arXiv：cs.LG· Sarah Shitrit, Ilai Bistritz·· 3 小时前AI 评分34
Cordial Learning：面向相关数据的分布式训练新方法
Cordial Learning: Distributed Training with Correlated Data
AI 导读
研究者提出 cordial（correlated and distributed）learning，通过智能体间仅共享低维输出、训练本地模型从同伴提取信息，解决联邦学习等方法在相关数据上表现不佳的问题。在线性模型假设下，该方法被证明以概率 1 收敛到全局最优解，尽管全局目标非凸。在结构化多位 MNIST 任务上，cordial learning 即使在高非线性场景下仍保持高效。
正文
Abstract:We consider a distributed learning task with agents that have correlated data. Specifically, the label of an agent depends on the input of other agents for the same sample, and these inputs are also correlated. Correlated data is the reality when agents share the same environment. Existing decentralized methods, such as federated learning, ignore the structure of the problem and perform poorly on correlated data. On the other hand, centralized approaches are infeasible due to privacy and communication constraints. We introduce cordial (correlated and distributed) learning to address this gap by sharing only low-dimensional outputs between the agents while training local models to extract informative signals from peers. This distributed learning induces a game in which the loss function of each agent depends on the models of others. Assuming a linear model, we prove that cordial learning converges with probability one to a globally optimal solution, despite the nonconvex global objective. Experiments on structured multi-digit MNIST tasks demonstrate that cordial learning remains highly effective even in highly nonlinear settings.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.03330 [cs.LG] |
| (or arXiv:2610.03330v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.03330 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ilai Bistritz [view email]
[v1]
Fri, 2 Oct 2026 14:03:06 UTC (1,050 KB)
来源：arXiv：cs.LG · arxiv.org