ASQ-FastBM3D: An Adaptive Denoising Framework for Defending Adversarial Attacks in Machine Learning Enabled Systems

Xu, Guangquan; Han, Zheng‐Bo; Gong, Lixiao; Jiao, Litao; Bai, Hongpeng; Liu, Shaoying; Zheng, Xi

doi:10.1109/tr.2022.3171420

Cited by 5 publications

(2 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this work, 7 advanced image reconstruction methods are adopted for comparison, namely, EPLL [15], NL-Means [16], BM3D [17], NL-Bayes [18], BLS-GSM [19], DCT [20] The peak-signal-to-noise (PSNR) ratio is adopted to demonstrate the effectiveness of the presented DGMR model and comparison image reconstruction methods. The PSNR is defined as follows:…”

Section: Comparison Methods and Evaluation Metricmentioning

confidence: 99%

Simultaneous High-Resolution Blind Image Reconstruction and Perturbation Defense

Xue,

Zheng,

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Image reconstruction plays an important role in modern electronic information systems and equipment and has attracted widespread intense attention in academic and industrial fields. Restoring delicate image particulars using traditional handcrafted prior image reconstruction methods is often difficult because of the inferior prior characterization abilities of these methods. In this article, an efficient image reconstruction model, named DGMR, is proposed based on deep Gaussian mixture learning. In particular, a channel attention mechanism is designed for image spatial correlation exploitation. Both the external and internal information are explored using deep Gaussian mixture. A residual Swin transformer module is constructed to learn Gaussian mixture priors, including the images ' means and variances; in contrast, existing methods calculate only the mean and ignore the variance. Moreover, sparse regularized united learning is developed to improve invariant representation learning ability, and the model is customized by internal learning with spatial constraints and regularization. Extensive qualitative and quantitative experiments are performed, confirming that DGMR is superior to the current advanced comparison approaches.INDEX TERMS Deep learning, Gaussian mixture, image reconstruction, prior learning.

show abstract

Section: Comparison Methods and Evaluation Metricmentioning

confidence: 99%

Simultaneous High-Resolution Blind Image Reconstruction and Perturbation Defense

Xue,

Zheng,

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…In order to defend against the threats posed by adversarial attacks to artificial intelligence security applications, researchers have investigated multiple network models for adversarial attack defense methods, which at this stage are mainly divided into three categories: (1) data preprocessing for adversarial examples; (2) enhancing the robustness of deep neural networks; and (3) detecting adversarial examples. Data preprocessing methods such as denoising ( Aneja et al, 2022 ; Xu et al, 2022 ) and data compression ( Chang et al, 2022 ; Zhang, Yi & Sang, 2022 ). The advantages of these methods are faster computation and no need to modify the network structure, the disadvantages are that denoising and data compression can cause loss of information in the image, the neural network cannot extract features adequately, which makes the neural network make wrong judgments.…”

Section: Introductionmentioning

confidence: 99%

Enhancing the robustness of vision transformer defense against adversarial attacks based on squeeze-and-excitation module

Chang

Hong

Wang

2023

PeerJ Computer Science

View full text Add to dashboard Cite

Vision Transformer (ViT) models have achieved good results in computer vision tasks, their performance has been shown to exceed that of convolutional neural networks (CNNs). However, the robustness of the ViT model has been less studied recently. To address this problem, we investigate the robustness of the ViT model in the face of adversarial attacks, and enhance the robustness of the model by introducing the ResNet- SE module, which acts on the Attention module of the ViT model. The Attention module not only learns edge and line information, but also can extract increasingly complex feature information; ResNet-SE module highlights the important information of each feature map and suppresses the minor information, which helps the model to perform the extraction of key features. The experimental results show that the accuracy of the proposed defense method is 19.812%, 17.083%, 18.802%, 21.490%, and 18.010% against Basic Iterative Method (BIM), C&W, DeepFool, DI2FGSM, and MDI2FGSM attacks, respectively. The defense method in this paper shows strong robustness compared with several other models.

show abstract

A Graph Neural Network Model for Live Face Anti-Spoofing Detection Camera Systems

Xu,

Lin,

Fan

et al. 2024

IEEE Internet Things J.

View full text Add to dashboard Cite

ASQ-FastBM3D: An Adaptive Denoising Framework for Defending Adversarial Attacks in Machine Learning Enabled Systems

Cited by 5 publications

References 33 publications

Simultaneous High-Resolution Blind Image Reconstruction and Perturbation Defense

Simultaneous High-Resolution Blind Image Reconstruction and Perturbation Defense

Enhancing the robustness of vision transformer defense against adversarial attacks based on squeeze-and-excitation module

A Graph Neural Network Model for Live Face Anti-Spoofing Detection Camera Systems

Contact Info

Product

Resources

About