Fahiem Bacchus

(Professor).
BSc (Alberta '79), MSc (Toronto '83), PhD (Alberta '88).

fbacchus@cs.toronto.edu

My research interests revolve around various reasoning and representation problems in AI. My main foci these days include CSPs (constraint satisfaction problems), planning, SAT (satisfiability), and Bayesian Inference. I am interested in developing algorithms for these problems that can achieve better performance by taking advantage of domain specific knowledge or structure.

For example, F. Kabanza and I developed an approach whereby extra domain specific knowledge can be represented declaratively and used to solve planning problems. Our approach can obtain orders of magnitude increases in performance from even very simple domain knowledge. The TLPlan planning system, developed with M. Ady, utilized this approach to planning and demonstrated the effectiveness of the approach by winning the AIPS2002 international planning competition).

Another example is my 2clseq SAT solver which utilizes extensive binary clause reasoning exploiting the "binary substructure" of a clausal theory. When run as a preprocessor this kind of reasoning greatly simplifies clausal theories, sometimes converting problems unsolvable by current SAT solvers into problems they can easily solve.

In the navigation frame to the left are links to on-line versions of most of my papers along with other material I have written; various pieces of software I have been involved in developing that are available for download (some with sources); some conferences whose organization I am involved in; and links to some of the courses I teach.

Fahiem Bacchus
Dept. Computer Science
6 King's College Road
University of Toronto
Toronto, Ont, M5S 3H5
CANADA
416-946-7174
FAX: 416-978-1455