Shape-Texture Debiased Neural Network Training

Li, Yingwei; Yu, Qihang; Tan, Mingxing; Mei, Jieru; Tang, Peng; Shen, Wei; Yuille, Alan; Xie, Cihang

doi:10.48550/arxiv.2010.05981

Cited by 15 publications

(14 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In contrast, a texture-biased model is trained if the translated image is labeled as lemon. To balance the bias, the translated image by style transferring is taken with two labels [65], both chimpanzee and lemon, which leads to a de-biased model. Further, inspired by Mixup [38], Hong et al propose…”

Section: Label-changingmentioning

confidence: 99%

A Comprehensive Survey of Image Augmentation Techniques for Deep Learning

Xu¹,

Yoon²,

Fuentes³

et al. 2022

Preprint

View full text Add to dashboard Cite

Deep learning has been achieving decent performance in computer vision requiring a large volume of images, however, collecting images is expensive and difficult in many scenarios. To alleviate this issue, many image augmentation algorithms have been proposed as effective and efficient strategies. Understanding current algorithms is essential to find suitable methods or develop novel techniques for given tasks. In this paper, we perform a comprehensive survey on image augmentation for deep learning with a novel informative taxonomy.To get the basic idea why we need image augmentation, we introduce the challenges in computer vision tasks and vicinity distribution. Then, the algorithms are split into three categories; model-free, model-based, and optimizing policybased. The model-free category employs image processing methods while the

show abstract

Section: Label-changingmentioning

confidence: 99%

A Comprehensive Survey of Image Augmentation Techniques for Deep Learning

Xu¹,

Yoon²,

Fuentes³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Both (Shafahi et al, 2019) and (Zhang et al, 2019) propose to merge the gradient for adversarial attacks and the gradient for network parameter updates into a single forward and backward pass to reduce computations. Wong et al (Wong et al, 2020) (Chen et al, 2021b;Xu et al, 2021;Shu et al, 2020;Chen et al, 2021a;Gong et al, 2021), under different learning paradigms Ho & Vasconcelos, 2020;Xu & Yang, 2020), with different adversarial data (Merchant et al, 2020;Li et al, 2020;Herrmann et al, 2021), enabling extremely large-batch training (Liu et al, 2022), etc. In this paper, rather than furthering performance, we aim to make AdvProp "free".…”

Section: Related Workmentioning

confidence: 99%

Fast AdvProp

Mei¹,

Han²,

Bai³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Adversarial Propagation (AdvProp) is an effective way to improve recognition models, leveraging adversarial examples. Nonetheless, AdvProp suffers from the extremely slow training speed, mainly because: a) extra forward and backward passes are required for generating adversarial examples; b) both original samples and their adversarial counterparts are used for training (i.e., 2× data). In this paper, we introduce Fast AdvProp, which aggressively revamps AdvProp's costly training components, rendering the method nearly as cheap as the vanilla training. Specifically, our modifications in Fast AdvProp are guided by the hypothesis that disentangled learning with adversarial examples is the key for performance improvements, while other training recipes (e.g., paired clean and adversarial training samples, multi-step adversarial attackers) could be largely simplified. Our empirical results show that, compared to the vanilla training baseline, Fast AdvProp is able to further model performance on a spectrum of visual benchmarks, without incurring extra training cost. Additionally, our ablations find Fast AdvProp scales better if larger models are used, is compatible with existing data augmentation methods (i.e., Mixup and CutMix), and can be easily adapted to other recognition tasks like object detection. The code is available here: https://github.com/meijieru/fast_advprop.

show abstract

“…The robustness research of CNNs has experienced explosive development in recent years. Numerous works conduct thorough study on the robustness of CNNs and aim to strengthen it in different ways, e.g., stronger data augmentation [14,16,17], carefully designed [18,19] or searched [20,21] network architecture, improved training strategy [22][23][24], quantization [25] and pruning [26] of the weights, better pooling [27,28] or activation functions [29], etc. Although the methods mentioned above are good performed on CNNs, there is no evidence that they also keep the effectiveness on ViTs.…”

Section: Robustness Study For Cnnsmentioning

confidence: 99%

Towards Robust Vision Transformer

Mao¹,

Qi²,

Chen³

et al. 2021

Preprint

View full text Add to dashboard Cite

Recent advances on Vision Transformer (ViT) and its improved variants have shown that self-attention-based networks surpass traditional Convolutional Neural Networks (CNNs) in most vision tasks. However, existing ViTs focus on the standard accuracy and computation cost, lacking the investigation of the intrinsic influence on model robustness and generalization. In this work, we conduct systematic evaluation on components of ViTs in terms of their impact on robustness to adversarial examples, common corruptions and distribution shifts. We find some components can be harmful to robustness. By using and combining robust components as building blocks of ViTs, we propose Robust Vision Transformer (RVT), which is a new vision transformer and has superior performance with strong robustness. We further propose two new plug-and-play techniques called positionaware attention scaling and patch-wise augmentation to augment our RVT, which we abbreviate as RVT * . The experimental results on ImageNet and six robustness benchmarks show the advanced robustness and generalization ability of RVT compared with previous ViTs and state-of-the-art CNNs. Furthermore, RVT-S * also achieves Top-1 rank on multiple robustness leaderboards including ImageNet-C and ImageNet-Sketch. The code will be available at https://git.io/Jswdk.Preprint. Under review.

show abstract

Shape-Texture Debiased Neural Network Training

Cited by 15 publications

References 34 publications

A Comprehensive Survey of Image Augmentation Techniques for Deep Learning

A Comprehensive Survey of Image Augmentation Techniques for Deep Learning

Fast AdvProp

Towards Robust Vision Transformer

Contact Info

Product

Resources

About