Jordanian Journal of Informatics and Computing

ISSN: 3080-6828 (Online)

Secure-Aggregation Federated Graph Attention Networks with SHAP-Based Edge Attribution for Advanced Persistent Threat Detection

by 

Ahmed Fakhri Abdullah ;

Ahmed Sedeeq Baker Al-Doori

PDF logoPDF

Published: 2026/11/09

Abstract

Advanced Persistent Threats (APTs) pose significant cybersecurity challenges due to their stealthy, prolonged activities and the limited visibility of malicious behaviour in host provenance data. Existing detection approaches often require centralized access to sensitive logs and provide limited interpretability, raising concerns about data confidentiality and analyst trust. To address these challenges, this study proposes a privacy-preserving and explainable federated learning framework for APT detection using provenance graphs. The proposed framework integrates edge-typed Graph Attention Networks (GATv2), self-supervised masked-feature learning, and federated prototype-based anomaly detection to identify suspicious activities without requiring labeled attack data during training. To preserve confidentiality, a dropout-resilient, Bonawitz-style secure aggregation protocol is employed to protect model updates and shared statistics, including normalization parameters, clustering information, and calibration distributions. Furthermore, edge-level Shapley value attribution is incorporated to provide interpretable explanations of detected anomalies and support security analysts during incident investigation. The framework is evaluated using the DARPA Transparent Computing Engagement 3 benchmark, specifically the THEIA and CADETS datasets, under a simulated four-client federated setting. Experimental results across three random seeds demonstrate strong detection performance, achieving Area Under the Receiver Operating Characteristic Curve (AUROC) values of 0.995 and 0.971 and Area Under the Precision–Recall Curve (AUPRC) values of 0.870 and 0.561 for THEIA and CADETS, respectively. The proposed secure aggregation mechanism maintains detection performance comparable to conventional federated averaging, introduces approximately 0.1% additional training time on THEIA, and reduces cross-round client-linkage attack accuracy from 85.6% to chance-level performance. Behaviour-level analysis demonstrates complete detection of scan-like attack activities at a calibrated 95% threshold, while other malicious activities exhibit threshold-dependent recall. The explainability analysis confirms the completeness of Shapley attributions and convergence of the highest-ranked contributing edges within 25 permutations, although their empirical fidelity is comparable to less computationally expensive explanation methods. Overall, the findings demonstrate the potential of combining secure federated learning, provenance-graph representation learning, and explainable AI for confidentiality-preserving APT detection, while highlighting important limitations related to threshold sensitivity, explanation fidelity, and semi-honest security assumptions.

Keywords

Advanced Persistent Threats (APTs)Federated LearningProvenance GraphsGraph Attention Networks (GATv2)Secure AggregationAnomaly Detectionand Cybersecurity.

Secure-Aggregation Federated Graph Attention Networks with SHAP-Based Edge Attribution for Advanced Persistent Threat Detection is licensed under CC BY 4.0CCBY

References

  1. Fotouhi, F., Balu, A., Jiang, Z., Esfandiari, Y., Jahani, S., & Sarkar, S. (2024). Dominating Set Model Aggregation for communication-efficient decentralized deep learning. Neural Networks, 171, 25–39. https://doi.org/10.1016/j.neunet.2023.11.057.
  2. Zhang, Y., Li, X., Luo, Q., Wang, Y., & Shen, Y. (2023). Practical and Efficient Secure Aggregation for Privacy-Preserving Machine Learning. Proceedings of the 2023 Asia Conference on Artificial Intelligence, Machine Learning and Robotics, 1–8. https://doi.org/10.1145/3625343.3625344
  3. Kairouz, P., & McMahan, H. B. (2021). Advances and Open Problems in Federated Learning. Foundations and Trends® in Machine Learning, 14(1-2), 1–210. https://doi.org/10.1561/2200000083
  4. Kannan, Y. (2024). Federated Learning in Cybersecurity: Applications, Challenges, and Future Directions. International Journal of Science and Research (IJSR), 13(7), 617–622. https://doi.org/10.21275/mr24706174710
  5. Romhányi, Z. A., & Kummer, A. (2026). Bayesian models in Federated Learning architectures. Machine Learning with Applications, 25, 100949. https://doi.org/10.1016/j.mlwa.2026.100949
  6. Zhao, S., & Keshtkar, S. (2026). Growing Domain-Specific LLMs Through Federated Split-Phase Learning: System Design and Analysis. 2026 2nd International Conference on Federated Learning and Intelligent Computing Systems (FLICS), 346–355. https://doi.org/10.1109/flics70075.2026.11621959
  7. So, J., Guler, B., & Avestimehr, A. S. (2021). Turbo-Aggregate: Breaking the Quadratic Aggregation Barrier in Secure Federated Learning. IEEE Journal on Selected Areas in Information Theory, 2(1), 479–489. https://doi.org/10.1109/jsait.2021.3054610
  8. Girgis, A. M., Data, D., & Diggavi, S. (2021). Differentially Private Federated Learning with Shuffling and Client Self-Sampling. 2021 IEEE International Symposium on Information Theory (ISIT), 338–343. https://doi.org/10.1109/isit45174.2021.9517906
  9. Dwork, C., & Roth, A. (2014). The Algorithmic Foundations of Differential Privacy. Foundations and Trends® in Theoretical Computer Science, 9(3-4), 211–487. https://doi.org/10.1561/0400000042.
  10. Moriai, S. (2019). Privacy-Preserving Deep Learning via Additively Homomorphic Encryption. 2019 IEEE 26th Symposium on Computer Arithmetic (ARITH), 198. https://doi.org/10.1109/arith.2019.00047
  11. Bidargaddi, A. P., D, D. A., & Prasanna, S. R. M. (2025). Enhancing Speaker Representations via Graph Attention Networks and Contrastive Learning. 2025 IEEE Conference on Engineering Informatics (ICEI), 1–8. https://doi.org/10.1109/icei68356.2025.11603690
  12. Wang, J., & Deng, Z. (2021). A Deep Graph Wavelet Convolutional Neural Network for Semi-supervised Node Classification. 2021 International Joint Conference on Neural Networks (IJCNN), 1–8. https://doi.org/10.1109/ijcnn52387.2021.9533634
  13. Huang, Q., Ren, H., Chen, P., Kržmanc, G., Zeng, D., Liang, P., & Leskovec, J. (2023). PRODIGY: Enabling In-context Learning Over Graphs. Advances in Neural Information Processing Systems 36, 16302–16317. https://doi.org/10.52202/075280-0718
  14. Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., & Yu, P. S. (2021). A Comprehensive Survey on Graph Neural Networks. IEEE Transactions on Neural Networks and Learning Systems, 32(1), 4–24. https://doi.org/10.1109/tnnls.2020.2978386
  15. Zhou, J., Cui, G., Hu, S., Zhang, Z., Yang, C., Liu, Z., Wang, L., Li, C., & Sun, M. (2020). Graph neural networks: A review of methods and applications. AI Open, 1, 57–81. https://doi.org/10.1016/j.aiopen.2021.01.001
  16. Kairouz, P., & McMahan, H. B. (2021). Advances and Open Problems in Federated Learning. Foundations and Trends® in Machine Learning, 14(1-2), 1–210. https://doi.org/10.1561/2200000083
  17. Ying, R., He, R., Chen, K., Eksombatchai, P., Hamilton, W. L., & Leskovec, J. (2018). Graph Convolutional Neural Networks for Web-Scale Recommender Systems. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining, 974–983. https://doi.org/10.1145/3219819.3219890
  18. Huang, S., Poursafaei, F., Danovitch, J., Fey, M., Hu, W., Rossi, E., Leskovec, J., Bronstein, M., Rabusseau, G., & Rabbany, R. (2023). Temporal Graph Benchmark for Machine Learning on Temporal Graphs. Advances in Neural Information Processing Systems 36, 2056–2073. https://doi.org/10.52202/075280-0099
  19. Al-shareeda, M., Musa, H. A., Jaafar, A., Adel Salman, A., Tami, Z. J., Hameed, H. M., Muaid, A., Hasan, H. A. H., & Bashkh, N. S. (2026). Design and Implementation of a Speech-to-Sign Robotic Arm for Deaf Communication. International Journal of Cybersecurity Engineering and Innovation, 2026(1), 25. https://doi.org/10.68281/ijcei.theitap.2026.1.3
  20. Alqutaish, A., & Alhumaid, K. (2026). A Federated Learning Framework for Privacy-Preserving Cyber Threat Detection. Journal of Information and Cybersecurity, 2026(1), 32. https://doi.org/10.68281/jics.theitap.2026.1.3
  21. Günnemann, S. (2022). Graph Neural Networks: Adversarial Robustness. Graph Neural Networks: Foundations, Frontiers, and Applications, 149–176. https://doi.org/10.1007/978-981-16-6054-2_8
  22. Schmalhofer, F. (2019). Unified Model of Knowledge Acquisition and Experimental Predictions. Constructive Knowledge Acquisition, 107–130. https://doi.org/10.4324/9781315789439-7
  23. Numeroso, D., & Bacciu, D. (2021). MEG: Generating Molecular Counterfactual Explanations for Deep Graph Networks. 2021 International Joint Conference on Neural Networks (IJCNN), 1–8. https://doi.org/10.1109/ijcnn52387.2021.9534266
  24. Li, J., Zhang, C., & Zhang, C. (2023). Heterogeneous Temporal Graph Neural Network Explainer. Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, 1298–1307. https://doi.org/10.1145/3583780.3614909
  25. Liu, L., Wan, P., Zhang, F., Zhang, Y., Liu, Q., & Wang, G. (2025). Learning model-level explanations of graph neural networks via subgraph order embedding space. Neural Networks, 191, 107815. https://doi.org/10.1016/j.neunet.2025.107815
  26. Duval, A., & Malliaros, F. D. (2021). GraphSVX: Shapley Value Explanations for Graph Neural Networks. Machine Learning and Knowledge Discovery in Databases. Research Track, 302–318. https://doi.org/10.1007/978-3-030-86520-7_19
  27. Mastropietro, A., Pasculli, G., Feldmann, C., Rodríguez-Pérez, R., & Bajorath, J. (2022). EdgeSHAPer: Bond-centric Shapley value-based explanation method for graph neural networks. iScience, 25(10), 105043. https://doi.org/10.1016/j.isci.2022.105043
  28. Pasquini, D., Francati, D., & Ateniese, G. (2022). Eluding Secure Aggregation in Federated Learning via Model Inconsistency. Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security, 2429–2443. https://doi.org/10.1145/3548606.3560557
  29. Garg, S., & Torra, V. (2026). PrunePrivyTune: Accelerating Language Models with Pruning and Differentially Private Fine-Tuning. Machine Learning, 115(5). https://doi.org/10.1007/s10994-026-07016-y
  30. A. Goyal, X. Han, G. Wang, and A. Bates, "Sometimes, you aren't what you do: Mimicry attacks against provenance graph host intrusion detection systems," in Proc. NDSS, 2023.
  31. Wu, J., Morikawa, T., & Mori, T. (2026). Evading Attention-based Dynamic Malware Detectors: An RL-Guided Method for Adversarial API Parameter Injection. Proceedings of the 41st ACM/SIGAPP Symposium on Applied Computing, 1810–1819. https://doi.org/10.1145/3748522.3779810
  32. Zipperle, M., Gottwalt, F., Chang, E., & Dillon, T. (2022). Provenance-based Intrusion Detection Systems: A Survey. ACM Computing Surveys, 55(7), 1–36. https://doi.org/10.1145/3539605
  33. Inam, M. A., Chen, Y., Goyal, A., Liu, J., Mink, J., Michael, N., Gaur, S., Bates, A., & Hassan, W. U. (2023). SoK: History is a Vast Early Warning System: Auditing the Provenance of System Intrusions. 2023 IEEE Symposium on Security and Privacy (SP), 2620–2638. https://doi.org/10.1109/sp46215.2023.10179405
  34. (Ed.). (2019). Hybrid Graph Convolutional Networks for Semi-Supervised Classification. Proceedings of 2019 the 9th International Workshop on Computer Science and Engineering. https://doi.org/10.18178/wcse.2019.06.016
  35. Zhou, W.-G., & Abbas, K. (2025). Unsupervised Inductive Node Representation Learning for Dynamic Graphs. IEEE Access, 13, 55034–55048. https://doi.org/10.1109/access.2025.3553377
  36. Dhami, D. S., Mathur, S., Yan, S., & Natarajan, S. (2025). Representing Multi-Relational Data via Statistical Relational Learning and Graph Convolutional Networks. Proceedings of the 13th ACM IKDD International Conference on Data Science, 153–163. https://doi.org/10.1145/3799830.3799847
  37. Kadhim, A. F., Hamzah, A. E., Al-Shareeda, M. A., Hussein, A. I., & Sapiee, N. M. (2026). Accurate Network Intrusion Detection using a Feedforward Neural Network and Bee Colony Optimization Algorithm. International Journal of Cybersecurity Engineering and Innovation, 2026(1), 49. https://doi.org/10.68281/ijcei.theitap.2026.1.7
  38. Lan, K., Li, G., Huang, W., Li, J., & Yu, Y. (2026). LLM-HGAN: LLM-Enhanced Heterogeneous Graph Attention Networks for Advanced Persistent Threat Detection. IEEE Transactions on Network Science and Engineering, 13, 8892–8910. https://doi.org/10.1109/tnse.2026.3687554
  39. Albahar, M. A. (2026). A Hybrid Self-Supervised Learning Framework for Advanced Persistent Threat Detection. Computers, Materials &Amp; Continua, 0(0), 1–10. https://doi.org/10.32604/cmc.2026.079941
  40. Krishnapriya, S., Singh, S., & Pramodkumar, A. (2026). FedDTE-IDS: A Federated Dual-Model Deep Transfer Ensemble Framework for Privacy-Preserving Advanced Persistent Threat Detection in Heterogeneous Industrial Internet of Things Networks. IEEE Transactions on Industrial Cyber-Physical Systems, 1–9. https://doi.org/10.1109/ticps.2026.3746211
  41. Almaiah, M. A., Khraisat, A., & Mamodiya, U. (2026). A Hybrid Post-Quantum Cryptographic Framework for Secure 5G/6G and Massive IoT Networks. Journal of Information and Cybersecurity, 2026(1), 45. https://doi.org/10.68281/jics.theitap.2026.1.4
  42. Zhang, K., Sun, L., Ding, B., Yiu, S. M., & Yang, C. (2024). Deep Efficient Private Neighbor Generation for Subgraph Federated Learning. Proceedings of the 2024 SIAM International Conference on Data Mining (SDM), 806–814. https://doi.org/10.1137/1.9781611978032.92
  43. Wu, C., Wu, F., Lyu, L., Qi, T., Huang, Y., & Xie, X. (2022). A federated graph neural network framework for privacy-preserving personalization. Nature Communications, 13(1). https://doi.org/10.1038/s41467-022-30714-9
  44. Nguyen, T. D., Marchal, S., Miettinen, M., Fereidooni, H., Asokan, N., & Sadeghi, A.-R. (2019). DÏoT: A Federated Self-learning Anomaly Detection System for IoT. 2019 IEEE 39th International Conference on Distributed Computing Systems (ICDCS), 756–767. https://doi.org/10.1109/icdcs.2019.00080
  45. Liu, Q., & Feng, Y. (2026). Federated learning and edge computing-based collaborative detection system for IoT anomaly behavior. Discover Artificial Intelligence, 6(1). https://doi.org/10.1007/s44163-026-02041-1