A Curious Case of Remarkable Resilience to Gradient Attacks via Fully Convolutional and Differentiable Front End with a Skip Connection

• URL: http://arxiv.org/abs/2402.17018v2
• Date: Fri, 22 Aug 2025 17:26:54 GMT
• Title: A Curious Case of Remarkable Resilience to Gradient Attacks via Fully Convolutional and Differentiable Front End with a Skip Connection
• Authors: Leonid Boytsov, Ameya Joshi, Filipe Condessa,
• Abstract summary: A differentiable and fully convolutional model with a skip connection is added before a frozen backbone classifier.<n>By training such composite models using a small learning rate, we obtained models that retained the accuracy of the backbone.<n>We estimate that these ensembles achieve near-SOTA AutoAttack accuracy on CIFAR10, CIFAR100, and ImageNet.
• Score: 2.409321776852724
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We experimented with front-end enhanced neural models where a differentiable and fully convolutional model with a skip connection is added before a frozen backbone classifier. By training such composite models using a small learning rate for about one epoch, we obtained models that retained the accuracy of the backbone classifier while being unusually resistant to gradient attacks-including APGD and FAB-T attacks from the AutoAttack package-which we attribute to gradient masking. Although gradient masking is not new, the degree we observe is striking for fully differentiable models without obvious gradient-shattering-e.g., JPEG compression-or gradient-diminishing components. The training recipe to produce such models is also remarkably stable and reproducible: We applied it to three datasets (CIFAR10, CIFAR100, and ImageNet) and several modern architectures (including vision Transformers) without a single failure case. While black-box attacks such as the SQUARE attack and zero-order PGD can partially overcome gradient masking, these attacks are easily defeated by simple randomized ensembles. We estimate that these ensembles achieve near-SOTA AutoAttack accuracy on CIFAR10, CIFAR100, and ImageNet (while retaining almost all clean accuracy of the original classifiers) despite having near-zero accuracy under adaptive attacks. Adversarially training the backbone further amplifies this front-end "robustness". On CIFAR10, the respective randomized ensemble achieved 90.8$\pm 2.5\%$ (99\% CI) accuracy under the full AutoAttack while having only 18.2$\pm 3.6\%$ accuracy under the adaptive attack ($\varepsilon=8/255$, $L^\infty$ norm). We conclude the paper with a discussion of whether randomized ensembling can serve as a practical defense. Code and instructions to reproduce key results are available. https://github.com/searchivarius/curious_case_of_gradient_masking

Related papers

• LISArD: Learning Image Similarity to Defend Against Gray-box Adversarial Attacks [13.154512864498912]
  Adversarial Training (AT) and Adversarial Distillation (AD) include adversarial examples during the training phase. This paper considers an even more realistic evaluation scenario: gray-box attacks, which assume that the attacker knows the architecture and the dataset used to train the target network, but cannot access its gradients. We provide empirical evidence that models are vulnerable to gray-box attacks and propose LISArD, a defense mechanism that does not increase computational and temporal costs but provides robustness against gray-box and white-box attacks without including AT.
  arXiv  Detail & Related papers  (2025-02-27T22:02:06Z)
• Model Inversion Attacks Through Target-Specific Conditional Diffusion Models [54.69008212790426]
  Model inversion attacks (MIAs) aim to reconstruct private images from a target classifier's training set, thereby raising privacy concerns in AI applications. Previous GAN-based MIAs tend to suffer from inferior generative fidelity due to GAN's inherent flaws and biased optimization within latent space. We propose Diffusion-based Model Inversion (Diff-MI) attacks to alleviate these issues.
  arXiv  Detail & Related papers  (2024-07-16T06:38:49Z)
• DiffAttack: Evasion Attacks Against Diffusion-Based Adversarial Purification [63.65630243675792]
  Diffusion-based purification defenses leverage diffusion models to remove crafted perturbations of adversarial examples. Recent studies show that even advanced attacks cannot break such defenses effectively. We propose a unified framework DiffAttack to perform effective and efficient attacks against diffusion-based purification defenses.
  arXiv  Detail & Related papers  (2023-10-27T15:17:50Z)
• White-box Membership Inference Attacks against Diffusion Models [13.425726946466423]
  Diffusion models have begun to overshadow GANs in industrial applications due to their superior image generation performance. We aim to design membership inference attacks (MIAs) catered to diffusion models. We first conduct an exhaustive analysis of existing MIAs on diffusion models, taking into account factors such as black-box/white-box models and the selection of attack features. We found that white-box attacks are highly applicable in real-world scenarios, and the most effective attacks presently are white-box.
  arXiv  Detail & Related papers  (2023-08-11T22:03:36Z)
• Carefully Blending Adversarial Training and Purification Improves Adversarial Robustness [1.2289361708127877]
  CARSO is able to defend itself against adaptive end-to-end white-box attacks devised for defences. Our method improves by a significant margin the state-of-the-art for CIFAR-10, CIFAR-100, and TinyImageNet-200.
  arXiv  Detail & Related papers  (2023-05-25T09:04:31Z)
• Class-Conditioned Transformation for Enhanced Robust Image Classification [19.738635819545554]
  We propose a novel test-time threat model algorithm that enhances Adrial-versa-Trained (AT) models. Our method operates through COnditional image transformation and DIstance-based Prediction (CODIP) The proposed method achieves state-of-the-art results demonstrated through extensive experiments on various models, AT methods, datasets, and attack types.
  arXiv  Detail & Related papers  (2023-03-27T17:28:20Z)
• Detection and Mitigation of Byzantine Attacks in Distributed Training [24.951227624475443]
  An abnormal Byzantine behavior of the worker nodes can derail the training and compromise the quality of the inference. Recent work considers a wide range of attack models and has explored robust aggregation and/or computational redundancy to correct the distorted gradients. In this work, we consider attack models ranging from strong ones: $q$ omniscient adversaries with full knowledge of the defense protocol that can change from iteration to iteration to weak ones: $q$ randomly chosen adversaries with limited collusion abilities.
  arXiv  Detail & Related papers  (2022-08-17T05:49:52Z)
• Adversarially Robust Classification by Conditional Generative Model Inversion [4.913248451323163]
  We propose a classification model that does not obfuscate gradients and is robust by construction without assuming prior knowledge about the attack. Our method casts classification as an optimization problem where we "invert" a conditional generator trained on unperturbed, natural images. We demonstrate that our model is extremely robust against black-box attacks and has improved robustness against white-box attacks.
  arXiv  Detail & Related papers  (2022-01-12T23:11:16Z)
• Stochastic Variance Reduced Ensemble Adversarial Attack for Boosting the Adversarial Transferability [20.255708227671573]
  Black-box adversarial attacks can be transferred from one model to another. In this work, we propose a novel ensemble attack method called the variance reduced ensemble attack. Empirical results on the standard ImageNet demonstrate that the proposed method could boost the adversarial transferability and outperforms existing ensemble attacks significantly.
  arXiv  Detail & Related papers  (2021-11-21T06:33:27Z)
• Meta Gradient Adversarial Attack [64.5070788261061]
  This paper proposes a novel architecture called Metaversa Gradient Adrial Attack (MGAA), which is plug-and-play and can be integrated with any existing gradient-based attack method. Specifically, we randomly sample multiple models from a model zoo to compose different tasks and iteratively simulate a white-box attack and a black-box attack in each task. By narrowing the gap between the gradient directions in white-box and black-box attacks, the transferability of adversarial examples on the black-box setting can be improved.
  arXiv  Detail & Related papers  (2021-08-09T17:44:19Z)
• Sparse and Imperceptible Adversarial Attack via a Homotopy Algorithm [93.80082636284922]
  Sparse adversarial attacks can fool deep networks (DNNs) by only perturbing a few pixels. Recent efforts combine it with another l_infty perturbation on magnitudes. We propose a homotopy algorithm to tackle the sparsity and neural perturbation framework.
  arXiv  Detail & Related papers  (2021-06-10T20:11:36Z)
• Adaptive Feature Alignment for Adversarial Training [56.17654691470554]
  CNNs are typically vulnerable to adversarial attacks, which pose a threat to security-sensitive applications. We propose the adaptive feature alignment (AFA) to generate features of arbitrary attacking strengths. Our method is trained to automatically align features of arbitrary attacking strength.
  arXiv  Detail & Related papers  (2021-05-31T17:01:05Z)
• Adversarial Robustness by Design through Analog Computing and Synthetic Gradients [80.60080084042666]
  We propose a new defense mechanism against adversarial attacks inspired by an optical co-processor. In the white-box setting, our defense works by obfuscating the parameters of the random projection. We find the combination of a random projection and binarization in the optical system also improves robustness against various types of black-box attacks.
  arXiv  Detail & Related papers  (2021-01-06T16:15:29Z)
• Learnable Boundary Guided Adversarial Training [66.57846365425598]
  We use the model logits from one clean model to guide learning of another one robust model. We achieve new state-of-the-art robustness on CIFAR-100 without additional real or synthetic data.
  arXiv  Detail & Related papers  (2020-11-23T01:36:05Z)
• Patch-wise Attack for Fooling Deep Neural Network [153.59832333877543]
  We propose a patch-wise iterative algorithm -- a black-box attack towards mainstream normally trained and defense models. We significantly improve the success rate by 9.2% for defense models and 3.7% for normally trained models on average.
  arXiv  Detail & Related papers  (2020-07-14T01:50:22Z)
• Orthogonal Deep Models As Defense Against Black-Box Attacks [71.23669614195195]
  We study the inherent weakness of deep models in black-box settings where the attacker may develop the attack using a model similar to the targeted model. We introduce a novel gradient regularization scheme that encourages the internal representation of a deep model to be orthogonal to another. We verify the effectiveness of our technique on a variety of large-scale models.
  arXiv  Detail & Related papers  (2020-06-26T08:29:05Z)
• Towards Query-Efficient Black-Box Adversary with Zeroth-Order Natural Gradient Descent [92.4348499398224]
  Black-box adversarial attack methods have received special attentions owing to their practicality and simplicity. We propose a zeroth-order natural gradient descent (ZO-NGD) method to design the adversarial attacks. ZO-NGD can obtain significantly lower model query complexities compared with state-of-the-art attack methods.
  arXiv  Detail & Related papers  (2020-02-18T21:48:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.