Patch Selection Denoiser: An Effective Approach Defending Against One-Pixel Attacks

Chen, Dong; Xu, Ruqiao; Han, Bo

doi:10.1007/978-3-030-36802-9_31

Cited by 13 publications

(11 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Propose a three-staged noise elimination and reconstructing algorithm to defend the N -pixel attack without affecting the image integrity. • Compare with [2], our proposed algorithm does not require retraining a new model, thus causing no additional cost of performance.…”

Section: Our Resultsmentioning

confidence: 99%

“…There are still less solutions to this new kind of adversarial threat. To date the most promising defending method is Patch Selection Denoiser (PSD) [2], proposed by Chen et al in 2019. The authors use another neural network and local-patch method to removes the potential attacking pixels.…”

Section: Related Workmentioning

confidence: 99%

“…Before our experiment, we follow [11] to train a CNN-ResNet model [5] using CIFAR-10 dataset, which contains 60,000 32x32 color images (50,000 training images and 10,000 test images) from 10 classes, each containing 6,000 images. 2 In experiment I, we first input each image in the original test dataset into the proposed algorithm to generate a new dataset, called reconstructed dataset. Then, we runs two times of inference by feeding original test dataset and reconstructed dataset into the trained models respectively.…”

Section: Experiments Setupmentioning

confidence: 99%

“…In other words, it is more difficult to resist this attack by only increasing the layer of neural network. The latest defense against one-pixel attack is proposed by Chen et al in 2019, called patch selection denoiser (PSD) [2], which uses local patches to remove the potential attacking pixels [2]. Although their architecture has a high probability of resisting N -pixel attack (i.e., multiple pixels are attacked), they must additionally train another new model to eliminate noise.…”

Section: Introduction 1backgroundmentioning

confidence: 99%

See 3 more Smart Citations

Defense against N-pixel Attacks based on Image Reconstruction

Liu

Wang

Hsiao

et al. 2020

Proceedings of the 8th International Workshop on Security in Blockchain and Cloud Computing

View full text Add to dashboard Cite

Since machine learning and deep learning are largely used for image recognition in real-world applications, how to avoid adversarial attacks become an important issue. It is common that attackers add adversarial perturbation to a normal image in order to fool the models. The N-pixel attack is one of the recently popular adversarial methods by simply changing a few pixels in the image. We observe that changing the few pixels leads to an obvious difference with its neighboring pixels. Therefore, this research aims to defend the N-pixel attacks based on image reconstruction. We develop a three-staged reconstructing algorithm to recover the fooling images. Experimental results show that the accuracy of CIFAR-10 test dataset can reach 92% after applying our proposed algorithm, indicating that the algorithm can maintain the original inference accuracy on normal dataset. Besides, the effectiveness of defending N-pixel attacks is also validated by reconstructing 500 attacked images using the proposed algorithm. The results show that we have a 90% to 92% chance of successful defense, where N =1,3,5,10,and 15. CCS CONCEPTS • Security and privacy → Security services; Systems security; • Computing methodologies → Artificial intelligence.

show abstract

Section: Our Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Experiments Setupmentioning

confidence: 99%

Section: Introduction 1backgroundmentioning

confidence: 99%

See 2 more Smart Citations

Defense against N-pixel Attacks based on Image Reconstruction

Liu

Wang

Hsiao

et al. 2020

Proceedings of the 8th International Workshop on Security in Blockchain and Cloud Computing

View full text Add to dashboard Cite

show abstract

“…For instance, in one-pixel attacks, altering only few pixels in an original image is enough to fool a deep neural network [6,7]. Therefore, filtering algorithms dedicated to the suppression of impulsive disturbances in color images and also considered as defensive methods against adversarial attacks, have attracted considerable interest among many researchers [8][9][10][11][12][13][14][15][16][17].…”

Section: Introductionmentioning

confidence: 99%

Is large improvement in efficiency of impulsive noise removal in color images still possible?

2021

View full text Add to dashboard Cite

The substantial improvement in the efficiency of switching filters, intended for the removal of impulsive noise within color images is described. Numerous noisy pixel detection and replacement techniques are evaluated, where the filtering performance for color images and subsequent results are assessed using statistical reasoning. Denoising efficiency for the applied detection and interpolation techniques are assessed when the location of corrupted pixels are identified by noisy pixel detection algorithms and also in the scenario when they are already known. The results show that improvement in objective quality measures can be achieved by using more robust detection techniques, combined with novel methods of corrupted pixel restoration. A significant increase in the image denoising performance is achieved for both pixel detection and interpolation, surpassing current filtering methods especially via the application of a convolutional network. The interpolation techniques used in the image inpainting methods also significantly increased the efficiency of impulsive noise removal.

show abstract