DOI:
https://doi.org/10.64539/sjer.v1i4.2025.349Keywords:
CNN, Adversarial attacks, Anti-noise predictor, Machine learning security, Image classification, Adversarial detectionAbstract
This work explores the vulnerability of Convolutional Neural Networks (CNNs) to adversarial attacks, particularly focusing on the Fast Gradient Sign Method (FGSM). Adversarial attacks, which subtly manipulate input images to deceive machine learning models, pose significant threats to the security and reliability of CNN-based systems. The research introduces an enhanced methodology for identifying and mitigating these adversarial threats by incorporating an anti-noise predictor to separate adversarial noise and images, thereby improving detection accuracy. The proposed method was evaluated against multiple adversarial attack strategies using the MNIST dataset, demonstrating superior detection performance compared to existing techniques. Additionally, the study highlights the integration of Fourier domain-based noise accommodation, enhancing robustness against attacks. The findings contribute to the development of more resilient CNN models capable of effectively countering adversarial manipulations, emphasizing the importance of continuous adaptation and multi-layered defense strategies in securing machine learning systems.
References
[1] Y. Lu, “Artificial intelligence: a survey on evolution, models, applications and future trends,” Journal of Management Analytics, vol. 6, no. 1, pp. 1–29, Jan. 2019, https://doi.org/10.1080/23270012.2019.1570365.
[2] S. A. Nawaz, J. Li, U. A. Bhatti, M. U. Shoukat, and R. M. Ahmad, “AI-based object detection latest trends in remote sensing, multimedia and agriculture applications,” Front Plant Sci, vol. 13, Nov. 2022, https://doi.org/10.3389/fpls.2022.1041514.
[3] A. M. Zbrzezny and A. E. Grzybowski, “Deceptive Tricks in Artificial Intelligence: Adversarial Attacks in Ophthalmology,” J Clin Med, vol. 12, no. 9, p. 3266, May 2023, https://doi.org/10.3390/jcm12093266.
[4] X. Yuan, P. He, Q. Zhu, and X. Li, “Adversarial Examples: Attacks and Defenses for Deep Learning,” IEEE Trans Neural Netw Learn Syst, vol. 30, no. 9, pp. 2805–2824, Sep. 2019, https://doi.org/10.1109/TNNLS.2018.2886017.
[5] A. Bajaj and D. K. Vishwakarma, “A state-of-the-art review on adversarial machine learning in image classification,” Multimed Tools Appl, vol. 83, no. 3, pp. 9351–9416, Jan. 2024, https://doi.org/10.1007/s11042-023-15883-z.
[6] T. Bai, J. Luo, J. Zhao, B. Wen, and Q. Wang, “Recent Advances in Adversarial Training for Adversarial Robustness,” arXiv preprint arXiv:2102.01356, 2021, https://doi.org/10.48550/arXiv.2102.01356.
[7] M. Ivgi and J. Berant, “Achieving Model Robustness through Discrete Adversarial Training,” arXiv preprint arXiv:2104.05062, 2021, https://doi.org/10.48550/arXiv.2104.05062.
[8] K. Zhao, L. Wang, F. Yu, B. Zeng, and Z. Pang, “FedMP: A multi-pronged defense algorithm against Byzantine poisoning attacks in federated learning,” Computer Networks, vol. 257, p. 110990, Feb. 2025, https://doi.org/10.1016/j.comnet.2024.110990.
[9] C.-Y. Hsu, P.-Y. Chen, S. Lu, S. Liu, and C.-M. Yu, “Adversarial Examples Can Be Effective Data Augmentation for Unsupervised Machine Learning,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 36, no. 6, pp. 6926–6934, Jun. 2022, https://doi.org/10.1609/aaai.v36i6.20650.
[10] I. H. Sarker, “Multi‐aspects AI‐based modeling and adversarial learning for cybersecurity intelligence and robustness: A comprehensive overview,” Security and Privacy, vol. 6, no. 5, Sep. 2023, https://doi.org/10.1002/spy2.295.
[11] M. Repetto, “Adaptive monitoring, detection, and response for agile digital service chains,” Comput Secur, vol. 132, p. 103343, Sep. 2023, https://doi.org/10.1016/j.cose.2023.103343.
[12] J. Fang, Y. Jiang, C. Jiang, Z. L. Jiang, C. Liu, and S.-M. Yiu, “State-of-the-art optical-based physical adversarial attacks for deep learning computer vision systems,” Expert Syst Appl, vol. 250, p. 123761, Sep. 2024, https://doi.org/10.1016/j.eswa.2024.123761.
[13] D. J. Miller, Z. Xiang, and G. Kesidis, “Adversarial Learning Targeting Deep Neural Network Classification: A Comprehensive Review of Defenses Against Attacks,” Proceedings of the IEEE, vol. 108, no. 3, pp. 402–433, Mar. 2020, https://doi.org/10.1109/JPROC.2020.2970615.
[14] A. Abomakhelb, K. A. Jalil, A. G. Buja, A. Alhammadi, and A. M. Alenezi, “A Comprehensive Review of Adversarial Attacks and Defense Strategies in Deep Neural Networks,” Technologies (Basel), vol. 13, no. 5, p. 202, May 2025, https://doi.org/10.3390/technologies13050202.
[15] M. Hassan, S. Younis, A. Rasheed, and M. Bilal, “Integrating single-shot Fast Gradient Sign Method (FGSM) with classical image processing techniques for generating adversarial attacks on deep learning classifiers,” in Fourteenth International Conference on Machine Vision (ICMV 2021), W. Osten, D. Nikolaev, and J. Zhou, Eds., SPIE, Mar. 2022, p. 48. https://doi.org/10.1117/12.2623585.
[16] A. Agarwal, R. Singh, and M. Vatsa, “The Role of ‘Sign’ and ‘Direction’ of Gradient on the Performance of CNN,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), IEEE, Jun. 2020, pp. 2748–2756. https://doi.org/10.1109/CVPRW50498.2020.00331.
[17] S. M. A. Naqvi, M. Shabaz, M. A. Khan, and S. I. Hassan, “Adversarial Attacks on Visual Objects Using the Fast Gradient Sign Method,” J Grid Comput, vol. 21, no. 4, p. 52, Dec. 2023, https://doi.org/10.1007/s10723-023-09684-9.
[18] W. Pedrycz and K.-C. Kwak, “The Development of Incremental Models,” IEEE Transactions on Fuzzy Systems, vol. 15, no. 3, pp. 507–518, Jun. 2007, https://doi.org/10.1109/TFUZZ.2006.889967.
[19] C. J. Trammell, M. G. Pleszkoch, R. C. Linger, and A. R. Hevner, “The incremental development process in Cleanroom software engineering,” Decis Support Syst, vol. 17, no. 1, pp. 55–71, Apr. 1996, https://doi.org/10.1016/0167-9236(95)00022-4.
[20] H. Ben Braiek and F. Khomh, “On testing machine learning programs,” Journal of Systems and Software, vol. 164, p. 110542, Jun. 2020, https://doi.org/10.1016/j.jss.2020.110542.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Maaz Nisar, Nabeel Fayyaz, Muhammad Abdullah Ahmed, Muhammad Usman Shams, Bushra Fareed
This work is licensed under a Creative Commons Attribution 4.0 International License.
