I work on federated learning for medical imaging: how to simulate the heterogeneity of real federations, which architectures train well under it, and how to aggregate under it. AI Scientist at Ultromics and doctoral researcher at City St George's, University of London.
| Project | Paper | Code and models |
|---|---|---|
| FDS: Federated Dataset Simulation | NeurIPS 2026, Evaluations and Datasets Track (arXiv) | fds |
| ANFR: Adaptive Normalization-Free Feature Recalibration | TMLR 2025 (arXiv) | ANFR, models on Hugging Face |
| FedCLAM: Client Adaptive Momentum with Foreground Intensity Matching | MICCAI 2025 (arXiv) | FedCLAM |
| ARIA: architectures, initialization and aggregation in federated visual classification | ISBI 2024 (arXiv) | ARIA |
| FedChest: a four-client chest X-ray federation | used in ANFR | FedChest |