Adversarial Perturbations Prevail in the Y-Channel of the YCbCr Color Space
arXiv:2003.00883
Abstract
Deep learning offers state of the art solutions for image recognition. However, deep models are vulnerable to adversarial perturbations in images that are subtle but significantly change the model's prediction. In a white-box attack, these perturbations are generally learned for deep models that operate on RGB images and, hence, the perturbations are equally distributed in the RGB color space. In this paper, we show that the adversarial perturbations prevail in the Y-channel of the YCbCr space. Our finding is motivated from the fact that the human vision and deep models are more responsive to shape and texture rather than color. Based on our finding, we propose a defense against adversarial images. Our defence, coined ResUpNet, removes perturbations only from the Y-channel by exploiting ResNet features in an upsampling framework without the need for a bottleneck. At the final stage, the untouched CbCr-channels are combined with the refined Y-channel to restore the clean image. Note that ResUpNet is model agnostic as it does not modify the DNN structure. ResUpNet is trained end-to-end in Pytorch and the results are compared to existing defence techniques in the input transformation category. Our results show that our approach achieves the best balance between defence against adversarial attacks such as FGSM, PGD and DDN and maintaining the original accuracies of VGG-16, ResNet50 and DenseNet121 on clean images. We perform another experiment to show that learning adversarial perturbations only for the Y-channel results in higher fooling rates for the same perturbation magnitude.
References in corpus (16)
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks
- Deep Learning for Single Image Super-Resolution: A Brief Review
- Towards Deep Neural Network Architectures Robust to Adversarial Examples
- Countering Adversarial Images using Input Transformations
- Understanding Adversarial Training: Increasing Local Stability of Neural Nets through Robust Optimization
- A study of the effect of JPG compression on adversarial images
- Keeping the Bad Guys Out: Protecting and Vaccinating Deep Learning with JPEG Compression
- Certified Adversarial Robustness with Additive Noise
- Simple Black-Box Adversarial Perturbations for Deep Networks
- advertorch v0.1: An Adversarial Robustness Toolbox based on PyTorch
- Deep Learning for Multiple-Image Super-Resolution
- Deep Learning for Image Super-resolution: A Survey
- Robust Neural Networks using Randomized Adversarial Training
- Sampling and Super-resolution of Sparse Signals Beyond the Fourier Domain
- An Empirical Investigation of Randomized Defenses against Adversarial Attacks
Cited by in corpus (5)
- Imperceptible Adversarial Examples by Spatial Chroma-Shift
- Adversarial Image Generation by Spatial Transformation in Perceptual Colorspaces
- Improving Robustness of Learning-based Autonomous Steering Using Adversarial Images
- Data Quality Matters For Adversarial Training: An Empirical Study
- Defense-friendly Images in Adversarial Attacks: Dataset and Metrics for Perturbation Difficulty