Publications

Selected papers and preprints.

2026

  1. A Van Trees Lower Bound for Fully Interactive Differentially Private Federated Learning
    T. Tony Cai and Yicheng Li
    arXiv preprint, arXiv:2605.19813v2, Jul 2026
    arXiv
  2. Large Dimensional Kernel Ridge Regression: Extending to Product Kernels
    Yang Zhou, Yicheng Li, Yuqian Cheng, and Qian Lin
    arXiv preprint, arXiv:2605.14524, May 2026
    arXiv
  3. Alignment-Sensitive Minimax Rates for Spectral Algorithms with Learned Kernels
    Dongming Huang, Zhifan Li, Yicheng Li, and Qian Lin
    arXiv preprint, arXiv:2509.20294, May 2026
    arXiv
  4. Minimax and Adaptive Covariance Matrix Estimation under Differential Privacy
    T. Tony Cai and Yicheng Li (alphabetical order)
    arXiv preprint, arXiv:2603.19703, Mar 2026
    arXiv

2025

  1. Several Supporting Evidences for the Adaptive Feature Program
    Yicheng Li and Qian Lin
    arXiv preprint, arXiv:2511.09425, Nov 2025
    arXiv
  2. Differentially Private Functional Linear regression: Optimal Algorithms and Theoretical Guarantees
    Yicheng Li and T. Tony Cai
    Sep 2025
  3. Functional Virtual Adversarial Training for Semi-Supervised Time Series Classification
    Qingyi Pan and Yicheng Li (Corresponding author)
    In 39th Conference on Neural Information Processing Systems (NeurIPS 2025), Sep 2025
    OpenReview
  4. Diagonal Over-Parameterization in Reproducing Kernel Hilbert Spaces as an Adaptive Feature Model: Generalization and Adaptivity
    Yicheng Li and Qian Lin
    arXiv preprint, arXiv:2501.08679, Jan 2025
    arXiv
  5. Optimal Rates of Kernel Ridge Regression under Source Condition in Large Dimensions
    Haobo Zhang, Yicheng Li, Weihao Lu, and Qian Lin
    Journal of Machine Learning Research, 2025
    URL

2024

  1. Towards a Statistical Understanding of Neural Networks: Beyond the Neural Tangent Kernel Theories
    Haobo Zhang, Jianfa Lai, Yicheng Li, Qian Lin, and Jun S. Liu
    arXiv preprint, arXiv:2412.18756, Dec 2024
    arXiv
  2. On the Impacts of the Random Initialization in the Neural Tangent Kernel Theory
    Guhan Chen, Yicheng Li, and Qian Lin
    In 38th Conference on Neural Information Processing Systems (NeurIPS 2024), Sep 2024
    DOI OpenReview
  3. Improving Adaptivity via Over-Parameterization in Sequence Models
    Yicheng Li and Qian Lin
    In 38th Conference on Neural Information Processing Systems (NeurIPS 2024), Sep 2024
    DOI OpenReview
  4. On the Saturation Effects of Spectral Algorithms in Large Dimensions
    Weihao Lu, Haobo Zhang, Yicheng Li, and Qian Lin
    In 38th Conference on Neural Information Processing Systems (NeurIPS 2024), Sep 2024
    DOI OpenReview
  5. Generalization Error Curves for Analytic Spectral Algorithms under Power-Law Decay
    Yicheng Li, Weiye Gan, Zuoqiang Shi, and Qian Lin
    arXiv preprint, arXiv:2401.01599, Jan 2024
    arXiv
  6. On the Eigenvalue Decay Rates of a Class of Neural-Network Related Kernel Functions Defined on General Domains
    Yicheng Li, Zixiong Yu, Guhan Chen, and Qian Lin
    Journal of Machine Learning Research, 2024
    URL
  7. Kernel Interpolation Generalizes Poorly
    Yicheng Li, Haobo Zhang, and Qian Lin
    Biometrika, 2024
    DOI URL
  8. On the Optimality of Misspecified Spectral Algorithms
    Haobo Zhang, Yicheng Li (Co-first author), and Qian Lin
    Journal of Machine Learning Research, 2024
    URL

2023

  1. Optimal Rate of Kernel Regression in Large Dimensions
    Weihao Lu, Haobo Zhang, Yicheng Li, Manyun Xu, and Qian Lin
    arXiv preprint, arXiv:2309.04268, Sep 2023
    arXiv
  2. On the Saturation Effect of Kernel Ridge Regression
    Yicheng Li, Haobo Zhang, and Qian Lin
    In 11th International Conference on Learning Representations (ICLR 2023), Feb 2023
    arXiv OpenReview
  3. On the Asymptotic Learning Curves of Kernel Ridge Regression under Power-Law Decay
    Yicheng Li, Haobo Zhang, and Qian Lin
    In 37th Conference on Neural Information Processing Systems (NeurIPS 2023), 2023
    OpenReview
  4. On the Optimality of Misspecified Kernel Ridge Regression
    Haobo Zhang, Yicheng Li, Weihao Lu, and Qian Lin
    In 40th International Conference on Machine Learning (ICML 2023), 2023
    OpenReview