Bio
Tom's PhD is in explainable reinforcement learning at the University of Bristol. He is interested in understanding the "full stack" of explainable agency, from the theoretical foundations (how can we construct accurate abstract models of evolving behaviour?) to human-computer interaction (what visualisations best communicate this complex information to a user?) Tom's ultimate goal is to integrate explainability methods into a pipeline for interactive reinforcement learning, in which humans and machine learning algorithms collaborate to construct aligned and trustworthy behavioural policies.