Exploring Transferable and Robust Adversarial Perturbation Generation from the Perspective of Network Hierarchy

Ruikui Wang, Yuanfang Guo, Ruijie Yang, Yunhong WangarXiv 2021 · arXiv

Adversarial attacks pose a significant threat to the reliability of deep neural networks (DNNs). While many methods can successfully fool a specific "source" model, they often struggle with two critical challenges in real-world black-box settings: transferability (the ability to fool a different "target" model) and robustness (the ability to remain effective after image processing like resizing or noise).

This paper introduces TRAP (Transferable and Robust Adversarial Perturbation), a novel framework that addresses these challenges by exploiting the hierarchical nature of neural networks. Instead of focusing on the highly specialized output layers, TRAP targets the input and intermediate feature stages to generate more generalized and resilient perturbations.

The Network Hierarchy Perspective

Neural networks process information in stages. The authors argue that the final output stage is overly tuned to the source model's specific architecture and training data, leading to "overfitting" of the adversarial perturbation. In contrast:

  1. Intermediate Stages contain more general features that are shared across different architectures.
  2. Input Stages are where the perturbation is physically applied, making them the primary battleground for robustness against image transformations.

The TRAP Framework

TRAP combines two key mechanisms to optimize perturbations across these stages:

1. Dynamically Guided Mechanism (DGM)

Previous intermediate-level attacks (ILA) used a static direction to guide feature optimization. However, as the perturbation evolves, this fixed direction becomes suboptimal. DGM adaptively updates the directional guidance at each iteration. This ensures that the optimization process continuously pushes features toward a space that is likely to cause misclassification in any model, not just the source.

2. Affine-Invariant Perturbation Enhancement (AIM)

To ensure robustness, TRAP introduces AIM at the input stage. Unlike existing methods that use simple augmentations (like random resizing), AIM employs a multi-form affine transformation. This combines translation, rotation, scaling, and shearing into a single differentiable operation. By training the perturbation to be effective across this diverse set of transformations, the resulting adversarial images become significantly more robust to real-world image processing.

Figure 1: Visual comparison of attack transferability and robustness. (a) shows TRAP's superior transferability across different architectures. (b) demonstrates TRAP's robustness to affine transformations compared to baseline methods.

Experimental Evaluation and Results

The researchers evaluated TRAP on the ImageNet dataset using a variety of source and target models, including both standard architectures (Inception-v3, ResNet-101, etc.) and models specifically trained with adversarial defenses.

Superior Transferability

TRAP consistently outperformed state-of-the-art (SOTA) methods. When attacking from an Inception-v3 source, TRAP achieved an Attack Success Rate (ASR) of over 92% on other normally trained models. Even against robustly defended models, TRAP showed a marked improvement, nearly doubling the success rate of previous methods like ILA++.

Robustness to Transformations and Noise

The multi-form affine transformations in AIM proved highly effective. Adversarial images generated by TRAP maintained their effectiveness even when subjected to significant Gaussian noise or blurring. The "destruction rate"—the frequency with which an attack is neutralized by such processing—was significantly lower for TRAP than for its competitors.

Figure 5: Attack Success Rate (ASR) vs. optimization steps for various target models. The proposed DG-ILA++ (a component of TRAP) consistently achieves higher ASR across different architectures compared to baseline methods like TAP* and ILA++.

Conclusion

By shifting the focus from the output layer to a hierarchical approach involving intermediate feature guidance and multi-form input transformations, TRAP sets a new benchmark for black-box adversarial attacks. Its ability to generate perturbations that are both highly transferable and robust to physical-world variations highlights a critical vulnerability in current DNNs and provides a robust tool for evaluating model security.

Summary updated

Abstract

The transferability and robustness of adversarial examples are two practical yet important properties for black-box adversarial attacks. In this paper, we explore effective mechanisms to boost both of them from the perspective of network hierarchy, where a typical network can be hierarchically divided into output stage, intermediate stage and input stage. Since over-specialization of source model, we can hardly improve the transferability and robustness of the adversarial perturbations in the output stage. Therefore, we focus on the intermediate and input stages in this paper and propose a transferable and robust adversarial perturbation generation (TRAP) method. Specifically, we propose the dynamically guided mechanism to continuously calculate accurate directional guidances for perturbation generation in the intermediate stage. In the input stage, instead of the single-form transformation augmentations adopted in the existing methods, we leverage multiform affine transformation augmentations to further enrich the input diversity and boost the robustness and transferability of the adversarial perturbations. Extensive experiments demonstrate that our TRAP achieves impressive transferability and high robustness against certain interferences.
BibTeX
@misc{wang2021exploring,
  author = {Ruikui Wang and Yuanfang Guo and Ruijie Yang and Yunhong Wang},
  title = {{Exploring Transferable and Robust Adversarial Perturbation Generation from the Perspective of Network Hierarchy}},
  year = {2021},
  eprint = {2108.07033},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2108.07033},
}
Figure