|
|
|
 |
Reasoning in artificial intelligence |
| |
|
15h L / 15h R / 1 OE / 2 ECTS credits / IIC_RIA |
| |
|
Fabrice POPINEAU |
| |
|
The goal of this course is the study of reasoning, how to formalize it and how to implement it on computers. The main line is the one of building an agent with an “intelligent” behavior, or rather a “rational” behavior with regard to its environment. Several methods for building such an agent are envisaged. First, search methods, both uninformed and informed are studied. Next, the key roles of logic in formalizing artificial intelligence problems are explained: real world formalization, reasoning formalization, proof theory. Classical logic is studied first with application to planning. Then common-sense reasoning is introduced thanks to non-classical logics: modal logics, non-monotonic logics. Finally, this tour of reasoning in artificial intelligence wouldn’t be complete without looking at the use of uncertain knowledge and probabilistic reasoning.
|
| |
|
Introduction : problem-solver agent |
| |
|
Definition of the AI paradigm in terms of agent and environment.
|
| |
|
State-space search |
| |
|
Uniformed and informed state-space search methods. Heuristics. 2-players games.
|
| |
|
Logic in AI |
| |
|
The roles of logic in AI. Logic programming and Prolog. SLD-resolution limitations. Meta-interpreters. Automated theorem proving techniques. Case study with Otter.
Formalization of reasoning with classical logic. Case study of situational calculus. The frame-problem.
Application to planning. Partial order planning.
|
| |
|
Non-monotonic reasoning in AI |
| |
|
Modal logics. Use of modal logic to express possibility, certainty, belief, knowledge or temporal progression.
Default logic. Introduction of models holding predicates with different extensions.
Model based methods: closed world assumption, predicate completion, circumscription.
|
| |
| |
| |
|
References
P. Norvig, S. Russell, Artificial Intelligence : A Modern Approach (2nd Edition), Prentice-Hall, 2002.
Z. Michalewicz, D. Fogel, How to Solve It: Modern Heuristics, Springer-Verlag, 1999.
A. Mackworth, R. Goebel, D. Poole, Computational Intelligence: A Logical Approach, Oxford University Press.
D. Gabbay (Ed.), Handbook of Logic in Artificial Intelligence, Clarendon Press, 1995.
I. Bratko, « Prolog Programming for Artificial Intelligence », Prentice-Hall, 2000.
|
| |
| |
| Last update 06/07/2007 by Cl.M. |
|
