The paper introduces stabilized proximal decentralized optimization methods that achieve the best known communication and computation complexities, optimal up to log factors when local functions are sufficiently similar.
Title resolution pending
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
-
Exploiting Similarity for Computation and Communication-Efficient Decentralized Optimization
The paper introduces stabilized proximal decentralized optimization methods that achieve the best known communication and computation complexities, optimal up to log factors when local functions are sufficiently similar.