Provable Offline Preference-Based Reinforcement Learning
This addresses the problem of learning from human preferences without explicit rewards in offline settings, which is incremental as it builds on existing PbRL methods by providing theoretical guarantees.
The paper tackles offline preference-based reinforcement learning with human feedback, proposing an algorithm that estimates implicit rewards and performs robust planning, achieving a polynomial sample guarantee for learning any target policy covered by the data, which is the first such result with general function approximation.
In this paper, we investigate the problem of offline Preference-based Reinforcement Learning (PbRL) with human feedback where feedback is available in the form of preference between trajectory pairs rather than explicit rewards. Our proposed algorithm consists of two main steps: (1) estimate the implicit reward using Maximum Likelihood Estimation (MLE) with general function approximation from offline data and (2) solve a distributionally robust planning problem over a confidence set around the MLE. We consider the general reward setting where the reward can be defined over the whole trajectory and provide a novel guarantee that allows us to learn any target policy with a polynomial number of samples, as long as the target policy is covered by the offline data. This guarantee is the first of its kind with general function approximation. To measure the coverage of the target policy, we introduce a new single-policy concentrability coefficient, which can be upper bounded by the per-trajectory concentrability coefficient. We also establish lower bounds that highlight the necessity of such concentrability and the difference from standard RL, where state-action-wise rewards are directly observed. We further extend and analyze our algorithm when the feedback is given over action pairs.