Hu Yao's doctoral defense focused on the research topic "Privacy-Preserving Deep Learning for Medical Image Analysis under Data Scarcity." During his doctoral studies, Hu Yao was devoted to investigating and designing effective privacy-preserving deep learning frameworks to solve the data scarcity problem from various perspectives, including federated learning (FL), source-free transfer learning (SFTL), and non-sensitive images utilization, for building efficient computer-aided diagnosis (CAD) models. His innovative achievements include: (1) Proposing a federated multi-task learning framework to exploit cross-institutional information compensation and data diversity in related mental disorders without compromising data privacy. (2) Proposing the first SFTL framework for CAD of mental disorders by utilizing inter-domain knowledge and exploring the unlabeled source data without private data transfer. (3) Developing a nature image-assisted framework for limited X-ray classification by formulating the CAD model training as an SSDA problem between X-ray data and the natural images. (4) Developing a heterogeneous structured FL framework to further meet the need of personalized models by pioneering the use of natural images as a proxy dataset to facilitate knowledge transfer from diverse heterogeneous local models to the global model.

 

Hu Yao received the B.S. degree in mining engineering and the M.Sc. degree in control science and engineering from the China University of Mining and Technology, Xuzhou, China, in 2017 and 2020, respectively. His current research focus is on privacy-preserving domain adaptation, federated learning, and the applied deep learning for medical image analysis. He has published four papers in prestigious research journals, as the first or corresponding author, including IEEE Transactions on Pattern Analysis and Machine Intelligence (IF: 23.6), IEEE Transactions on Emerging Topics in Computational Intelligence (IF: 5.3), and IEEE Transactions on Biomedical Engineering (IF: 4.54). Additionally, Hu Yao has granted six national invention patents.

 

Dr. Hu will join The Hong Kong Polytechnic University as a postdoctoral fellow, where he will continue his research in few-shot learning, medical image analysis, and the application of medical deep learning. Wishing him to have a fruitful and fulfilling postdoc journey ahead!