Niemonotoniczne rozumowanie w filozofii i sztucznej inteligencji 3800-NMR23-S
Classical logic is monotone in the following sense: whenever a sentence A can be deduced from a set of sentences S, then A can be deduced from an arbitrary superset of S. Intuitively, this means that classical reasoning is non-defeasible: a conclusion which has been obtained cannot be undermined by the discovery of new facts. Yet, common sense reasoning is different. In various contexts (e.g., everyday reasoning, medical diagnosis, scientific reasoning) the knowledge base at our disposal is incomplete, and we need to draw plausible but uncertain conclusions which we might need to retract in light of further information. For example, based on the information that Mary and Tom are married, one can plausibly draw the conclusion that they live together. However, upon further discovery that Mary and Tom work in Poland and Portugal, respectively, one could revise the previously obtained conclusion.
The aim of this course is to study some of the best known formal approaches to non-monotonic reasoning. In particular, we will study the so called default logic as introduced in the classical paper Reiter (1980), and we will look at some related approaches, such as autoepistemic logic
and circumscription. The underlying idea behind default logic is that in an incompletely specified world, there is a need to draw (defeasible) conclusions by default, making assumptions of the form “in the absence of any information to the contrary, assume that ...”. Non-monotonic logics have a wide variety of applications in philosophy and artificial intelligence. We will study some of them, depending on time and the interests of participants.
Some seminar talks are planned to be given by invited guests, from both Philosophy and Computer Science.
Type of course
Prerequisites (description)
Course coordinators
Learning outcomes
- Acquired knowledge:
Students will obtain a solid understanding of some key aspects of the formalisms developed to model non-monotonic reasoning.
- Acquired skills:
Students will study how to implement sophisticated formal techniques in philosophy and artificial intelligence.
- Acquired social competences:
Students will have the ability to work in a team, and will understand and appreciate the need for training and professional development.
Bibliography
1. Bochman, A. (2008). Default logic generalized and simplified. Annals of Mathematics and Artificial Intelligence, 53, 21–49.
2. McCarthy, J. (1980). Circumscription – A Form of Non-Monotonic Reasoning. Artifical Intelligence, 13: 27–29.
3. Gabbay, D., Hogger, C., & Robinson, J. (eds.) (1994). Handbook of Logic in Artificial Intelligence and Logic Programming, Volume 3: Nonmonotonic Reasoning and Uncertain Reasoning. Oxford and New York: Oxford University Press.
4. Halpern, J. Y. (2003). Reasoning about uncertainty. Cambridge, Mass.: MIT Press.
5. Raymond, R. (1980). A Logic for Default Reasoning. Artifical Intelligence, 13: 81–132.
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: