SyncAI.news, a Varaisys broadcasting
SPADE-DFL: Communication-Efficient Decentralized Federated Learning via Derivative-Free Linearized ADMM
MW

Mengli Wei, Mengkai Zhu, Jiawen Chen, Wenwu Yu, Duxin Che

· 1 min read

ResearcharXiv cs.LG

SPADE-DFL: Communication-Efficient Decentralized Federated Learning via Derivative-Free Linearized ADMM

arXiv:2609.29446v1 Announce Type: new Abstract: Reducing communication in derivative-free decentralized learning requires controlling the disagreement accumulated over multiple local updates. This paper develops SPADE-DFL, a primal--dual method that allows the number of local function-value updates between neighbor exchanges to grow with the computation budget while preserving the nonprivate convergence order. For smooth nonconvex objectives under uniform query-moment bounds, the prescribed nonprivate schedule achieves a time-averaged stationarity and consensus bound of $\mathcal{O}(T^{-1/3})$ using only $\Theta(T^{2/3})$ communication rounds, where $T$ is the number of local updates per client. For private training, the accumulated data-dependent increment is isolated from the graph correction, allowing one protected state per client and round to generate all outgoing messages. We prove client-level differential privacy for the full interactive transcript and quantify the resulting optimization error over a finite horizon. Experiments on four classification tasks show that SPADE-DFL achieves higher mean test accuracy than existing decentralized learning methods.

Original source

This story was published by arXiv cs.LG and written by Mengli Wei, Mengkai Zhu, Jiawen Chen, Wenwu Yu, Duxin Che. SyncAI.news shows a preview; the complete article is on the publisher's site.

Read the full story on arxiv.org

Similar News