Sub-symbolic AI Methods

This course will be taught every year. It is the same as IT3708 (Subsymbolic AI) but for PhD students.
The main focus of the course is to build intelligent systems based on two key natural concepts: the brain, and evolution by natural selection. In computer-science, the analogs for these are artificial neural networks (ANNs) and evolutionary algorithms (EAs). Both methods have thousands of useful applications in fields as diverse as control theory, telecommunications, music and art. This course discusses both methods in great detail along with providing a bit of the biological basis for each.

Students will get both theoretical and practical programming experience with two of the best known sub-symbolic AI methods: artificial neural networks and evolutionary algorithms.

Regular lectures, homeworks and a projects, along with a take-home final exam. The final grade is based 50% on the homeworks/projects and 50% on the take-home exam.
This course is VERY programming intensive, with each homework taking 2-4 weeks to complete. There are normally 4-5 such homework assignments.
Group work on homeworks is acceptable, but group size cannot
exceed 2 members. The take home exam is to be done individually, with absolutely no discussion with other students. Violation of this rule will result in a failing mark for the course.