Adversarial Vertex Mixup: Toward Better Adversarially Robust Generalization

被引:72
|
作者
Lee, Saehyung
Lee, Hyungyu
Yoon, Sungroh [1 ]
机构
[1] Seoul Natl Univ, Elect & Comp Engn, ASRI, INMC, Seoul 08826, South Korea
基金
新加坡国家研究基金会;
关键词
D O I
10.1109/CVPR42600.2020.00035
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Adversarial examples cause neural networks to produce incorrect outputs with high confidence. Although adversarial training is one of the most effective forms of defense against adversarial examples, unfortunately, a large gap exists between test accuracy and training accuracy in adversarial training. In this paper, we identify Adversarial Feature Overfitting (AFO), which may cause poor adversarially robust generalization, and we show that adversarial training can overshoot the optimal point in terms of robust generalization, leading to AFO in our simple Gaussian model. Considering these theoretical results, we present soft labeling as a solution to the AFO problem. Furthermore, we propose Adversarial Vertex mixup (AVmixup), a soft-labeled data augmentation approach for improving adversarially robust generalization. We complement our theoretical analysis with experiments on CIFAR10, CIFAR100, SVHN, and Tiny ImageNet, and show that AVmixup significantly improves the robust generalization performance and that it reduces the trade-off between standard accuracy and adversarial robustness.
引用
收藏
页码:269 / 278
页数:10
相关论文
共 50 条
  • [31] Signal Augmentation Method based on Mixing and Adversarial Training for Better Robustness and Generalization
    Zhang, Li
    Zhou, Gang
    Sun, Gangyin
    Wu, Chaopeng
    JOURNAL OF COMMUNICATIONS AND NETWORKS, 2024, 26 (06) : 679 - 688
  • [32] Reinforcement Based Learning on Classification Task Yields Better Generalization and Adversarial Accuracy
    Gupta, Shashi Kant
    THIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2021, 35 : 15793 - 15794
  • [33] Toward Enhanced Adversarial Robustness Generalization in Object Detection: Feature Disentangled Domain Adaptation for Adversarial Training
    Jung, Yoojin
    Song, Byung Cheol
    IEEE ACCESS, 2024, 12 : 179065 - 179076
  • [34] Better generalization with less data using robust gradient descent
    Holland, Matthew J.
    Ikeda, Kazushi
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 97, 2019, 97
  • [35] Adversarially Robust Models may not Transfer Better: Sufficient Conditions for Domain Transferability from the View of Regularization
    Xu, Xiaojun
    Zhang, Jacky Yibo
    Ma, Evelyn
    Son, Danny
    Koyejo, Oluwasanmi
    Li, Bo
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 162, 2022,
  • [36] Improving Generalization of Adversarial Training via Robust Critical Fine-Tuning
    Zhu, Kaijie
    Hu, Xixu
    Wang, Jindong
    Xie, Xing
    Yang, Ge
    2023 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION, ICCV, 2023, : 4401 - 4411
  • [37] Toward Robust Spiking Neural Network Against Adversarial Perturbation
    Liang, Ling
    Xu, Kaidi
    Hu, Xing
    Deng, Lei
    Xie, Yuan
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35 (NEURIPS 2022), 2022,
  • [38] Adversarially Enhanced Learning (AEL): Robust lightweight deep learning approach for radiology image classification against adversarial attacks
    Singh, Anshu
    Singh, Maheshwari Prasad
    Singh, Amit Kumar
    IMAGE AND VISION COMPUTING, 2025, 154
  • [39] Toward Statistical Model-Based Robust IMPT Planning: CrossValidation and Robust Generalization
    Zhang, Y.
    Mascia, A.
    MEDICAL PHYSICS, 2018, 45 (06) : E374 - E374
  • [40] THE CURSE OF OVERPARAMETRIZATION IN ADVERSARIAL TRAINING: PRECISE ANALYSIS OF ROBUST GENERALIZATION FOR RANDOM FEATURES REGRESSION
    Hassani, Hamed
    Javanmard, Adel
    ANNALS OF STATISTICS, 2024, 52 (02): : 441 - 465