CED-EF: Compressed Exact Diffusion with Error Feedback for Multi-Agent Learning

2026-08-24Distributed, Parallel, and Cluster Computing

Distributed, Parallel, and Cluster Computing
AI summary

The authors propose a new method called CED-EF for decentralized optimization, where many agents work together to solve a problem while communicating compressed information to save bandwidth. Their method effectively handles biased compression and guarantees good convergence rates for smooth, nonconvex problems with noisy gradients. They also show improved efficiency under certain mathematical conditions compared to previous approaches. Experiments on common tasks like logistic regression demonstrate that CED-EF performs better in practice.

decentralized optimizationstochastic optimizationcompressed communicationerror feedbacknonconvex optimizationPolyak–Łojasiewicz conditionconvergence ratespectral gapδ-contractive compressorlogistic regression
Authors
Sulaiman A. Alghunaim, Kun Yuan
Abstract
We study decentralized stochastic optimization over a network of $N$ agents under compressed communication. We propose CED-EF, an exact diffusion-based method with error feedback that directly accommodates biased $δ$-contractive compressors while communicating one compressed model-sized vector per node per iteration. For smooth nonconvex objectives with unbiased stochastic gradients whose variance is bounded by $σ^2$, where $σ\geq0$, we establish a convergence rate whose leading stochastic term is $\mathcal O(σ/\sqrt{NK})$. For $σ>0$, the dominant dependence of the corresponding transient time on the number of agents, compression level, and spectral gap $Δ_λ$ is $\mathcal O(N^3/(δ^4Δ_λ^4))$, with fixed problem-dependent factors suppressed. Under the Polyak--Łojasiewicz condition, CED-EF attains a leading stochastic term $\widetilde{\mathcal O}(σ^2/(NK))$ with transient time on the order of $\widetilde{\mathcal O}(N/(δ^2Δ_λ^2))$. These dependencies improve the compression and/or network dependence of existing results. Numerical experiments on least-squares and logistic-regression problems illustrate the performance advantages of CED-EF.