AI) made a big leap forward during the last 20-30 years of its existence. It continues to develop quite rapidly.

However, when compared to other tasks requiring intelligence, chess playing is relatively simple and that is why it turned out to be one of AI's early big successes. The chess game is subject to rather simple rules which, together with the overall objective of the game, can be easily represented in computer's memory using the language of classical mathematics. We still have to "teach" the computer how to use these mathematically encoded rules in order to achieve the desired goal but at least we have a well-defined knowledge representation language in which we can encode the task and its objective. Using such a representation, we can, at a minimum, use the incredible computing power of modern computers to analyze as many moves as possible and select the most promising one as our next move. This is, in fact, how most commercially available computerized chess games actually work.

The situation becomes much more complicated when it comes to dealing with more sophisticated forms of intelligent behavior. One of the most important and indispensable forms of such behavior is human reasoning. We reason when we ask and respond to questions while being engaged in a conversation, when we try to understand what someone else said or wrote, when we make plans and undertake decisions, when we react to new and unexpected situations. One can probably safely say that not a minute goes by without us having to engage in some - more or less complex - form of reasoning. Consequently, one cannot even imagine "smart" computers that would not be able to engage in intelligent reasoning.

As opposed to the game of chess, however, there in no obvious language in which one could represent the acquired knowledge and reason about it. Consequently, there is no obvious way of its adequate encoding in computer's memory. It has been pointed out long ago that classical mathematical logic, a natural candidate, in entirely unsuited to be a knowledge representation language. The main reason lies in the fact that human reasoning is strikingly different from that performed in mathematical logic. While logic allows us to derive only those conclusions that can be mathematically proved from the existing knowledge, human reasoning bases most of its conclusions on default assumptions and various forms of belief that are used in place of the missing factual information. That is why we often use the term commonsense reasoning to describe human reasoning.

Contrary to what one might expect, the "commonsense" nature of human reasoning is not a drawback but rather a very useful feature. Since most of our knowledge is greatly incomplete, it precludes logical derivation of almost any conclusion. Just try to actually prove (rather than to perform an experiment!) such a seemingly trivial fact that there is nothing that prevents you from walking to the opposite side of the room. Clearly, inability to reason in the absence of complete information would render us completely powerless.

Accordingly, an important task facing Artificial Intelligence is finding a suitable knowledge representation language in which one could encode the acquired knowledge and efficiently reason about it, update and revise it, as well as learn and communicate knowledge. The resolution of this problem is of fundamental importance to Artificial Intelligence, in particular, and to Computer Science as a whole.

Most of Professor Przymusinski's recent research work was devoted to the study of this important problem knowledge representation in AI. In particular, he studied the issues of formalizing commonsense reasoning within the framework of logic programming and its extensions. Logic programming constitutes a relatively narrow domain of commonsense reasoning which is therefore easier to study and more efficient to implement. Consequently, logic programs provide fruitful test beds to study the highly complex issues involved in formalizing commonsense reasoning. The results of his studies led to the introduction of powerful extensions of the logic programming paradigm that allow much more expressive representation of commonsense knowledge.

See List of Publications for more information.


Contact Information

 

Department of Computer Science and Engineering

College of Engineering

Engineering Bldg. II, Room 419

University of California

Riverside, CA 92521, USA

Office phone: (951) 827-5015

Department phone: (951) 827-5639

Fax: (951) 827-4643

WWW: http://www.cs.ucr.edu/~teodor

teodor@cs.ucr.edu


Counter

This page received   hits since August 1, 1997.

Last modified on 3/15/2007


CS&E Home Page