Analyzing Adversarial Robustness of Deep Neural Networks in Pixel Space: a Semantic Perspective
Not on the map. Why?
Abstract (arXiv)
The vulnerability of deep neural networks to adversarial examples, which are crafted maliciously by modifying the inputs with imperceptible perturbations to misled the network produce incorrect outputs, reveals the lack of robustness and poses security concerns. Previous works study the adversarial robustness of image classifiers on image level and use all the pixel information in an image indiscriminately, lacking of exploration of regions with different semantic meanings in the pixel space of an image. In this work, we fill this gap and explore the pixel space of the adversarial image by proposing an algorithm to looking for possible perturbations pixel by pixel in different regions of the segmented image. The extensive experimental results on CIFAR-10 and ImageNet verify that searching for the modified pixel in only some pixels of an image can successfully launch the one-pixel adversarial attacks without requiring all the pixels of the entire image, and there exist multiple vulnerable points scattered in different regions of an image. We also demonstrate that the adversarial robustness of different regions on the image varies with the amount of semantic information contained.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 32 of its 43 references that are in the map. First posted on arXiv 18 Jun 2021.
Builds on
32 of 43 references.
References · 32
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
- 2014Explaining and Harnessing Adversarial Examplescited by 7,364
- 2013Intriguing properties of neural networkscited by 5,155
- 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
- 2016Adversarial examples in the physical worldcited by 1,769
- 2015DeepFool: a simple and accurate method to fool deep neural networkscited by 1,318
- 2014Striving for Simplicity: The All Convolutional Netcited by 990
- 2016Adversarial Machine Learning at Scalecited by 936
- 2016Practical Black-Box Attacks against Machine Learningcited by 903
- 2015The Limitations of Deep Learning in Adversarial Settingscited by 790
20 more
- 2016Universal adversarial perturbationscited by 784
- 2017ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Modelscited by 666
- 2013Network In Networkcited by 605
- 2017One pixel attack for fooling deep neural networkscited by 582
- 2016Delving into Transferable Adversarial Examples and Black-box Attackscited by 536
- 2017Decision-Based Adversarial Attacks: Reliable Attacks Against Black-Box Machine Learning Modelscited by 462
- 2016Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samplescited by 446
- 2020U2-Net: Going Deeper with Nested U-Structure for Salient Object Detectioncited by 412
- 2014Towards Deep Neural Network Architectures Robust to Adversarial Examplescited by 146
- 2015Ask Me Anything: Dynamic Memory Networks for Natural Language Processingcited by 117
- 2016Robustness of classifiers: from adversarial to random noisecited by 88
- 2016A Boundary Tilting Persepective on the Phenomenon of Adversarial Examplescited by 70
- 2017Dense Associative Memory is Robust to Adversarial Inputscited by 45
- 2016Adversarial Perturbations Against Deep Neural Networks for Malware Classificationcited by 42
- 2015Exploring the Space of Adversarial Imagescited by 28
- 2017Intriguing Properties of Adversarial Examplescited by 20
- 2017On the Limitation of Convolutional Neural Networks in Recognizing Negative Imagescited by 17
- 2016Are Accuracy and Robustness Correlated?cited by 11
- 2016Towards Robust Deep Neural Networks with BANGcited by 8
- 2020Improving adversarial robustness of deep neural networks by using semantic informationcited by 2
Built on by
Not cited yet.