Understanding human decision making using machine learning

In this project we develop new machine learning methods to understand how humans take decisions given the information they are presented with. Existing machine learning methods like behaviour cloning or inverse reinforcement learning are not interpretable and often are not able to learn correctly the underlying informational and causal structure involved in making decisions or the underlying rewards motivating these decisions. This research will develop new methods for inverse reinforcement learning as well as uncertainty estimates associated with what decision will be taken, when and why given the available partially incomplete and potentially noisy information they are presented with.