Institute of Computer Vision, College of Computer Science and Technology, Zhejiang University of Technology
We ask why a model fails when the subject, the recording device or the recording session changes, and how to prevent that failure. Real observations carry three defects: missing data, misaligned channels and distribution shift. Instead of patching them with hand-tuned preprocessing that only fits the data it was tuned on, we build their handling into the model itself, so that the space in which the model represents its input — the latent space — does not change with acquisition conditions.
Current direction: robust EEG feature extraction under multimodal constraints — using microscopic neural recordings (calcium imaging, ECoG) during training to constrain the EEG latent space, so that brain–computer interfaces stay reliable across people and sessions while needing only EEG at inference.
Figures as of September 2026.
A machine-and-human reader study (Nature Communications, 2021, first author) used a generative adversarial network to change only the lesion region of mammograms....
Read more →In a feed-forward network the designer controls three things: the architecture, the activation function and the loss. We let each layer choose its own activation...
Read more →A multicentre study (eClinicalMedicine, 2025, co-corresponding author; master's student Chen Jin co-first author) predicts the response of breast cancer to...
Read more →EEG differs greatly between people, which keeps brain–computer interfaces from practical use. Ph.D. student Kunkun Zhang proposed a masked self-supervised contrastive...
Read more →
Four of our deep-learning models run on the lab GPU. Upload de-identified images and get the result in seconds; nothing you upload is kept.
We welcome students, researchers and partners who want to make AI models that keep working when the data change. Come to study with us, do research with us, or bring a problem from your clinic, lab or company.