FedHPro learns hyper-prototypes via gradient matching and mutual-contrastive learning to achieve state-of-the-art performance in heterogeneous federated learning.
Referring to (Wang et al., 2024), we consider two non-iid scenarios on the AG News:
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FedHPro: Federated Hyper-Prototype Learning via Gradient Matching
FedHPro learns hyper-prototypes via gradient matching and mutual-contrastive learning to achieve state-of-the-art performance in heterogeneous federated learning.