SEP 9, 2026 · PREPRINT
Cascading Gradient Inversion via LT-Code Inspired Peeling in Federated Learning
arXiv
A novel computational attack on federated learning that demonstrates theoretical vulnerability through algorithmic construction and benchmarking, but does not address clinical, patient, or real-world deployment outcomes.
Reported
Prior attack batch recovery rateabout half of a batch of size 100
Passive attacker recovery on Imag…94–100%
ImageNet batch size (passive atta…128