Asynchronous federated learning over TDMA channels is claimed to converge at rate O(G^2/√K) with device group count G, but the proof relies on a false inequality, and an intentional delay reduces staleness in experiments.
Parallel restarted SGD with faster con- vergence and less communication: Demystifying why model averaging works for deep learning,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Asynchronous Federated Learning Using Outdated Local Updates Over TDMA Channel
Asynchronous federated learning over TDMA channels is claimed to converge at rate O(G^2/√K) with device group count G, but the proof relies on a false inequality, and an intentional delay reduces staleness in experiments.