Open Access Open Access  Restricted Access Subscription or Fee Access

Adversarial Examples in Processing of High-Tech Imaging Using DNN

A. Vatian, N. Gusarova, I. Tomilov, A. Shalyto


The article investigates the author-detected effect of noise from devices forming reconstructed high-tech images on their classification performance by means of deep neural networks. The main observation is that Computed Tomography and Magnetic Resonance Imaging devices introduce noise, which causes an effect called “natural adversarial examples”, which is similar to the adversarial examples attack. These noisy images alter a manifold in the intermediate feature space of a deep neural network, making the network less robust. The article suggests simulating their effects using the fast gradient sign method (FGSM) attack model. To mitigate their influence, it is proposed to augment the training dataset with noises of a similar nature as natural adversarial examples (AEs) (rather than trying to remove the latter from the original dataset). The paper describes a training pipeline and its parametrization for achieving best performance and robustness to the natural adversarial examples for high-tech medical images classification using deep neural networks (DNN). As a result of the research, we have shown that due to the successful selection of the architecture and network parameters, primarily the variants of the activation function and augmentation of the original dataset, it is possible to achieve a suboptimal ratio of accuracy and f1-score, on the one hand, and robustness to natural AE, on the other hand.


Adversarial examples, high-tech images, reconstructed images, deep neural networks, noise, robustness.

Full Text:


Disclaimer/Regarding indexing issue:

We have provided the online access of all issues and papers to the indexing agencies (as given on journal web site). It’s depend on indexing agencies when, how and what manner they can index or not. Hence, we like to inform that on the basis of earlier indexing, we can’t predict the today or future indexing policy of third party (i.e. indexing agencies) as they have right to discontinue any journal at any time without prior information to the journal. So, please neither sends any question nor expects any answer from us on the behalf of third party i.e. indexing agencies.Hence, we will not issue any certificate or letter for indexing issue. Our role is just to provide the online access to them. So we do properly this and one can visit indexing agencies website to get the authentic information. Also: DOI is paid service which provided by a third party. We never mentioned that we go for this for our any journal. However, journal have no objection if author go directly for this paid DOI service.