7 papers
Sensitivity Shaping for Latent Modeling
Hongzhan Yu, Chenghao Li, Ruipeng Zhang +2
Generative dynamics models enable planning in challenging systems, but safe deployment requires detecting policy-induced out-of-distribution (OOD) transitions. Existing methods typ…
P-DPO: Grounding Hallucination in Perceptual Processing via Calibration Direct Preference Optimization
Ruipeng Zhang, Zhihao Li, Haozhang Yuan +2
Hallucination has recently garnered significant research attention in Large Vision-Language Models (LVLMs). Direct Preference Optimization (DPO) aims to learn directly from the cor…
Steer Where It Matters: Token-Level Visual-Sensitivity Steering for LVLMs Hallucination Mitigation
Ruipeng Zhang, Zhihao Li, C. L. Philip Chen +1
Large vision language models (LVLMs) have made rapid advancements and are deployed across various applications, yet hallucinations remain a major challenge. Activation steering is…
Learning Quadruped Walking from Seconds of Demonstration
Ruipeng Zhang, Hongzhan Yu, Ya-Chien Chang +3
Quadruped locomotion provides a natural setting for understanding when model-free learning can outperform model-based control design, by exploiting data patterns to bypass the diff…
Sequence Modeling for Time-Optimal Quadrotor Trajectory Optimization with Sampling-based Robustness Analysis
Katherine Mao, Hongzhan Yu, Ruipeng Zhang +4
Time-optimal trajectories drive quadrotors to their dynamic limits, but computing such trajectories involves solving non-convex problems via iterative nonlinear optimization, makin…
When Maximum Entropy Misleads Policy Optimization
Ruipeng Zhang, Ya-Chien Chang, Sicun Gao
The Maximum Entropy Reinforcement Learning (MaxEnt RL) framework is a leading approach for achieving efficient learning and robust performance across many RL tasks. However, MaxEnt…
Improving Value Estimation Critically Enhances Vanilla Policy Gradient
Tao Wang, Ruipeng Zhang, Sicun Gao
Modern policy gradient algorithms, such as TRPO and PPO, outperform vanilla policy gradient in many RL tasks. Questioning the common belief that enforcing approximate trust regions…