Dr. Tadeusz Dobrowiecki, Department of Measurement and Information Systems

  mathematical logic, probability theory, computer science basics

A kötelező előtanulmányi rend az adott szak honlapján és képzési programjában található.

The aim of the subject is a short, yet substantial presentation of the field of artificial intelligence. The principal presented topics are (1) expressing intelligent behavior with computational models, (2) analysis and application of the formal and heuristic methods of artificial intelligence, (3) methods and problems of practical implementations. The subject is intended to develop the abilities and skills of the students of informatics in the area of:

- studying novel applications of the computing,

- developing effective methods to solve computational problems,

- understanding the technological and conceptual limits of the computer science,

- intellectual understanding of the central role of the algorithm in information systems.

Agent paradigm: Intelligent system and its environment. Formal modeling and solving of complex problems within agent paradigm. Comparing problem solving methods (search strategies). Heuristics for reducing complexity. Knowledge intensive approach and complexity. Experimenting with the scheduling problems: modeling within the paradigm and solving with the search algorithms.

Planning: Planning as a tool of problem solving. Basic representations for planning. The basics of the modern planning algorithms. Hierarchical and conditional planning. The question of the resource constraints. Integrated planning and execution. Experimenting with the assembly problems: developing plans taking into account various problems of increasing complexity.

Knowledge intensive systems. Formal representation and manipulation of knowledge. Logic based methods. Using first order logic to describe problems and to compute solutions. The functioning of rule-based systems. Inference methods for uncertain knowledge. Probabilistic inference systems. Representing vague meaning with fuzzy sets. Experimenting with the diagnostic problem with knowledge of different levels of uncertainty, using suitable methods, or experimenting with building a fuzzy system (rule-based language, fuzzy software packages, etc.).

Learning. Learning within agent paradigm. Inductive logical learning (decision trees, learning general logical expressions). Learning in neural and Bayesian networks. Reinforcement learning. Genetic algorithms and evolutionary programming. Experimenting with multiple learning problems, using suitable software packages.

During the term:

one midterm exam (scheduled beyond the weekly lecture), which can be reinstated acc. to the Code of Studies and Exams. The required minimum level is 40% (i.e. 20 points).

Acc. to the Code of Studies and Exams.

Stuart Russell and Peter Norvig: Artificial Intelligence. The Modern Approach, Prentice-Hall, 1995

Stuart Russell and Peter Norvig: Artificial Intelligence. The Modern Approach, 2^nd ed. Prentice-Hall, 2000