Foundations of imitation learning

The research project Foundations of imitation learning aims to develop a statistically principled approach for specific machine learning problems, studying how to learn a task based only on demonstrations represented by trajectories of correlated data that are provided by experts.

Building on the proposed framework for imitation learning, this project will develop optimization-based algorithms that fundamentally contribute towards understanding observed behaviour in a data-driven and mathematically rigorous manner.

We plan to verify and test the developed methods and algorithms in specific projects in interdisciplinary collaboration within the Centre for the Advanced Study of Collective Behaviour (CASCB).