Associate Prof Jonathan Shock 

2nd semester 
20 credits / 30 lectures

In this project-based module on modern methods of artificial intelligence, the student will investigate advanced topics in Reinforcement Learning (RL). In the past, these have included multi-agent RL, curiosity based RL, causality, meta-RL and many more. There will be weekly discussion sessions where students will give mini-presentations, and share ideas. The first three weeks gets everyone up to scratch on deep neural networks applied to RL. The final grade is based on a mini-thesis.

Prerequisites are to have read "Reinforcement Learning" by Sutton and Barto: and to have a good grounding in Python.