Focus Period Linköping 2022: PhD Course

Focus period Linköping 2022

PhD Course – Optimal Control and Reinforcement Learning (3 credits)

In connection with the ELLIIT focus period on Hybrid AI, a PhD course on optimal control and reinforcement learning will be offered.

Intended audience

Students who want to get an introduction to optimal control and reinforcement learning.

The course is based on a manuscript by Anders Hansson, Linkoping University and Martin Andersen, Technical University of Denmark, to be published by Wiley. We will cover Partially Separable Optimization Problems, Finite Horizon Optimal Control, Parametric Approximation, Infinite Horizon Optimal Control, Value Iterations, Policy Iterations, Linear Programming Formulations, Stochastic Optimization, Markov Decision Processes, Reinforcement Learning with Value Iterations and with Policy Iterations, Reinforcement Learning using Linear Programming, Reinforcement Learning using Approximations in Policy Space, Stochastic Optimization Methods, Root Finding Algorithms, Iterative Learning Control, and Iterative Feedback Tuning.

Organization

There will be 5 lectures. The students are expected to solve 4 homework problems for passing the course. There is no exam.

Lectures

October 18, 13:15–15:00
October 21, 08:15–10:00
October 24, 13:15–15:00
October 26, 08:15–10:00
October 28, 08:15–10:00

Lecures will be held in Systemet (B-house, 3rd floor).

Homeworks

All homeworks should be handed in at the lecture on paper. In addition to that matlab-files used for solving the homeworks should be sent as zipped files using e-mail to the examiner. You will have approximately two days to solve each homework.

Literature

Copies of the manuscript will be handed out at the lectures. Slides for the lectures will be made available as pdf-files.

Examiner

Anders Hansson, Department of Electrical Engineering (anders.g.hansson@liu.se, phone: +46 703004401)

Distance Mode

It is not possible to follow the course in distance mode.