Provable Reward-Agnostic Preference-Based Reinforcement Learning
This work addresses the problem of reducing human feedback in preference-based RL for AI applications like language model fine-tuning, though it is incremental by building on existing theoretical models.
The paper tackles the gap between theoretical and practical preference-based reinforcement learning by proposing a reward-agnostic framework that acquires exploratory trajectories before human feedback, requiring less feedback for optimal policy learning under linear models and unknown transitions, with sample-efficient incorporation of linear and low-rank MDPs.
Preference-based Reinforcement Learning (PbRL) is a paradigm in which an RL agent learns to optimize a task using pair-wise preference-based feedback over trajectories, rather than explicit reward signals. While PbRL has demonstrated practical success in fine-tuning language models, existing theoretical work focuses on regret minimization and fails to capture most of the practical frameworks. In this study, we fill in such a gap between theoretical PbRL and practical algorithms by proposing a theoretical reward-agnostic PbRL framework where exploratory trajectories that enable accurate learning of hidden reward functions are acquired before collecting any human feedback. Theoretical analysis demonstrates that our algorithm requires less human feedback for learning the optimal policy under preference-based models with linear parameterization and unknown transitions, compared to the existing theoretical literature. Specifically, our framework can incorporate linear and low-rank MDPs with efficient sample complexity. Additionally, we investigate reward-agnostic RL with action-based comparison feedback and introduce an efficient querying algorithm tailored to this scenario.