SLIDE 1
V Goranko
V Goranko
Logic in Computer Science, Artificial Intelligence and Multi-agent - - PowerPoint PPT Presentation
Logic in Computer Science, Artificial Intelligence and Multi-agent - - PowerPoint PPT Presentation
Logic in Computer Science, Artificial Intelligence and Multi-agent Systems Introduction Valentin Goranko AlgoLog, DTU Informatics Aug 30, 2010 V Goranko Lecturers info Lecturer: Assoc. Prof. Valentin Goranko Office: bld. 322 / room 117
SLIDE 2
SLIDE 3
V Goranko
Course info
◮ Weekly teaching slots: Mondays, 08.00-12.00, in 322/033. Tentative schedule:
- 08.00 - 08.30: Tutorial session: questions and discussion on
previous lecture, on exercises, assignments, etc.
- 08.30 - 10.30: Lectures.
- 10.40 - 12.00: Exercise session.
◮ Course webpage: see link on CampusNet. ◮ Lecture notes and additional course material: will be provided during the course. ◮ Exercises: lists of selected exercises will be provided weekly. ◮ Assignments: two mandatory written individual assignments, given in weeks 6 and 11. ◮ Exam: 4 hours written exam, to be discussed later.
SLIDE 4
V Goranko
The role of Logic in CS, AI and MAS
During the second half of the 20th century Formal Logic becomes a fundamental tool in CS, AI, Linguistics, and more recently, in the study of agents and MAS, as well. Logic provides a powerful and practically useful framework for:
- precise mathematical characterization of the notions of
deduction and computation and of the limits of deducibility and computability;
- knowledge representation and reasoning;
- deductive reasoning, incl. tools for automated reasoning;
- specification, modelling, and formal verification of systems.
The role and influence of Logic in CS and AI is compared to the role of Calculus in Natural Sciences over the past 300 years. For the success of Logic in CS see: J. Halpern, et al: ”On the Unusual Effectiveness of Logic in Computer Science”.
SLIDE 5
V Goranko
Logical reasoning with classical logic
Formal logical reasoning typically uses classical (first-order) logic. Advantages:
- rich and uniform language for knowledge representation,
- relatively simple syntax and well-understood semantics;
- well-developed deductive systems and tools for automated
reasoning. Disadvantages:
- cannot express some important properties, e.g. finiteness.
- cannot capture well some aspects natural language;
- cannot capture adequately specific modes of reasoning;
- algorithmic undecidability of logical consequence and validity.
SLIDE 6
V Goranko
Aspects of logical reasoning in AI with non-classical logics
More flexible, task-specific, and computationally better behaved reasoning frameworks are based on various modal logics, including:
- Reasoning about programs and processes.
Dynamic and process logics.
- Temporal and spatial reasoning. Temporal and spatial logics.
- Reasoning about knowledge and beliefs. Epistemic logics.
- Reasoning about obligations and rights. Deontic logics.
- Reasoning about ontologies. Description logics.
- Reasoning about agents and their knowledge, beliefs,
intentions, desires, actions, strategies, etc. Logics for multi-agent systems.
- Non-monotonic reasoning and logics.
SLIDE 7
V Goranko