Task-grained Federated Continual Learning Method for Maritime Defense Data

doi:10.19678/j.issn.1000-3428.0252615

Abstract

Abstract: With the rapid informatization of the maritime defense, federated learning becomes crucial for secure distributed data sharing. However, maritime defense sites face two challenges: 1) significant discrepancies in local data and tasks leading to negative knowledge transfer, and 2) difficulties in sharing knowledge due to limited communication bandwidth (remote locations of maritime defense sites) when continuously updating local model. Existing federated continual learning (FCL) methods primarily focus on extracting and aggregating the latest local model, yet lack in-depth research at the task level. This paper proposes FedTask, a task-grained FCL method, designed to adapt to multimodal maritime defense data (images, text, tabular data) while supporting task-level knowledge extraction, compact weight compression for narrow bandwidth. Through identification of similar task pairs among multiple clients (coastal defense sites), FedTask also enables positive knowledge transfer in both centralized and decentralized scenarios. Meanwhile, a unified architecture for federated continual learning is established. The system provides an integrated data processing pipeline and a scalable algorithm container, supporting low-code collaborative training across distributed nodes. Evaluations on real maritime defense datasets demonstrate that, compared to state-of-the-art 9 FCL methods, FedTask improves accuracy by 23% and 5% in centralized and decentralized scenarios, and reduces communication overhead by 75% and 85% under equivalent training time. The code is available at: https://github.com/LINC-BIT/FCLOnMDefenseData.

摘要： 随着海防部队信息化加速，联邦学习成为分布式数据安全共享的重要手段。然而，海防站点（如前线/后勤部队、设备中心）面临任务差异大导致知识负迁移，和本地模型持续更新但受限于通信带宽（边海防地处偏远）难以共享知识两大难题。现有联邦持续学习方法主要聚焦模型参数与数据特征，缺乏对任务层面的持续学习研究。本文提出任务粒度联邦持续学习方法FedTask，适配图像、文本、表格等多模态海防数据，支持任务粒度的模型知识提取，面向通信窄带宽的限制实现紧凑权重压缩。在中心化/去中心化场景下，通过对相似任务对的识别和聚合，FedTask实现了分布式客户端（海防站点）之间的知识正迁移。同时，本文构建了联邦持续学习系统架构，该系统提供了一体化的数据处理流程和可扩展的算法容器，并支持分布式节点的低代码化协同训练。基于真实的海防数据，本文对比了最新的9种联邦持续方法，测试结果表明，在同等训练时间下，FedTask分别提升中心化和去中心化联邦学习任务平均准确率14.78%和9.78%，并减少75%和85%的通信时间。

Xiaoning Wu, Siqi Du, Yaxin Luopan, Rui Han, Ke Qiu, Haiting Hou, Zhen Chen, Yu Zhao, Shuo Wang. Task-grained Federated Continual Learning Method for Maritime Defense Data[J]. Computer Engineering, doi: 10.19678/j.issn.1000-3428.0252615.

吴晓宁, 杜思琪, 罗潘亚欣, 韩锐, 邱可, 侯海婷, 陈铮, 赵玉, 王硕. 面向海防数据的任务粒度联邦持续学习方法[J]. 计算机工程, doi: 10.19678/j.issn.1000-3428.0252615.

/ Recommend / Download Citations

URL: https://www.ecice06.com/EN/10.19678/j.issn.1000-3428.0252615

References

[1] 梁其椿,许锡锴.智慧边海防的架构与推进策略[J].兵工自动化,2024,43(09):43-45. LIANG Qichun, XU Xikai. Structure and promotion strategy of intelligent border and coastal defense[J]. Ordnance Industry Automation, 2024,43(09):43-45.
[2] ZHOU Zhi, CHEN Xu, LI En, et al. Edge intelligence: Paving the last mile of artificial intelligence with edge computing[J]. Proceedings of the IEEE, 2019, 107(8): 1738-1762.
[3] HUANG Yanhong, LIU Wen, LIN Yijing, et al. FLCSDet: Federated Learning-Driven Cross-Spatial Vessel Detection for Maritime Surveillance With Privacy Preservation[J]. IEEE Transactions on Intelligent Transportation Systems, 2024.
[4] YANG Xin, YU Hao, GAO Xin, et al. Federated continual learning via knowledge fusion: A survey[J]. IEEE. Transactions on Knowledge and Data Engineering, 2024.
[5] LUOPAN Yaxin, HAN Rui, ZHANG Qinglong, et al. Fedknow: Federated continual learning with signature task knowledge integration at edge[C]. ICDE’23. 2023: 341-354.
[6] ZUO Xiaojiang, LUOPAN Yaxin, HAN Rui, et al. FedViT: Federated continual learning of vision transformer at edge[J]. Future Generation Computer Systems, 2024, 154: 1-15.
[7] HUANG Yixing, BERT C, FISCHER S, et al. Continual learning for peer-to-peer federated learning: A study on automated brain metastasis identification[J]. arXiv preprint arXiv:2204.13591, 2022.
[8] YAO Xin, SUN Lifeng. Continual local training for better initialization of federated models[C]. ICIP’2020. 2020: 1736-1740.
[9] Contributors M. MMCV: OpenMMLab Computer Vision Foundation[Z]. https://github.com/openmmlab/mmcv. 2018 [10] WOLF T, DEBUT L, SANH V, et al. Transformers: State-of-the-art natural language processing[C]. EMNLP’20. Online: Association for Computational Linguistics, 2020: 38-45.
[11] 李燕,张诗雅,李建玉,等.海防部队数据系统建设研究[J].火力与指挥控制,2024,49(07):184-193. LI Yan, ZHANG Shiya, LI Jianyu, et al. Study on data system construction of coastal defense force[J]. Fire Control & Command Control, 2024,49(07):184-193.
[12] MCMAHAN B, MOORE E, RAMAGE D, et al. Communication-efficient learning of deep networks from decentralized data[C]. AIR’2017. 2017: 1273-1282.
[13] SHELLER M J, REINA G A, EDWARDS B, et al. Multi-institutional deep learning modeling without sharing patient data: A feasibility study on brain tumor segmentation[C]. MICCAI’18. 2019: 92-104.
[14] LIAO Yunming, XU Yang, XU Hongli, et al. Adaptive configuration for heterogeneous participants in decentralized federated learning[C]. INFOCOM’23. 2023: 1-10.
[15] MA Zhenguo, XU Yang, XU Hongli, et al. Like attracts like: Personalized federated learning in decentralized edge computing[J]. IEEE Transactions on Mobile Computing, 2022, 23(2): 1080-1096.
[16] SHOHAM N, AVIDOR T, KEREN A, et al. Overcoming forgetting in federated learning on non-iid data[J]. arXiv preprint arXiv:1910.07796, 2019.
[17] YAO Xin, SUN Lifeng. Continual local training for better initialization of federated models[C]. ICIP’2020. 2020: 1736-1740.
[18] ZHU Zhuangdi, HONG Junyuan, ZHOU Jiayu. Data-free knowledge distillation for heterogeneous federated learning[C]. International conference on machine learning. 2021: 12878-12889.
[19] ZHANG Zhouyangzi, GUO Bin, SUN Wen, et al. Cross-fcl: Toward a cross-edge federated continual learning framework in mobile edge computing systems[J]. IEEE Transactions on Mobile Computing, 2022, 23(1): 313-326.
[20] WANG Qiang, LIU Bingyan, LI Yawen. Traceable federated continual learning[C]. CVPR’24. 2024: 12872-12881.
[21] WEI Guoyizhe, LI xin. Knowledge lock: Overcoming catastrophic forgetting in federated learning[C]. Pacific Asia Conference on Knowledge Discovery and Data Mining. 2022: 601-612.
[22] LIU Quande, CHEN Cheng, QIN Jing, et al. Feddg: Federated domain generalization on medical image segmentation via episodic learning in continuous frequency space[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 1013-1023.
[23] RISSANEN J J. Fisher information and stochastic complexity[J]. IEEE Transactions on Information Theory, 1996, 42(1): 40-47.
[24] YU Sixing, QIAN Wei, JANNESARI A. Resource-aware federated learning using knowledge extraction and multi-model fusion[J]. arXiv preprint arXiv:2208.07978, 2022.
[25] LI Daliang, WANG Junpu. Fedmd: Heterogenous federated learning via model distillation[J]. arXiv preprint arXiv:1910.03581, 2019.
[26] ZHU Zhuangdi, HONG Junyuan, ZHOU Jiayu. Data-free knowledge distillation for heterogeneous federated learning[C]. International conference on machine learning. 2021: 12878-12889.
[27] ONOSZKO N, KARLSSON G, MOGREN O, et al. Decentralized federated learning of deep neural networks on non-iid data[J]. arXiv preprint arXiv:2107.08517, 2021.
[28] ZHANG Jie, CHEN Li, CHEN Xiaohui, et al. A Novel Hierarchically Decentralized Federated Learning Framework in 6G Wireless Networks[C]. INFOCOM’23. 2023: 1-6.
[29] CHEN Suo, XU Yang, XU Hongli, et al. Decentralized federated learning with intermediate results in mobile edge computing[J]. IEEE Transactions on Mobile Computing, 2022, 23(1): 341-358.

Please choose a citation manager

Content to export