Comparison of Adversarial Robustness of Convolutional Neural Networks for Handwritten Digit Recognition

DOI

10.2991/978-94-6463-540-9_54How to use a DOI?

Keywords

Convolutional Neural Network; Fast Gradient Sign Method; Adversarial Samples

Abstract

Machine learning has found widespread application in contemporary society, yet it remains vulnerable to the corrosive effects of adversarial samples. These refer to input data that has been deliberately modified in a certain way to mislead machine learning models. While these modifications may be undetectable to human observers, they are sufficient to trigger erroneous outputs from machine learning models, thereby compromising their robustness, and exposing their weaknesses. The purpose of this paper is to examine the vulnerability of machine learning models to adversarial samples. The Fast Gradient Sign Method (FGSM) is used to create adversarial samples from the Modified National Institute of Standards and Technology (MNIST) dataset, which are then used to attack the LeNet and a basic convolutional neural network (CNN) model. The findings reveal that the LeNet model exhibits a higher degree of sensitivity compared to the simple CNN model. As time progresses and models continue to innovate, they are becoming less prone to interference from adversarial samples. This study could offer valuable insights for future endeavors aimed at designing more secure and resilient machine learning models.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Zhen Ren
PY  - 2024
DA  - 2024/10/16
TI  - Comparison of Adversarial Robustness of Convolutional Neural Networks for Handwritten Digit Recognition
BT  - Proceedings of the 2024 2nd International Conference on Image, Algorithms and Artificial Intelligence (ICIAAI 2024)
PB  - Atlantis Press
SP  - 547
EP  - 552
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-540-9_54
DO  - 10.2991/978-94-6463-540-9_54
ID  - Ren2024
ER  -