arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.00343cs.DCcs.AIcs.PF
收藏

绿色联邦学习的标准化方法与建议

Standardized Methods and Recommendations for Green Federated Learning

Children’s National Hospital(儿童医院);NVIDIA(英伟达);Children’s National Hospital George Washington University(儿童医院乔治华盛顿大学)

Austin Tapp, Holger R. Roth, Ziyue Xu, Abhijeet Parida, Hareem Nisar, Marius George Linguraru

AI总结:

提出基于NVFlare和CodeCarbon的联邦学习碳核算方法,通过实验验证系统慢速和协调效应可显著增加碳排放,强调标准化碳核算对可复现绿色FL评估的必要性。

摘要:

联邦学习(FL)能够在隐私敏感的分布式数据上进行协作模型训练,但由于不一致的测量边界和异构的报告方式,其环境影响难以跨研究进行比较。我们提出了一种实用的碳核算方法,用于FL的CO2e跟踪,使用NVIDIA NVFlare和CodeCarbon进行显式的、阶段感知的任务(初始化、每轮训练、评估和空闲/协调)。为了捕捉非计算效应,我们还根据网络可配置能量模型,从传输的模型更新大小估计通信排放。我们在两个代表性工作负载上验证了所提出的方法:CIFAR-10图像分类和视网膜视盘分割。在CIFAR-10中,受控的客户端效率场景表明,在原本固定的FL协议下,系统级慢速和协调效应可能对碳足迹产生显著影响,相对于高效率基线,总CO2e增加了8.34倍(中等)和21.73倍(低)。在视网膜分割中,交换GPU层级(H100 vs. V100)产生了1.7倍的运行时间差距(290 vs. 503分钟),同时在不同站点间总能量和CO2e的变化不均匀,强调了按站点和按轮报告的必要性。总体而言,我们的结果支持一种标准化的碳核算方法,作为可复现的“绿色”FL评估的前提。我们的代码可在以下网址获取:https://this https URL。

英文摘要:

Federated learning (FL) enables collaborative model training over privacy-sensitive, distributed data, but its environmental impact is difficult to compare across studies due to inconsistent measurement boundaries and heterogeneous reporting. We present a practical carbon-accounting methodology for FL CO2e tracking using NVIDIA NVFlare and CodeCarbon for explicit, phase-aware tasks (initialization, per-round training, evaluation, and idle/coordination). To capture non-compute effects, we additionally estimate communication emissions from transmitted model-update sizes under a network-configurable energy model. We validate the proposed approach on two representative workloads: CIFAR-10 image classification and retinal optic disk segmentation. In CIFAR-10, controlled client-efficiency scenarios show that system-level slowdowns and coordination effects can contribute meaningfully to carbon footprint under an otherwise fixed FL protocol, increasing total CO2e by 8.34x (medium) and 21.73x (low) relative to the high-efficiency baseline. In retinal segmentation, swapping GPU tiers (H100 vs.\ V100) yields a consistent 1.7x runtime gap (290 vs. 503 minutes) while producing non-uniform changes in total energy and CO2e across sites, underscoring the need for per-site and per-round reporting. Overall, our results support a standardized carbon accounting method that acts as a prerequisite for reproducible 'green' FL evaluation. Our code is available at https://github.com/Pediatric-Accelerated-Intelligence-Lab/carbon_footprint.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑