Preprint
Oct 2026
Differentially Private Sparse Reward Estimation with Preference Feedback
Reinforcement Learning with Human Feedback (RLHF) has become a central paradigm for aligning Large Language Models with human values. However, the reliance on sensitive preference data necessitates rigorous differential privacy (DP) guarantees to prevent user information leakage. Existing private alignment methods typi...
Meng Ding, Ming-Xi Lei, Jie Zhang et al.
· 1 citation