Jingyan Shen

reward modeling (2) data poisoning (1) principal component analysis (1) data valuation (1) reinforcement learning from human feedback (1) outlier detection (1) model alignment (1) backdoor attack (1) pluralistic alignment (1) human preference (1) interpretable ai (1) backdoor attack detection (1) mixture of preference (1) personalized preference learning (1) binary preference dataset (1) out-of-bag estimation (1) backdoor trigger (1) fine-grained outlier (1) data contribution (1)