My main research interests lie in privacy-preserving machine learning (PPML), Bayesian Learning, Uncertainty estimation and their applications in medical fields, such as medical image analysis, drug discovery and EHR data analysis.
In particular, I'm interested in developing theoretical tools for explainable privacy-preserving machine learning.
I also aim to combine different techniques such as differential privacy, federated learning, and Bayesian learning for
developing PPML algorithms which have better trade-offs among model accuracy, privacy protection strength, computation efficiency and
On this basis, I apply these methods and tools for rich applications such as pathological image classification, federated GNN,
recommender system and privacy-preserving Deep Learning as a Service.