Publications

“Deep lifts”: Neural networks can generalize even with the dumbest optimizer
Sunil Madhow and Yu-Xiang Wang.
Manuscript, 2026.
arXiv

Learnable Chernoff Baselines for Inference-Time Alignment
Sunil Madhow, Yuchen Liang, Ness Shroff, Yingbin Liang, and Yu-Xiang Wang.
Advances in Neural Information Processing Systems (NeurIPS), 2026.
arXiv

AKORN: Adaptive knots generated online for regression splines
Sunil Madhow, Dheeraj Baby, and Yu-Xiang Wang.
International Conference on Machine Learning (ICML), 2025.
Paper · Code

Rates for offline reinforcement learning with adaptively collected data
Sunil Madhow, Dan Qiao, Ming Yin, and Yu-Xiang Wang.
Learning for Dynamics and Control (L4DC), 2025.
Paper · arXiv