Abstract

The increasing sophistication of adversarial attacks in machine learning systems requires advanced detection and mitigation strategies. Existing models fail to identify such attacks in a timely and accurate manner, mainly because of their inability to adapt to various conditions and lack of predictive capabilities. To address these shortcomings, this paper introduces an innovative pre-emption model based on the capabilities of Deep Dyna Q Learning combined with VARMAx Operations. The Deep Dyna Q Learning framework is very efficient for finding the salient performance indicators accuracy, confidence, training loss, prediction performance, anomaly occurrence, and explainability performance. An advanced GridCAM++ process tracks these performance indicators while providing insights regarding their interconnection, thus a basis for formulating a well-performing model process of prediction for adversarial attacks. Further enhancing the model's resilience, VARMAx Operations are employed to integrate both endogenous variables (derived from the system's performance metrics) and exogenous variables (representing noise or external factors), providing a comprehensive view for preempting adversarial attacks. This combination enhances predictive accuracy and also allows for preempting possible security breaches. The model uses GAN-based sample generation along with a digital twin framework that significantly increases classification performance by properly cross-comparing and eliminating attack vectors for different scenarios. The empirical tests performed on real-time networks show the superiority of this model over the existing methods. Notably, it achieved an increase of 8.5% precision in adversarial attack pre-emption, accuracy improved up to 8.3%, higher recall rate by up to 7.5%, reduced delay by 4.9%, and AUC rose by 10.4%, and specificity also enhanced by 9.4% in process.

Keywords

Deep Dyna Q Learning, VARMAx Operations, Adversarial Attack Pre-emption, GAN-Based Sample Generation, Scenarios,

Downloads

Download data is not yet available.

References

  1. A. Guesmi, M.A. Hanif, B. Ouni, M. Shafique, Physical Adversarial Attacks for Camera-Based Smart Systems: Current Trends, Categorization, Applications, Research Challenges, and Future Outlook. IEEE Access, 11, (2023) 109617-109668. https://doi.org/10.1109/ACCESS.2023.3321118
  2. R. Huang, Y. Li, Adversarial Attack Mitigation Strategy for Machine Learning-Based Network Attack Detection Model in Power System. IEEE Transactions on Smart Grid, 14(3), (2023) 2367-2376. https://doi.org/10.1109/TSG.2022.3217060
  3. W. Feng, Xu, N., Zhang, T., Wu, B., & Zhang, Y. Robust and generalized physical adversarial attacks via meta-GAN. IEEE Transactions on Information Forensics and Security, 19, (2023) 1112-1125. https://doi.org/10.1109/TIFS.2023.3288426
  4. S. He, R. Wang, T. Liu, C. Yi, X. Jin, R. Liu, W. Zhou, Type-I generative adversarial attack. IEEE Transactions on Dependable and Secure Computing, 20(3), (2022) 2593-2606. https://doi.org/10.1109/TDSC.2022.3186918
  5. S. Zhao, W. Wang, Z. Du, J. Chen, Z. Duan, A Black-Box Adversarial Attack Method via Nesterov Accelerated Gradient and Rewiring Towards Attacking Graph Neural Networks. IEEE Transactions on Big Data, 9(6), (2023) 1586-1597. https://doi.org/10.1109/TBDATA.2023.3296936
  6. F. He, Y. Chen, R. Chen, W. Nie, Point Cloud Adversarial Perturbation Generation for Adversarial Attacks. IEEE Access, 11, (2023) 2767-2774. https://doi.org/10.1109/ACCESS.2023.3234313
  7. Y. Wang, T. Sun, S. Li, X. Yuan, W. Ni, E. Hossain, H.V. Poor, Adversarial attacks and defenses in machine learning-empowered communication systems and networks: A contemporary survey. IEEE Communications Surveys & Tutorials, 25(4), (2023) 2245-2298. https://doi.org/10.1109/COMST.2023.3319492
  8. S.M.K.A. Kazmi, N. Aafaq, M.A. Khan, M. Khalil, A. Saleem, From Pixel to Peril: Investigating Adversarial Attacks on Aerial Imagery Through Comprehensive Review and Prospective Trajectories. IEEE Access, 11, (2023) 81256-81278. https://doi.org/10.1109/ACCESS.2023.3299878
  9. Y. Shi, Y. Han, Q. Hu, Y. Yang, Q. Tian, Query-Efficient Black-Box Adversarial Attack With Customized Iteration and Sampling. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(2), (2023) 2226-2245. https://doi.org/10.1109/TPAMI.2022.3169802
  10. C. Shi, M. Zhang, Z. Lv, Q. Miao, C.M. Pun, Universal Object-Level Adversarial Attack in Hyperspectral Image Classification. IEEE Transactions on Geoscience and Remote Sensing, 61, (2023) 1-14. https://doi.org/10.1109/TGRS.2023.3336734
  11. W. Jiang, H. Li, G. Xu, T. Zhang, R. Lu, Physical Black-Box Adversarial Attacks Through Transformations. IEEE Transactions on Big Data, 9(3), (2023) 964-974. https://doi.org/10.1109/TBDATA.2022.3227318
  12. K. Mo, W. Tang, J. Li, X. Yuan, Attacking Deep Reinforcement Learning With Decoupled Adversarial Policy. IEEE Transactions on Dependable and Secure Computing, 20(1), (2023) 758-768. https://doi.org/10.1109/TDSC.2022.3143566
  13. L. Sun, Y. Dou, C. Yang, K. Zhang, J. Wang, S.Y. Philip, L. He, B. Li, Adversarial attack and defense on graph data: A survey. IEEE Transactions on Knowledge and Data Engineering, 35(8), (2022) 7693-7711. https://doi.org/10.1109/TKDE.2022.3201243
  14. L. Nguyen Vu, T.P. Doan, M. Bui, K. Hong, S. Jung, On the Defense of Spoofing Countermeasures Against Adversarial Attacks. IEEE Access, 11, (2023) 94563-94574. https://doi.org/10.1109/ACCESS.2023.3310809
  15. C. Wan, F. Huang and X. Zhao, Average Gradient-Based Adversarial Attack. IEEE Transactions on Multimedia, 25, (2023) 9572-9585. https://doi.org/10.1109/TMM.2023.3255742
  16. H. Teryak, A. Albaseer, M. Abdallah, S. Al-Kuwari, M. Qaraqe, Double-Edged Defense: Thwarting Cyber Attacks and Adversarial Machine Learning in IEC 60870-5-104 Smart Grids. IEEE Open Journal of the Industrial Electronics Society, 4, (2023) 629-642. https://doi.org/10.1109/OJIES.2023.3336234
  17. C. Qin, Y. Chen, K. Chen, X. Dong, W. Zhang, X. Mao, Y. He, N. Yu, Feature fusion based adversarial example detection against second-round adversarial attacks. IEEE Transactions on Artificial Intelligence, 4(5), (2022)1029-1040. https://doi.org/10.1109/TAI.2022.3190816
  18. R. Gipiškis, D. Chiaro, M. Preziosi, E. Prezioso, F. Piccialli, The Impact of Adversarial Attacks on Interpretable Semantic Segmentation in Cyber–Physical Systems. IEEE Systems Journal, 17(4), (2023) 5327-5334. https://doi.org/10.1109/JSYST.2023.3281079
  19. T. Chen, Z. Ma, Toward Robust Neural Image Compression: Adversarial Attack and Model Finetuning. IEEE Transactions on Circuits and Systems for Video Technology, 33(12), (2023) 7842-7856. https://doi.org/10.1109/TCSVT.2023.3276442
  20. S. Yan, J. Ren, W. Wang, L. Sun, W. Zhang, Q. Yu, A Survey of Adversarial Attack and Defense Methods for Malware Classification in Cyber Security. IEEE Communications Surveys & Tutorials, 25(1), (2023) 467-496. https://doi.org/10.1109/COMST.2022.3225137
  21. J. Pi, J. Zeng, Q. Lu, N. Jiang, H. Wu, L. Zeng, Z. Wu, Adv-Eye: A Transfer-Based Natural Eye Makeup Attack on Face Recognition. IEEE Access, 11, (2023) 89369-89382. https://doi.org/10.1109/ACCESS.2023.3307132
  22. R. Li, H. Liao, J. An, C. Yuen, L. Gan, IntraClass Universal Adversarial Attacks on Deep Learning-Based Modulation Classifiers. IEEE Communications Letters, 27(5), (2023) 1297-1301. https://doi.org/10.1109/LCOMM.2023.3261423
  23. X. Yuan, Z. Zhang, X. Wang, L. Wu, Semantic-Aware Adversarial Training for Reliable Deep Hashing Retrieval. IEEE Transactions on Information Forensics and Security, 18, (2023) 4681-4694. https://doi.org/10.1109/TIFS.2023.3297791
  24. L. Xu, J. Zhai, DCVAE-adv: A universal adversarial example generation method for white and black box attacks. Tsinghua Science and Technology, 29(2), (2023) 430-446. https://doi.org/10.26599/TST.2023.9010004
  25. L. Chen, Q.X. Zhu, Y.L. He, Adversarial attacks for neural network-based industrial soft sensors: Mirror output attack and translation mirror output attack. IEEE Transactions on Industrial Informatics, 20(2), (2023) 2378-2386. https://doi.org/10.1109/TII.2023.3291717