Identifiability and Generalizability in Constrained Inverse Reinforcement Learning
This work addresses theoretical challenges in designing safe reward functions for reinforcement learning, but it is incremental as it extends prior results to constrained settings.
The authors tackled the problem of reward identifiability and generalizability in constrained inverse reinforcement learning, showing that identifiability up to potential shaping may not hold with other regularizations or safety constraints, and derived a finite sample guarantee for suboptimality.
Two main challenges in Reinforcement Learning (RL) are designing appropriate reward functions and ensuring the safety of the learned policy. To address these challenges, we present a theoretical framework for Inverse Reinforcement Learning (IRL) in constrained Markov decision processes. From a convex-analytic perspective, we extend prior results on reward identifiability and generalizability to both the constrained setting and a more general class of regularizations. In particular, we show that identifiability up to potential shaping (Cao et al., 2021) is a consequence of entropy regularization and may generally no longer hold for other regularizations or in the presence of safety constraints. We also show that to ensure generalizability to new transition laws and constraints, the true reward must be identified up to a constant. Additionally, we derive a finite sample guarantee for the suboptimality of the learned rewards, and validate our results in a gridworld environment.