Showing cs.LGShow all
3 papers · 1 filter
cs.LG2026
Revealing After Overwriting: An Exponential POMDP OPE Lower Bound under History-Dependent Logging
Youyu Luo, Pengzhan Zhou, Zhida Qin +4
Multi-step revealing can make off-policy evaluation tractable under memoryless logging. With history-dependent logging, state decodability and target-relevant evidence can separate…
cs.LG2026
Mitigating Gradient Pathology in PINNs through Aligned Constraint
Yichen Luo, Peiyu Zhu, Dongxiao Hu +5
While Physics-Informed Neural Networks (PINNs) are powerful for solving Partial Differential Equations (PDEs), their training is often paralyzed by gradient pathology. The gradient…
cs.LG2026
CFG: Control Classifier-Free Guidance via Score Discrepancy Analysis
Jiayang Gao, Tianyi Zheng, Jiayang Zou +9
Classifier-Free Guidance (CFG) is a cornerstone of modern conditional diffusion models, yet its reliance on the fixed or heuristic dynamic guidance weight is predominantly empirica…