Deep Reinforcement Learning

What it is about

You will learn modern methods of Reinforcement Learning (RL) and apply them. The module conveys both the theoretical foundations (e.g., Markov decision processes, Monte Carlo, Temporal Difference) as well as current Deep-RL methods (e.g., Q-Learning, DQN, Policy-Gradient methods, PPO, Actor-Critic). By the end you can formulate real problems as RL tasks and implement solution concepts in Python.

What you will be able to do

• Deepened understanding of the concepts of reinforcement learning
• Explanation of classical algorithms (e.g., Q-Learning, SARSA, DQN, Policy Gradient)
• Modeling real problems as reinforcement learning tasks
• Implementation of RL solutions in Python

What the module consists of

• VorlesungDelivery of theoretical foundations and illustration of examples and applications
• Projekt/ÜbungPractical application of the learned algorithms in group work and homework

Teaching method

• VorlesungspräsentationenIntroduction to theory and demonstration of examples
• Projektarbeit und HausaufgabenPractical application and deepening of the algorithms through implementation and report
• Diskussionen/PräsentationenReflection of results and critical engagement in the final presentation