In this course, we look into methods and ideas of Artificial Intelligence that are especially interesting or widely applicable in practice. Therefore, these methods are important in advanced applications of computers, as well as models of how human intelligence works. We will find answers to questions like:
- How to construct a plan for a given task that is to be solved jointly by a number of agents, e.g. robots or other machines, when operations may be executed in parallel? We face such problems in logistics, scheduling, or organization of complex operations. •
- How can an agent, e.g. a robot, learn to achieve its goals in a completely unknown environment, whereby the agent also has to discover the laws of its environment? •
- How can a computer automatically construct a program when nothing more than examples of input and output data for the program are given? We will study algorithms for automatic programming from examples. • How can optimization problems be solved by mimicking natural evolution – with natural selection when only the fittest survive? Genetic algorithms are based on this idea. •
- How can a computer predict the course of events by »common sense« reasoning, as people do in everyday life: without numbers, differential equations or numerical mathematical models? In most cases, »naive«, intuitive physics that children learn from everyday experience suffices.
Contents: space-efficient search algorithms, partial-order planning, reinforcement learning, genetic algorithms, qualitative reasoning and modelling, inductive logic programing, limits of learnability – what can be learned in limited time, and what can't.
***If foreign students are enrolled in the course it is held in English.
TeacherProf. Ivan Bratko, PhD