Segpgd: An effective and efficient adversarial attack for evaluating and boosting segmentation robustness
Deep neural network-based image classifications are vulnerable to adversarial
perturbations. The image classifications can be easily fooled by adding artificial small and …
perturbations. The image classifications can be easily fooled by adding artificial small and …
Towards efficient adversarial training on vision transformers
Abstract Vision Transformer (ViT), as a powerful alternative to Convolutional Neural Network
(CNN), has received much attention. Recent work showed that ViTs are also vulnerable to …
(CNN), has received much attention. Recent work showed that ViTs are also vulnerable to …
Prior-guided adversarial initialization for fast adversarial training
Fast adversarial training (FAT) effectively improves the efficiency of standard adversarial
training (SAT). However, initial FAT encounters catastrophic overfitting, ie, the robust …
training (SAT). However, initial FAT encounters catastrophic overfitting, ie, the robust …
Efficient and effective augmentation strategy for adversarial training
S Addepalli, S Jain - Advances in Neural Information …, 2022 - proceedings.neurips.cc
Abstract Adversarial training of Deep Neural Networks is known to be significantly more data-
hungry when compared to standard training. Furthermore, complex data augmentations …
hungry when compared to standard training. Furthermore, complex data augmentations …
Make some noise: Reliable and efficient single-step adversarial training
Abstract Recently, Wong et al.(2020) showed that adversarial training with single-step FGSM
leads to a characteristic failure mode named catastrophic overfitting (CO), in which a model …
leads to a characteristic failure mode named catastrophic overfitting (CO), in which a model …
Comparative analysis of binary and one-class classification techniques for credit card fraud data
The yearly increase in incidents of credit card fraud can be attributed to the rapid growth of e-
commerce. To address this issue, effective fraud detection methods are essential. Our …
commerce. To address this issue, effective fraud detection methods are essential. Our …
Improving fast adversarial training with prior-guided knowledge
Fast adversarial training (FAT) is an efficient method to improve robustness in white-box
attack scenarios. However, the original FAT suffers from catastrophic overfitting, which …
attack scenarios. However, the original FAT suffers from catastrophic overfitting, which …
Shift from texture-bias to shape-bias: Edge deformation-based augmentation for robust object recognition
Recent studies have shown the vulnerability of CNNs under perturbation noises, which is
partially caused by the reason that the well-trained CNNs are too biased toward the object …
partially caused by the reason that the well-trained CNNs are too biased toward the object …
Learning defense transformations for counterattacking adversarial examples
Deep neural networks (DNNs) are vulnerable to adversarial examples with small
perturbations. Adversarial defense thus has been an important means which improves the …
perturbations. Adversarial defense thus has been an important means which improves the …
Towards intrinsic adversarial robustness through probabilistic training
Modern deep neural networks have made numerous breakthroughs in real-world
applications, yet they remain vulnerable to some imperceptible adversarial perturbations …
applications, yet they remain vulnerable to some imperceptible adversarial perturbations …