HOEK AND WITTEVEEN

Logic and Computation
ABDUCTION AND DIAGNOSIS: LOGICAL AND COMPUTATIONAL ASPECTS

Introductory course

WIEBE VAN DER HOEK, CEES WITTEVEEN and ANDRE BOS

Department of Computer Science, University of Utrecht,

Department of Computer Science, Technical University of Delft and

The Netherlands Organization of Applied Research (TNO), Delft

Second week
wiebe@cs.ruu.nl, witt@cs.tudelft.nl and bos-a@tpd.tno.nl

Course description

We present the main ideas behind abduction and diagnosis, discuss their logical foundations and computational aspects of applications.

We discuss some main principles of Nonmonotonic Reasoning and present several formalisms.

In particular, the plan of the course is as follows:

1) General introduction to abduction and diagnosis. We introduce the main ideas behind abduction and diagnosis, their relation to non-monotonic reasoning and some of their formalizations. Examples and exercises show this formalism at work.

2) The relation to non-monotonic reasoning and theory-revision. We show that both abduction and diagnosis can be modeled as special forms of non-monotonic reasoning and that there also is a clear relation to the field of theory-revision.

3) Computational aspects of abduction and diagnosis. We give an overview of the computational (in)tractability results for several forms of abduction and diagnosis, showing that, in general, these reasoning processes are computationally intractable.

4) Approaches to tractable implementations of diagnosis and abduction. In this part we discuss some general strategies to overcome the computational difficulties encountered in the previous part. We show how model-compilation might be applied to guarantee tractability in some cases and we also show some recent approaches using probabilistic and approximation techniques to make diagnosis and abduction feasible processes.

Prerequisites
None

Literature
No specific recommendation

HOME PROGRAMME CONTACT REGISTRATION

Logic and Computation	ABDUCTION AND DIAGNOSIS: LOGICAL AND COMPUTATIONAL ASPECTS
Introductory course	WIEBE VAN DER HOEK, CEES WITTEVEEN and ANDRE BOS Department of Computer Science, University of Utrecht, Department of Computer Science, Technical University of Delft and The Netherlands Organization of Applied Research (TNO), Delft
Second week	wiebe@cs.ruu.nl, witt@cs.tudelft.nl and bos-a@tpd.tno.nl
Course description	We present the main ideas behind abduction and diagnosis, discuss their logical foundations and computational aspects of applications. We discuss some main principles of Nonmonotonic Reasoning and present several formalisms. In particular, the plan of the course is as follows: 1) General introduction to abduction and diagnosis. We introduce the main ideas behind abduction and diagnosis, their relation to non-monotonic reasoning and some of their formalizations. Examples and exercises show this formalism at work. 2) The relation to non-monotonic reasoning and theory-revision. We show that both abduction and diagnosis can be modeled as special forms of non-monotonic reasoning and that there also is a clear relation to the field of theory-revision. 3) Computational aspects of abduction and diagnosis. We give an overview of the computational (in)tractability results for several forms of abduction and diagnosis, showing that, in general, these reasoning processes are computationally intractable. 4) Approaches to tractable implementations of diagnosis and abduction. In this part we discuss some general strategies to overcome the computational difficulties encountered in the previous part. We show how model-compilation might be applied to guarantee tractability in some cases and we also show some recent approaches using probabilistic and approximation techniques to make diagnosis and abduction feasible processes.
Prerequisites	None
Literature	No specific recommendation