Secure and Efficient Federated Learning for Multi-domain Data Scenarios
JIN Chunhua1, LI Lulu1, WANG Jiahao1, JI Ling1, LIU Xinying1, CHEN Liqing1,2, ZHANG Hao1, WENG Jian3
1. Faculty of Computer and Software Engineering, Huaiyin Institute of Technology, Huaian 223003; 2. Fujian Provincial Key Laboratory of Network Security and Cryp-tology, Fujian Normal University, Fuzhou 350007; 3. College of Information Science and Technology, Jinan University, Guangzhou 510632
Abstract:To tackle the challenges of poor generalization, catastrophic forgetting and privacy attacks that federated learning faces in multi-domain data training, a scheme for secure and efficient federated learning for multi-domain scenarios(SEFL-MDS) is proposed. In the local training phase, knowledge distillation technology is employed to prevent catastrophic forgetting during multi-domain data training, while accelerating knowledge transfer across domains to improve training efficiency. In the uploading phase, Gaussian noise is added to locally updated gradients and generalization differences across domains using the Gaussian differential privacy mechanism to ensure secure data uploads and enhance the confidentiality of the training process. In the aggregation phase, a dynamic generalization-weighted algorithm is utilized to reduce generalization differences across domains, thereby enhancing the generalization capability. Theoretical analysis demonstrates the high robustness of the proposed scheme. Experiments on PACS and office-Home datasets show that the proposed scheme achieves higher accuracy with reduced training time.
[1] CHEN K Y, CHEN B W, LIU C Y, et al. RSMamba: Remote Sensing Image Classification with State Space Model. IEEE Geosci-ence and Remote Sensing Letters, 2024, 21. DOI: 10.1109/LGRS.2024.3407111. [2] VIJAYAKUMAR A, VAIRAVASUNDARAM S.YOLO-Based Object Detection Models: A Review and Its Applications. Multimedia Tools and Applications, 2024, 83: 83535-83574. [3] SHAMSHIRI A, RYU K R, PARK J Y.Text Mining and Natural Language Processing in Construction. Automation in Construction, 2024, 158. DOI: 10.1016/j.autcon.2023.105200. [4] MCMAHAN B, MOORE E, RAMAGE D, et al. Communication-Efficient Learning of Deep Networks from Decentralized Data. Journal of Machine Learning Research, 2017, 54: 1273-1282. [5] KARIMIREDDY S P, KALE S, MOHRI M, et al. SCAFFOLD: Stochastic Controlled Averaging for Federated Learning. Journal of Machine Learning Research, 2020, 119: 5132-5143. [6] LI T, SAHU A K, ZAHEER M, et al. Federated Optimization in Heterogeneous Networks[C/OL].[2024-06-19]. http://arxiv.org/pdf/1812.06127v5. [7] DONG J H, WANG L X, FANG Z, et al. Federated Class-Incremental Learning//Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 10154-10163. [8] KHOSLA A, ZHOU T, MALISIEWICZ T, et al. Undoing the Damage of Dataset Bias//Proc of the European Conference on Computer Vision. Berlin, German: Springer. 2012: 158-171. [9] GHIFARY M, BALDUZZI D, KLEIJN W B, et al. Scatter Component Analysis: A Unified Framework for Domain Adaptation and Domain Generalization. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(7): 1414-1430. [10] SUN B C, SAENKO K.Deep CORAL: Correlation Alignment for Deep Domain Adaptation//Proc of the European Conference on Computer Vision. Berlin, German: Springer, 2016: 443-450 [11] HU H S, ZHANG X Y, SALCIC Z, et al. Source Inference Atta-cks: Beyond Membership Inference Attacks in Federated Learning. IEEE Transactions on Dependable and Secure Computing, 2024, 21(4): 3012-3029. [12] RAO B S, ZHANG J L, WU D, et al. Privacy Inference Attack and Defense in Centralized and Federated Learning: A Comprehensive Survey. IEEE Transactions on Artificial Intelligence, 2024. DOI: 10.1109/TAI.2024.3363670. [13] 管桂林,支婷,陶政坪,等.物联网中多密钥同态加密的联邦学习隐私保护方法.信息安全研究, 2024, 10(10): 958-966. (GUAN G Z, ZHI T, TAO Z P, et al. A Federated Learning Privacy Protection Method for Multi-key Homomorphic Encryption in the Internet of Things. Journal of International Security Research, 2024, 10(10): 958-966.) [14] 王莉芳,罗明星.基于用户级差分隐私的联邦学习方案研究[J/OL].[2024-06-19]. https://link.cnki.net/urlid/50.1181.n.20240715.1758.011. (WANG L F, LUO M X. Research on Federated Learning Scheme Based on User-Level Differential Privacy[J/OL].[2024-06-19]. https://link.cnki.net/urlid/50.1181.n.20240715.1758.011.) [15] ZHANG C, ZHOU B Y, HE Z Q, et al. OBLIVION: Poisoning Federated Learning by Inducing Catastrophic Forgetting//Proc of the IEEE Conference on Computer Communications. Washington, USA: IEEE, 2023. DOI: 10.1109/INFOCOM53939.2023.10228981. [16] LUO K Y, LI X, LAN Y S, et al. GradMA: A Gradient-Memory-Based Accelerated Federated Learning with Alleviated Catastrophic Forgetting//Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 3708-3717. [17] GAO Z P, ZHANG J Y, YU X L.Addressing Catastrophic Forge-tting in Federated Learning on Resource-Constrained Devices: A Feature Replay Approach//Proc of the 20th International Confe-rence on Intelligent Computing Technology and Applications. Berlin, German: Springer, 2024: 336-348. [18] DOU Q, CASTRO D C, KAMNITSAS K, et al. Domain Generalization via Model-Agnostic Learning of Semantic Features//Proc of the 33rd International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2019: 6450-6461. [19] HUANG J X, GUAN D Y, XIAO A R, et al. FSDR: Frequency Space Domain Randomization for Domain Generalization//Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 6887-6898. [20] WANG S J, YU L Q, LI C Z,et al. Learning from Extrinsic and Intrinsic Supervisions for Domain Generalization//Proc of the European Conference on Computer Vision. Berlin, German: Sprin-ger, 2020: 159-176. [21] CARLUCCI F M, D'INNOCENTE A, BUCCI S, et al. Domain Generalization by Solving Jigsaw Puzzles//Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2019: 2224-2233. [22] LIU Q D, CHEN C, QIN J, et al. FedDG: Federated Domain Generalization on Medical Image Segmentation via Episodic Lear-ning in Continuous Frequency Space//Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 1013-1023. [23] JIANG M R, WANG Z R, DOU Q.HarmoFL: Harmonizing Local and Global Drifts in Federated Learning on Heterogeneous Medical Images. Proceedings of the AAAI Conference on Artificial Intelligence, 2022, 36(1): 1087-1095. [24] WANG J Y, LIU Q H, LIANG H, et al. Tackling the Objective Inconsistency Problem in Heterogeneous Federated Optimization//Proc of the 34th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2020: 7611-7623. [25] LI S H, NGAI E, YE F H, et al. Auto-Weighted Robust Federated Learning with Corrupted Data Sources. ACM Transactions on Intelligent Systems and Technology, 2022, 13(5). DOI: 10.1145/3517821. [26] PARK J, HAN D J, KIM J, et al. StableFDG: Style and Attention Based Learning for Federated Domain Generalization[C/OL].[2024-06-19]. https://arxiv.org/pdf/2311.00227.00227. [27] ZHANG R P, XU Q W, YAO J C, et al. Federated Domain Ge-neralization with Generalization Adjustment//Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 3954-3963. [28] WEI Y K, HAN Y H.Multi-source Collaborative Gradient Discrepancy Minimization for Federated Domain Generalization. Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 38(14): 15805-15813. [29] PERKONIGG M, HOFMANNINGER J, HEROLD C J, et al. Dynamic Memory to Alleviate Catastrophic Forgetting in Continual Learning with Medical Imaging. Nature Communications, 2021, 12(1). DOI: 10.1038/s41467-021-25858-z. [30] TADROS T, KRISHNAN G P, RAMYAA R, et al. Sleep-Like Unsupervised Replay Reduces Catastrophic Forgetting in Artificial Neural Networks. Nature Communications, 2022, 13(1). DOI: 10.1038/s41467-022-34938-7. [31] LEE G, JEONG M, SHIN Y, et al. Preservation of the Global Know-ledge by Not-True Distillation in Federated Learning//Proc of the 36th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2022: 38461-38474. [32] MA Y H, XIE Z L, WANG J, et al. Continual Federated Learning Based on Knowledge Distillation//Proc of the 31st International Joint Conference on Artificial Intelligence. San Francisco, USA: IJCAI, 2022: 2182-2188. [33] ZHENG S W, HU J, MIN G Y, et al. Mutual Knowledge Disti-llation based Personalized Federated Learning for Smart Edge Computing. IEEE Transactions on Consumer Electronics, 2024. DOI: 10.1109/TCE.2024.3412817. [34] CHEN L J, XIAO D, YU Z Y, et al. Secure and Efficient Federated Learning via Novel Multi-party Computation and Compressed Sensing. Information Sciences, 2024, 667. DOI: 10.1016/j.ins.2024.120481. [35] SHEN C, ZHANG W, ZHOU T P, et al. A Security-Enhanced Federated Learning Scheme Based on Homomorphic Encryption and Secret Sharing. Mathematics, 2024, 12(13). DOI: 10.3390/math12131993. [36] BATOOL H, ANJUM A, KHAN A, et al. A Secure and Privacy Preserved Infrastructure for VANETs Based on Federated Learning with Local Differential Privacy. Information Sciences, 2024, 652. DOI: 10.1016/j.ins.2023.119717. [37] CHA J, CHUN S, LEE K, et al. SWAD: Domain Generalization by Seeking Flat Minima//Proc of the 35th International Confe-rence on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2021: 22405-22418. [38] XU Q W, ZHANG R P, ZHANG Y, et al. A Fourier-Based Frame-work for Domain Generalization//Proc of the IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 14378-14387. [39] HUANG Z Y, WANG H H, XING E P, et al. Self-Challenging Improves Cross-Domain Generalization//Proc of the 16th Euro-pean Conference on Computer Vision. Berlin, German: Springer, 2020: 124-140. [40] QU Z, LI X Y, DUAN R, et al. Generalized Federated Learning via Sharpness Aware Minimization. Journal of Machine Learning Research, 2022, 162: 18250-18280.