arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2501.01293cs.LGcs.AIcs.DCcs.NI
收藏

LEO-Split:一种面向低轨卫星网络的半监督分割学习框架

LEO-Split: A Semi-Supervised Split Learning Framework over LEO Satellite Networks

Institute of Space Internet, Fudan University(复旦大学空间互联网研究所);School of Computer Science, Fudan University(复旦大学计算机科学学院);Department of Electrical and Electronic Engineering, University of Hong Kong(香港大学电机与电子工程学系);Department of Computer Science, City University of Hong Kong(香港城市大学计算机科学系);School of Computer Science and Engineering, Nanyang Technological University(南洋理工大学计算机科学与工程学院)

Zheng Lin, Yuxin Zhang, Zhe Chen, Zihan Fang, Cong Wu, Xianhao Chen, Yue Gao, Jun Luo

AI总结:

本文提出LEO-Split半监督分割学习框架,通过辅助模型、伪标签算法和自适应激活插值,解决低轨卫星网络中数据稀缺、间歇连接和数据异构问题,实验表明其性能优于现有基准。

摘要:

近年来,低轨(LEO)卫星系统的日益部署使各种空间分析(例如农作物和气候监测)成为可能,这些分析高度依赖深度学习(DL)的进步。然而,LEO卫星与地面站(GS)之间的间歇性连接严重阻碍了将原始数据及时传输到地面站进行集中式学习,而规模不断扩大的深度学习模型也妨碍了在资源受限的LEO卫星上进行分布式学习。尽管分割学习(SL)通过划分模型并将主要训练负载卸载到地面站,可能是解决这些问题的一种潜在方案,但劳动密集型的标注过程仍然是一个障碍,同时间歇性连接和数据异构性也是其他挑战。在本文中,我们提出了LEO-Split,这是一种专为卫星网络量身定制的半监督(SS)分割学习设计,以应对这些挑战。利用半监督学习来处理(有标签)数据稀缺问题,我们构建了一个辅助模型来解决卫星与地面站非接触时段的训练失败问题。此外,我们提出了一种伪标签算法来纠正卫星之间的数据不平衡。最后,设计了一种自适应激活插值方案,以防止地面站服务器端子模型训练的过拟合。基于真实LEO卫星轨迹(例如Starlink)的大量实验表明,与最先进的基准方法相比,我们的LEO-Split框架取得了更优越的性能。

英文摘要:

Recently, the increasing deployment of LEO satellite systems has enabled various space analytics (e.g., crop and climate monitoring), which heavily relies on the advancements in deep learning (DL). However, the intermittent connectivity between LEO satellites and ground station (GS) significantly hinders the timely transmission of raw data to GS for centralized learning, while the scaled-up DL models hamper distributed learning on resource-constrained LEO satellites. Though split learning (SL) can be a potential solution to these problems by partitioning a model and offloading primary training workload to GS, the labor-intensive labeling process remains an obstacle, with intermittent connectivity and data heterogeneity being other challenges. In this paper, we propose LEO-Split, a semi-supervised (SS) SL design tailored for satellite networks to combat these challenges. Leveraging SS learning to handle (labeled) data scarcity, we construct an auxiliary model to tackle the training failure of the satellite-GS non-contact time. Moreover, we propose a pseudo-labeling algorithm to rectify data imbalances across satellites. Lastly, an adaptive activation interpolation scheme is devised to prevent the overfitting of server-side sub-model training at GS. Extensive experiments with real-world LEO satellite traces (e.g., Starlink) demonstrate that our LEO-Split framework achieves superior performance compared to state-ofthe-art benchmarks.

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。

↑