FedScalar achieves federated learning with constant scalar uploads via random vector inner products, proving O(d/sqrt(K)) convergence to stationary points for smooth non-convex losses while reducing variance with Rademacher vectors.
On a formula for the product-moment coefficient of any order of a normal frequency distribution in any number of variables,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FedScalar: Federated Learning with Scalar Communication for Bandwidth-Constrained Networks
FedScalar achieves federated learning with constant scalar uploads via random vector inner products, proving O(d/sqrt(K)) convergence to stationary points for smooth non-convex losses while reducing variance with Rademacher vectors.