Regular biography
Fengxiang He is a Lecturer in AI Driven Business Informatics and Financial Com at The University of Edinburgh, Department of CS. He can be contacted via email at F.He@ed.ac.uk, and his profile page is available at https://www.ed.ac.uk/Fengxiang_He.html. His research interests and existing bio details are not provided in the available information.
Scholar-generated biography
Fengxiang He is a researcher at the University of Edinburgh with a focus on machine learning. His work explores theoretical and practical aspects of generalization in deep learning, including residual networks, federated learning, and few-shot learning. He investigates how models can generalize well by controlling batch size and learning rate, and he develops methods for communication-efficient personalized federated learning. His research also addresses challenges in image inpainting, video super-resolution, and domain adaptation. He emphasizes the role of semantic alignment and topology-aware generalization in improving model performance. His publications highlight the importance of understanding overfitting, noise memorization, and catastrophic forgetting in deep learning systems.