SIFT-Graph: Benchmarking Multimodal Defense Against Image Adversarial Attacks With Robust Feature Graph
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917075475234816 |
|---|---|
| author | He, Jingjie Liang, Weijie Shan, Zihan Caesar, Matthew |
| author_facet | He, Jingjie Liang, Weijie Shan, Zihan Caesar, Matthew |
| contents | Adversarial attacks expose a fundamental vulnerability in modern deep vision models by exploiting their dependence on dense, pixel-level representations that are highly sensitive to imperceptible perturbations. Traditional defense strategies typically operate within this fragile pixel domain, lacking mechanisms to incorporate inherently robust visual features. In this work, we introduce SIFT-Graph, a multimodal defense framework that enhances the robustness of traditional vision models by aggregating structurally meaningful features extracted from raw images using both handcrafted and learned modalities. Specifically, we integrate Scale-Invariant Feature Transform keypoints with a Graph Attention Network to capture scale and rotation invariant local structures that are resilient to perturbations. These robust feature embeddings are then fused with traditional vision model, such as Vision Transformer and Convolutional Neural Network, to form a unified, structure-aware and perturbation defensive model. Preliminary results demonstrate that our method effectively improves the visual model robustness against gradient-based white box adversarial attacks, while incurring only a marginal drop in clean accuracy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_08810 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SIFT-Graph: Benchmarking Multimodal Defense Against Image Adversarial Attacks With Robust Feature Graph He, Jingjie Liang, Weijie Shan, Zihan Caesar, Matthew Computer Vision and Pattern Recognition Adversarial attacks expose a fundamental vulnerability in modern deep vision models by exploiting their dependence on dense, pixel-level representations that are highly sensitive to imperceptible perturbations. Traditional defense strategies typically operate within this fragile pixel domain, lacking mechanisms to incorporate inherently robust visual features. In this work, we introduce SIFT-Graph, a multimodal defense framework that enhances the robustness of traditional vision models by aggregating structurally meaningful features extracted from raw images using both handcrafted and learned modalities. Specifically, we integrate Scale-Invariant Feature Transform keypoints with a Graph Attention Network to capture scale and rotation invariant local structures that are resilient to perturbations. These robust feature embeddings are then fused with traditional vision model, such as Vision Transformer and Convolutional Neural Network, to form a unified, structure-aware and perturbation defensive model. Preliminary results demonstrate that our method effectively improves the visual model robustness against gradient-based white box adversarial attacks, while incurring only a marginal drop in clean accuracy. |
| title | SIFT-Graph: Benchmarking Multimodal Defense Against Image Adversarial Attacks With Robust Feature Graph |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2511.08810 |