On the Feasibility of Learning, Rather than Assuming, Human Biases for Reward Inference

Rohin Shah, Noah Gundotra, Pieter Abbeel, and Anca Dragan. ICML 2019. (Poster PDF · Paper)