A set of three techniques (update-level decomposition, block-wise Kronecker factorization, and aggregation-aware decoupling) improves the accuracy of low-rank compressed federated learning at the same communication budget.
Federated dynamic sparse training: Computing less, communicating less, yet learning better
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Panaceas for Improving Low-Rank Decomposition in Communication-Efficient Federated Learning
A set of three techniques (update-level decomposition, block-wise Kronecker factorization, and aggregation-aware decoupling) improves the accuracy of low-rank compressed federated learning at the same communication budget.