UA-PDFL measures client data skew using model outputs on a fixed unit input and switches between layer-wise personalization and whole-model replacement, improving decentralized federated learning accuracy on non-IID data.
Dropout as a bayesian approximation: Insights and applications, in: deep learning workshop, ICML, p
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
UA-PDFL: A Personalized Approach for Decentralized Federated Learning
UA-PDFL measures client data skew using model outputs on a fixed unit input and switches between layer-wise personalization and whole-model replacement, improving decentralized federated learning accuracy on non-IID data.