2024 ICML ICML 2024

Achieving Lossless Gradient Sparsification via Mapping to Alternative Space in Federated Learning