Email this article 
Study of neural network robustness in the task of pattern recognition
- Authors: Kharrasov K.R.1, Moseva M.S.1, Gorodnichev M.G.1
-
Affiliations:
- Moscow Technical University of Communication and Informatics
- Issue: Vol 31, No 2 (2025)
- Pages: 87-92
- Section: Neural network technologies
- Published: 15.02.2025
- URL: https://journals.eco-vector.com/1684-6400/article/view/702195
- DOI: https://doi.org/10.17587/it.31.87-92
- ID: 702195
Cite item
Open Access
Access granted
Subscription or Fee Access
Abstract
The problem of stable pattern recognition in an image is considered. Types and types of attacks on machine learning systems and methods of defense against them are discussed. An experiment with the application of the described approach of robust image recognition to adversarial attacks is carried out and the reliability of conventional and robust neural network classifiers is compared on the basis of the resulting metrics.
About the authors
K. R. Kharrasov
Moscow Technical University of Communication and Informatics
Author for correspondence.
Email: k.r.harrasov@edu.mtuci.ru
Assistant
Russian Federation, MoscowM. S. Moseva
Moscow Technical University of Communication and Informatics
Email: m.s.moseva@mtuci.ru
PhD, Senior Lecturer
Russian Federation, MoscowM. G. Gorodnichev
Moscow Technical University of Communication and Informatics
Email: m.g.gorodnichev@mtuci.ru
PhD, Assistant Professor
Russian Federation, MoscowReferences
- Goodfellow I., Shlens J., Szegedy C. Explaining and Harnessing Adversarial Examples, arXiv: 1412.6572.
- Goodfellow I., Yoshua B. Deep learning, Cambridge, Massachusetts, MIT Press, 2016, pp. 180—184.
- The MNIST dataset of handwritten digits, available at: https://www.kaggle.com/datasets/hojjatk/mnist-dataset)
- Goodfellow I., Warde-Farley D., Mirza M., Courville A., Yoshua B. Maxout Networks, arXiv: 1302.4389.
- Moosavi-Dezfooli S.-M., Fawzi A., Frossard P. DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). P. 2574—2582.
- Su J., Vargas D. V., Kouichi S. One pixel attack for fooling deep neural networks, arXiv: 1710.08864.
- Li H., Namiot D. A Survey of Adversarial Attacks and Defenses for image data on Deep Learning, International Journal of Open Information Technologies, 2022, vol. 10, no. 5, pp. 9—16.
- Namiot D., Ilyushin E., Chizhov I. The rationale for working on robust machine learning, International Journal of Open Information Technologies, 2021, vol. 9, no. 11, pp. 68—74.
- Namiot D., Ilyushin E., Chizhov I. Artificial intelligence and cybersecurity, International Journal of Open Information Technologies, 2022, vol. 10, no. 9, pp. 135—147.
- Schott L., Rauber J., Bethge M., Brendel W. Towards the first adversarially robust neural network model on MNIST, arXiv: 1805.09190.
- Song Y., Kim T., Nowozin S., Ermon S., Kushman N. PixelDefend: Leveraging Generative Models to Understand and Defend against Adversarial Examples, arXiv: 1710.10766.
- Madry A., Makelov A., Schmidt L., Tsipras D., Vladu A. Towards Deep Learning Models Resistant to Adversarial Attacks, arXiv: 1706.06083.
Supplementary files
