A two-stage Local GD with learning-rate warmup achieves O(1/(K R)) convergence for heterogeneous distributed logistic regression, proving that local steps can provably reduce communication rounds.
Characterizing implicit bias in terms of optimization geometry
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
-
Local Steps Speed Up Local GD for Heterogeneous Distributed Logistic Regression
A two-stage Local GD with learning-rate warmup achieves O(1/(K R)) convergence for heterogeneous distributed logistic regression, proving that local steps can provably reduce communication rounds.