Addressing the Vulnerability of Neural Networks to Adversarial Attacks: Challenges, Implications and Solutions for Safety-Critical Applications

References

[1] Akhtar, N., & Mian, A. (2018). Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey. IEEE Access.

[2] Dhillon, G. S., Azizzadenesheli, K., Khanna, A., Kantorov, V., & Anandkumar, A. (2018). Stochastic Activation Pruning for Robust Adversarial Defense. Advances in Neural Information Processing Systems (NeurIPS).

[3] Uesato, J., O’Donoghue, B., & Kohli, P. (2018). Adversarial Risk and the Dangers of Evaluating Against Weak Attacks. Advances in Neural Information Processing Systems (NeurIPS).

[4] Wong, E., & Kolter, J. Z. (2018). Provable Defenses against Adversarial Examples via the Convex Outer Adversarial Polytope. International Conference on Learning Representations (ICLR).

[5] Wong, E., & Kolter, J. Z. (2018). Scaling provable adversarial defenses. Advances in Neural Information Processing Systems (NeurIPS).

[6] Zhang, C., Cisse, M., Dauphin, Y. N., & Lopez-Paz, D. (2019). Mixup: Beyond Empirical Risk Minimization. International Conference on Learning Representations (ICLR).

[7] Wang, H., & Shan, S. (2019). Using Adversarial Examples to Understand the Robustness of Deep Learning Models for Biomedical Image Segmentation. IEEE Journal of Biomedical and Health Informatics.

[8] Schott, L., Rauber, J., & Bethge, M. (2019). Towards the first adversarially robust neural network model on MNIST. International Conference on Learning Representations (ICLR).

[9] Gowal, S., Dvijotham, K., Stanforth, R., Qin, C., Uesato, J., Mann, T., & Kohli, P. (2019). Exploring the Landscape of Spatial Robustness. International Conference on Learning Representations (ICLR).

[10] Hein, M., Andriushchenko, M., & Bitterwolf, J. (2019). Why ReLU networks yield high-confidence predictions far away from the training data and how to mitigate the problem. International Conference on Learning Representations (ICLR).

[11] Zhang, Y., Wang, Z., Ji, X., & Wang, X. (2020). Combating Adversarial Attacks with Sparse Adversarial Perturbations. AAAI Conference on Artificial Intelligence (AAAI).

[12] Song, Y., Kim, T., Nowozin, S., Ermon, S., & Kushman, N. (2020). Improving Adversarial Robustness Requires Revisiting Misclassified Examples. International Conference on Learning Representations (ICLR).

[13] Gowal, S., Dvijotham, K., Stanforth, R., Bunel, R., Qin, C., Uesato, J., ... & Kohli, P. (2020). On the Effectiveness of Low-Frequency Perturbations. International Conference on Learning Representations (ICLR).

[14] Xie, C., Wang, J., Zhang, Z., Ren, Z., & Yuille, A. (2020). Adversarial Examples for Semantic Segmentation and Object Detection. IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15] Gowal, S., Dvijotham, K., Stanforth, R., Uesato, J., Mann, T., Kohli, P., & Maddison, C. J. (2020). On the effectiveness of low-frequency perturbations. International Conference on Learning Representations (ICLR).

@article{1705997,
      author = {Nagaraj C, Dr. Hemalatha B, Dr. K. Jamberi},
      title = {Addressing the Vulnerability of Neural Networks to Adversarial Attacks: Challenges, Implications and Solutions for Safety-Critical Applications},
      journal = {Iconic Research And Engineering Journals},
      year = {2024},
      volume = {8},
      number = {1},
      pages = {43-47},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1705997.pdf},
      abstract = {Neural networks have demonstrated unparalleled success in various domains, yet challenges persist regarding their robustness and generalization capabilities. A significant concern is their vulnerability to adversarial attacks, where imperceptible perturbations in input data can cause erroneous predictions. This paper offers a comprehensive examination of the phenomenon of adversarial attacks on neural networks. Through empirical analysis and theoretical insights, we elucidate the mechanisms underlying these attacks and their implications for real-world deployment. Additionally, we investigate state-of-the-art defense mechanisms and mitigation strategies aimed at bolstering the robustness of neural networks against adversarial manipulation. By addressing these challenges head-on, we aim to contribute to the advancement of neural network security and reliability, facilitating their safe and effective integration into safety-critical systems.},
      keywords = {Neural networks, Adversarial attacks, Robustness, Generalization, Safety-critical applications, Defense mechanisms, Mitigation strategies.},
      month = {July},
  }