Hongji Wei
Home
Research
Teaching

Research Interests

  1. Statistical transfer learning theory and methodology.

    Human learners appear to have the talent to transfer knowledge from one task to another different but related task. However, in statistical learning, most procedures are developed to solve one single task, or to learn one single distribution based on observations from the same setting. In a wide range of real-world applications, it is important to gain improvement of learning in a new task through the transfer of knowledge from a related task that has already been learned.

  2. Distributed Learning under communication constraints, with possible adaptation and different types of protocols.

    In the conventional statistical decision theoretical framework, the focus is on the cen- tralized setting where all the data are collected together and directly available. However, as the modern data sets become huge and complicated, statistician gradually realizes that the classical optimal procedures sometimes cannot be applied to distributed learning settings where communication resources would become a bottleneck. My research focuses on understanding optimal and communication-efficient distributed learnin algorithms, as well as the trade-off between communication costs, statistical accuracy, adaptation, etc.

Papers and Preprints

  • Distributed Adaptive Gaussian Mean Estimation with Unknown Variance: Interactive Protocols Help Adaptation
    T. Tony Cai and Hongji Wei
    The Annals of Statistics, to appear, 2021+. [PDF]

  • Distributed Nonparametric Function Estimation: Optimal Rate of Convergence and Cost of Adaptation
    T. Tony Cai and Hongji Wei
    The Annals of Statistics, to appear, 2021+. [PDF]

  • Transfer Learning for Nonparametric Classification: Minimax Rate and Adaptive Classifier
    T. Tony Cai and Hongji Wei
    The Annals of Statistics, 49.1 (2021): 100-128. [PDF]

  • Inference of Evolutionary Modules by Low Rank Matrix Decomposition
    Hongji Wei and Jinzhu Jia
    Preprint, 2021.

  • Distributed Gaussian Mean Estimation under Communication Constraints: Optimal Fates and Communication-efficient Algorithms
    T. Tony Cai and Hongji Wei
    Preprint, 2019. [PDF]

Working Papers

  • Distributed Sparse Gaussian Sequence Detection and Classification: Phase Transition and Cost of Adaptation
    T. Tony Cai and Hongji Wei

  • Transfer Learning for Low Rank Matrices
    T. Tony Cai and Hongji Wei

  • Optimal Crowdsourcing with Low-rank Matrix Approximation
    T. Tony Cai and Hongji Wei