Research

Research Interests

• Statistical Learning Theory
• Differential Privacy
• Algorithmic Stability
• LLM Post-Training

Publications

* for equal contribution, # for correspondence

• Feature Bagging Provides Stability. ICML 2026.
  Yuheng Ma, Qiang Sun.
• PrAda-GAN: A Private Adaptive Generative Adversarial Network with Bayes Network Structure. AAAI 2026.
  Ke Jia*, Yuheng Ma*, Yang Li, Feifei Wang.
• Bagged Regularized k-Distances for Anomaly Detection. JMLR 2025.
  Yuchao Cai, Hanfang Yang#, Yuheng Ma#, Hanyuan Hang.
• ALTER: Augmentation for Large-Table-Based Reasoning. NAACL 2025.
  Han Zhang, Yuheng Ma, Hanfang Yang.
• Locally Private Estimation with Public Features. AISTATS 2025.
  Yuheng Ma, Ke Jia, Hanfang Yang.
• Optimal Locally Private Nonparametric Classification with Public Data. JMLR 2024.
  Yuheng Ma, Hanfang Yang.
• Better Locally Private Sparse Estimation Given Multiple Samples Per User. ICML 2024.
  Yuheng Ma, Ke Jia, Hanfang Yang.
• Decision Tree for Locally Private Estimation with Public Data. NeurIPS 2023.
  Yuheng Ma, Han Zhang, Yuchao Cai, Hanfang Yang.
• Extrapolated Random Tree for Regression. ICML 2023.
  Yuchao Cai*, Yuheng Ma*, Yiwei Dong, Hanfang Yang.

Working Papers

• Locally Private Nonparametric Contextual Multi-armed Bandits with Transfer Learning.
  Yuheng Ma, Feiyu Jiang, Zifeng Zhao, Hanfang Yang, Yi Yu.
• Feature-Subsampled and Shared-Embedding Ensemble Networks for Uplift Modeling.
  Shaofan Liu, Shikai Luo, Yuheng Ma, Xvdong Zheng, Qiang Sun.

Services

* for outstanding reviewer

Conference reviewer: ICML (2024, 2025, 2026*), NeurIPS (2024*, 2025), AAAI (2025, 2026), ICLR (2025, 2026), AISTATS (2025, 2026).
Journal reviewer: JMLR, Biometrika, Bernoulli, TMLR.