Entry requirements

120 credits including Probability and Statistics, Linear Algebra II, Single Variable Calculus and 10 credits of programming. Proficiency in English equivalent to the Swedish upper secondary course English 6.

Learning outcomes

On completion of the course, the student should be able to:

• explain possibilities and limitations of reinforcement learning,
• explain the connection between reinforcement learning and feedback control,
• analyse relevant applications, decide if they can be formulated as a reinforcement learning problem, and in such a case formulate it,
• implement and use central algorithms for reinforcement learning,
• analyse and evaluate methods through different performance criteria,
• implement, evaluate and present for the course relevant methods from the research literature.

Content

This course gives a solid introduction to the modern tools used to devise, implement and analyse reinforcement learning algorithms. The course covers Markov decision processes, feedback control systems, planning by dynamic programming, model-free prediction and control, the trade-off between exploration and exploitation, function approximations and policy-gradient methods. It also introduces deep reinforcement learning. Applications discussed in the course include classical control problems such as the inverted pendulum, as well as robotics and computer games.

Instruction

Lectures, seminars, computer labs, project.

Assessment

Oral and written examination of assignments (2 credits), project presentation (2.5 credits), and a written examination (3 credits).

If there are special reasons for doing so, an examiner may make an exception from the method of assessment indicated and allow a student to be assessed by another method. An example of special reasons might be a certificate regarding special pedagogical support from the disability coordinator of the university.

Other directives

The course cannot be included in the same degree as 1RT745 Reinforcement Learning.