Achieving Optimal Complexity in Decentralized Learning over Row-Stochastic Networks

提交: 2024-09-25更新: 2024-10-31

TL;DR

We investigate lower bound and propose novel algorithms to match our lower bound in Row-Only decentralized learning.

关键词

decentralized stochastic optimizationdirected graphrow-stochastic matrixgradient tracking

评审与讨论

撤稿通知

2024-10-31

We realized some fundemantal problems in the proposed algorithm and this method will be improved in the future.