Quantifying the Sensitivity of Inverse Reinforcement Learning to Misspecification

Open in new window