Sparse and Transferable Universal Singular Vectors Attack
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913209079824384 |
|---|---|
| author | Kuvshinova, Kseniia Tsymboi, Olga Oseledets, Ivan |
| author_facet | Kuvshinova, Kseniia Tsymboi, Olga Oseledets, Ivan |
| contents | The research in the field of adversarial attacks and models' vulnerability is one of the fundamental directions in modern machine learning. Recent studies reveal the vulnerability phenomenon, and understanding the mechanisms behind this is essential for improving neural network characteristics and interpretability. In this paper, we propose a novel sparse universal white-box adversarial attack. Our approach is based on truncated power iteration providing sparsity to $(p,q)$-singular vectors of the hidden layers of Jacobian matrices. Using the ImageNet benchmark validation subset, we analyze the proposed method in various settings, achieving results comparable to dense baselines with more than a 50% fooling rate while damaging only 5% of pixels and utilizing 256 samples for perturbation fitting. We also show that our algorithm admits higher attack magnitude without affecting the human ability to solve the task. Furthermore, we investigate that the constructed perturbations are highly transferable among different models without significantly decreasing the fooling rate. Our findings demonstrate the vulnerability of state-of-the-art models to sparse attacks and highlight the importance of developing robust machine learning systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_14031 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Sparse and Transferable Universal Singular Vectors Attack Kuvshinova, Kseniia Tsymboi, Olga Oseledets, Ivan Machine Learning Cryptography and Security Computer Vision and Pattern Recognition The research in the field of adversarial attacks and models' vulnerability is one of the fundamental directions in modern machine learning. Recent studies reveal the vulnerability phenomenon, and understanding the mechanisms behind this is essential for improving neural network characteristics and interpretability. In this paper, we propose a novel sparse universal white-box adversarial attack. Our approach is based on truncated power iteration providing sparsity to $(p,q)$-singular vectors of the hidden layers of Jacobian matrices. Using the ImageNet benchmark validation subset, we analyze the proposed method in various settings, achieving results comparable to dense baselines with more than a 50% fooling rate while damaging only 5% of pixels and utilizing 256 samples for perturbation fitting. We also show that our algorithm admits higher attack magnitude without affecting the human ability to solve the task. Furthermore, we investigate that the constructed perturbations are highly transferable among different models without significantly decreasing the fooling rate. Our findings demonstrate the vulnerability of state-of-the-art models to sparse attacks and highlight the importance of developing robust machine learning systems. |
| title | Sparse and Transferable Universal Singular Vectors Attack |
| topic | Machine Learning Cryptography and Security Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2401.14031 |