A federated learning flight operation data sharing algorithm for balancing privacy and utility
-
摘要:
数据安全与隐私保护的需要导致部分航班运行数据难以直接进行共享。联邦学习(FL)实现了数据可用不可见，但FL面对诚实但好奇的敌手时，仍面临反转攻击的威胁，存在隐私泄露的风险。针对FL在平衡隐私保护和模型效用方面的较差性能，提出基于人工蜂群和双Rényi差分隐私的联邦学习(ABC-2RDP-FL)航班运行数据数据共享算法。在ABC-2RDP-FL中，设计了双Rényi差分隐私(RDP)隐私保护机制，以更严格地测量隐私预算，提升隐私保护性能。提出基于人工蜂群(ABC)的FL超参数优化算法，提升模型性能，平衡隐私保护与模型效用。使用公共数据和航班运行数据验证了所提算法的有效性。
Abstract:The need for data security and privacy protection makes it difficult to directly share some flight operation data. Federated learning (FL) achieves data availability that is invisible, but when faced with honest but curious opponents, it still faces the threat of reverse attacks and the risk of privacy leakage. The inadequate performance of FL in striking a balance between privacy protection and model utility is addressed by the suggested artificial bee colony and dual Rényi differential privacy federated learning (ABC-2RDP-FL) flight operation data privacy protection algorithm, which is based on dual Rényi differential privacy (RDP) and artificial bee colony (ABC). In ABC-2RDP-FL, a dual RDP protection mechanism is designed to measure privacy budgets more strictly and improve privacy protection performance. After that, an ABC-based FL hyperparameter optimization approach is suggested to enhance model performance while striking a compromise between model utility and privacy protection. Finally, the effectiveness of the proposed method was validated using public data and flight operation data.
-
图 3 超参数优化前后算法性能对比
Figure 3. Comparison of algorithm performance before and after hyperparameter optimization
图 5 不同隐私预算下的算法性能对比
Figure 5. Comparison of algorithm performance under different privacy budgets
表 1 航班延误数据信息
Table 1. Flight delay data information
航班信息 环境信息 机型代码、起飞机场、计划起飞日期、实际起飞日期、落地日期、入位日期（飞机滑入停机位日期）、计划到达日期 起飞机场气温、起飞机场风向、起飞机场风速、起飞机场能见度、起飞机场云量等级、起飞机场云底高度、目的地机场气温、目的地机场风向、目的地机场风速、目的地机场能见度、目的地机场云量等级、目的地机场云底高度 
下载: 导出CSV
表 2 ABC算法搜索出的超参数值
Table 2. Hyperparameter values searched by ABC algorithm
超参数 算法 $ \eta $ $ H $ $ B $ $ Z $ 超参数经验值 0.01 3 128 0.0001 超参数搜索范围 0.00001 ～0.11～15 4～512 0.00001 ～1MNIST数据集超参数搜索结果 ABC-FL 0.0074 10 5 0.6623 ABC-2RDP-FL 0.0076 10 24 0.2492 CIFAR10数据集超参数搜索结果 ABC-FL 0.0048 3 135 0.8384 ABC-2RDP-FL 0.0043 9 56 0.8393 航班延误数据集超参数搜索结果 ABC-FL 0.0079 11 146 0.9677 ABC-2RDP-FL 0.0031 12 436 0.9137
下载: 导出CSV
表 3 超参数优化前后算法测试结果
Table 3. Algorithm test results before and after hyperparameter optimization
算法 测试精度/% 损失值 MNIST数据集 CIFAR10数据集 航班延误数据集 MNIST数据集 CIFAR10数据集 航班延误数据集 FL 97.74 73.27 91.05 0.0769 0.7762 0.2463 2RDP-FL 96.86 68.70 88.95 0.1028 0.9170 0.3185 ABC-FL 99.31 82.51 92.62 0.0249 0.6458 0.2054 ABC-2RDP-FL 99.06 77.16 91.46 0.0282 0.6890 0.2274
下载: 导出CSV
表 4 相同隐私预算下算法对比实验结果
Table 4. Comparative experiment results of algorithms under the same privacy budget
算法 测试精度/% 损失值 MNIST数据集
（$ \varepsilon =1.5 $）CIFAR10数据集
（$ \varepsilon =2 $）航班延误数据集
（$ \varepsilon =8 $）MNIST数据集
（$ \varepsilon =1.5 $）CIFAR10数据集
（$ \varepsilon =2 $）航班延误数据集
（$ \varepsilon =8 $）DP-FL 96.23 67.50 86.51 0.1210 0.9347 0.3385 LDP-FL 96.61 66.64 85.26 0.1136 0.9862 0.4212 NbAFL 96.44 67.23 86.89 0.1167 0.9682 0.3326 2RDP-FL 96.86 68.70 88.95 0.1028 0.9170 0.3185 ABC-2RDP-FL 99.06 77.16 91.46 0.0282 0.6889 0.2274
下载: 导出CSV
-
[1] Fu A M, Yu S, Zhang Y Q, et al. NPP: a new privacy-aware public auditing scheme for cloud data sharing with group users[J]. IEEE Transactions on Big Data, 2022, 8(1): 14-24. [2] Su Z, Wang Y T, Xu Q C, et al. LVBS: lightweight vehicular blockchain for secure data sharing in disaster rescue[J]. IEEE Transactions on Dependable and Secure Computing, 2022, 19(1): 19-32. [3] Mcmahan H B, Moore E, Ramage D, et al. Communication-efficient learning of deep networks from decentralized data[C]//Proceedings of the International Conference on Artificial Intelligence and Statistics. San Juan: PMLR, 2016: 1273-1282. [4] Chen Y Q, Qin X, Wang J D, et al. FedHealth: a federated transfer learning framework for wearable healthcare[J]. IEEE Intelligent Systems, 2020, 35(4): 83-93. [5] Tan L, Yu K P, Shi N, et al. Towards secure and privacy-preserving data sharing for COVID-19 medical records: a blockchain-empowered approach[J]. IEEE Transactions on Network Science and Engineering, 2022, 9(1): 271-281. [6] Xiao H Z, Zhao J, Pei Q Q, et al. Vehicle selection and resource optimization for federated learning in vehicular edge computing[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(8): 11073-11087. [7] Kong Q L, Yin F, Lu R X, et al. Privacy-preserving aggregation for federated learning-based navigation in vehicular fog[J]. IEEE Transactions on Industrial Informatics, 2021, 17(12): 8453-8463. [8] Lu Y L, Huang X H, Dai Y Y, et al. Blockchain and federated learning for privacy-preserved data sharing in industrial IoT[J]. IEEE Transactions on Industrial Informatics, 2020, 16(6): 4177-4186. [9] Ma X, Wen C L, Wen T. An asynchronous and real-time update paradigm of federated learning for fault diagnosis[J]. IEEE Transactions on Industrial Informatics, 2021, 17(12): 8531-8540. [10] Wang Y T, Su Z, Zhang N, et al. Learning in the air: secure federated learning for UAV-assisted crowdsensing[J]. IEEE Transactions on Network Science and Engineering, 2021, 8(2): 1055-1069. [11] Zhang J L, Chen B, Cheng X, et al. PoisonGAN: generative poisoning attacks against federated learning in edge computing systems[J]. IEEE Internet of Things Journal, 2021, 8(5): 3310-3322. [12] Li Z H, Yu H F, Zhou T Y, et al. Byzantine resistant secure blockchained federated learning at the edge[J]. IEEE Network, 2021, 35(4): 295-301. [13] Shokri R, Shmatikov V. Privacy-preserving deep learning[C]//Proceedings of the 2015 53rd Annual Allerton Conference on Communication, Control, and Computing. Piscataway: IEEE Press, 2016: 909-910. [14] Zhou C Y, Fu A M, Yu S, et al. Privacy-preserving federated learning in fog computing[J]. IEEE Internet of Things Journal, 2020, 7(11): 10782-10793. [15] Li Y R, Li H W, Xu G W, et al. Practical privacy-preserving federated learning in vehicular fog computing[J]. IEEE Transactions on Vehicular Technology, 2022, 71(5): 4692-4705. [16] Abadi M, Chu A, Goodfellow I, et al. Deep learning with differential privacy[C]//Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security. New York: ACM, 2016: 308-318. [17] Zhang L L, Shen B C, Barnawi A, et al. FedDPGAN: federated differentially private generative adversarial networks framework for the detection of COVID-19 pneumonia[J]. Information Systems Frontiers, 2021, 23(6): 1403-1415. [18] Xin B Z, Geng Y Y, Hu T, et al. Federated synthetic data generation with differential privacy[J]. Neurocomputing, 2022, 468: 1-10. [19] Chen S Z, Yu D X, Zou Y F, et al. Decentralized wireless federated learning with differential privacy[J]. IEEE Transactions on Industrial Informatics, 2022, 18(9): 6273-6282. [20] Ribero M, Henderson J, Williamson S, et al. Federating recommendations using differentially private prototypes[J]. Pattern Recognition, 2022, 129: 108746. [21] Hao M, Li H W, Luo X Z, et al. Efficient and privacy-enhanced federated learning for industrial artificial intelligence[J]. IEEE Transactions on Industrial Informatics, 2020, 16(10): 6532-6542. [22] Wei K, Li J, Ding M, et al. Federated learning with differential privacy: algorithms and performance analysis[J]. IEEE Transactions on Information Forensics and Security, 2020, 15: 3454-3469. [23] Lu Y L, Huang X H, Dai Y Y, et al. Differentially private asynchronous federated learning for mobile edge computing in urban informatics[J]. IEEE Transactions on Industrial Informatics, 2020, 16(3): 2134-2143. [24] Zhang T, Song A X, Dong X W, et al. Privacy-preserving asynchronous grouped federated learning for IoT[J]. IEEE Internet of Things Journal, 2022, 9(7): 5511-5523. [25] Deng W, Liu H L, Xu J J, et al. An improved quantum-inspired differential evolution algorithm for deep belief network[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69(10): 7319-7327. [26] Deng W, Xu J J, Gao X Z, et al. An enhanced MSIQDE algorithm with novel multiple strategies for global optimization problems[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2022, 52(3): 1578-1587. [27] Deng W, Xu J J, Zhao H M, et al. A novel gate resource allocation method using improved PSO-based QEA[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(3): 1737-1745. [28] Yamany W, Moustafa N, Turnbull B. OQFL: an optimized quantum-based federated learning framework for defending against adversarial attacks in intelligent transportation systems[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(1): 893-903. [29] Kishore P, Barisal S K, Mohapatra D P. Particle swarm optimized federated learning for securing IoT devices[C]//Proceedings of the 2021 IEEE Globecom Workshops. Piscataway: IEEE Press, 2021: 1-6. [30] Qolomany B, Ahmad K, Al-fuqaha A, et al. Particle swarm optimized federated learning for industrial IoT and smart city services[C]//Proceedings of the 2020 IEEE Global Communications Conference. Piscataway: IEEE Press, 2021: 1-6. [31] Karaboga D, Basturk B. On the performance of artificial bee colony (ABC) algorithm[J]. Applied Soft Computing, 2008, 8(1): 687-697. [32] Mironov I. Rényi differential privacy[C]//Proceedings of the 2017 IEEE 30th Computer Security Foundations Symposium. Piscataway: IEEE Press, 2017: 263-275. -
下载:
点击查看大图
计量
- 文章访问数: 453
- HTML全文浏览量: 167
- PDF下载量: 13
- 被引次数: 0
