发表机构
Durham University; University of Cambridge（杜伦大学; 剑桥大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究揭示去中心化学习扩展受数据分配、拓扑混合和通信容量的耦合影响，环图因谱间隙衰减导致高分歧，而自适应拓扑如LFHE保持共识并提升精度，但增加度数阈值会提高通信成本。
AI 中文摘要
扩展去中心化学习不仅改变了客户端数量N，也改变了信息传播和共识的动力学。我们认为，不能孤立地理解增加N所带来的影响，因为数据分配、依赖拓扑的混合以及通信容量可能同时发生变化。我们在CIFAR-10上研究了这些耦合效应，其中N∈{10,50,100,200}，比较了二阶环图、静态随机图和局部优先启发式演化（LFHE）——一种基于朋友之友发现的局部自适应拓扑过程。环图提供了一种分析上透明的失败模式：其Metropolis谱间隙按Θ(N^{-2})衰减，这意味着随着群体规模的增长，模型分歧的收缩速度逐渐变慢。实验表明，保持名义上的局部数据集大小固定，可以显著减少当固定总数据集被分配给更多客户端时观察到的表观人口惩罚。剩余的退化在很大程度上取决于通信结构：环图进入高分歧状态，而静态随机图和LFHE保持接近共识。提高LFHE的度数阈值进一步提高了准确性和共识，但代价是模型传输成本显著增加。这些结果表明，去中心化的扩展是由学习和通信动力学的耦合所决定的，而非仅由客户端数量决定。
英文摘要
Scaling decentralized learning changes not only the number of clients $N$, but also the dynamics of information propagation and consensus. We argue that the effect of increasing $N$ cannot be understood in isolation, because data allocation, topology-dependent mixing, and communication capacity may change simultaneously. We study these coupled effects on CIFAR-10 with $N\in\{10,50,100,200\}$, comparing a degree-two Ring, a Static Random graph, and Local-First Heuristic Evolution (LFHE), a locally adaptive topology process based on friend-of-friend discovery. The Ring provides an analytically transparent failure mode: its Metropolis spectral gap decays as $Θ(N^{-2})$, implying progressively slower contraction of model disagreement as the population grows. Experiments show that holding the nominal local dataset size fixed substantially reduces the apparent population penalty observed when a fixed total dataset is divided among more clients. The remaining degradation depends strongly on communication structure: Ring enters a high-disagreement regime, whereas Static Random and LFHE remain close to consensus. Increasing LFHE's degree threshold further improves accuracy and consensus, but at a substantially higher model-transmission cost. These results show that decentralized scaling is governed by coupled learning and communication dynamics, rather than by the number of clients alone.
Comments11 pages, 6 figures. Accepted as a poster at DynaFront @ NeurIPS 2026
