David Boyle

Inducing, Detecting and Characterising Neural Modules: A Pipeline for Functional Interpretability in Reinforcement Learning

Constrained Reinforcement Learning as Wasserstein Variational Inference: Formal Methods for Interpretability