Research map / Adversarial robustness and security / Certified and empirical robustness
Sparse Adversarial Attack via Perturbation Factorization
Yanbo Fan, Baoyuan Wu, Tuanhui Li, Yong Zhang, Mingyang Li, Zhifeng Li, Yujiu Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| jointly optimize perturbation magnitude and binary selection factors (SAPF) instead of C&W-l0 heuristic position selection | sparse targeted attack · CIFAR-10 average case | l2-norm: 0.805 -> 0.435 | Improved |
| SAPF instead of StrAttack | sparse targeted attack · CIFAR-10 average case | l1-norm: 9.494 -> 8.513 | Improved |
| SAPF instead of C&W-l0 | sparse targeted attack · ImageNet average case | l0-norm: 70885 -> 56218 | Improved |
| SAPF instead of StrAttack | sparse targeted attack · ImageNet average case | l2-norm: 0.974 -> 0.719 | Improved |
| SAPF instead of SparseFool | sparse targeted attack · ImageNet worst case | ASR (%): 46 -> 100 | Improved |
| SparseFool instead of SAPF (lower linf) | sparse targeted attack · CIFAR-10 average case | linf-norm: 0.064 -> 0.047 | Hurt |
| increase cardinality k (number of perturbed positions) (ablation) | SAPF · CIFAR-10 and ImageNet | l1/l2/linf-norm | Mixed |
| add group-wise sparsity (SAPF-GS, lambda2=10 vs 0) (ablation) | SAPF · ImageNet examples (350 super-pixel regions) | perturbation locality and norms | Mixed |
| use variance-based pixel weights instead of uniform weights (SAPF-VI) (ablation) | SAPF · ImageNet example (conch to fountain) | perturbed positions location | Improved |
| add brightness (variance-mean weight) to variance weight (ablation) | SAPF-VI · ImageNet example (conch to fountain) | perturbed positions location | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SparseFool — SparseFool: A Few Pixels Make a Big Difference (CVPR 2019) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Sparse and Imperceivable Adversarial Attacks (ICCV 2019 · similarity 0.93)
- Transferable Structural Sparse Adversarial Attack Via Exact Group Sparsity Training (CVPR 2024 · similarity 0.91)
- GSE: Group-wise Sparse and Explainable Adversarial Attacks (ICLR 2025 · similarity 0.90)
- Towards Efficient Training and Evaluation of Robust Models against $l_0$ Bounded Adversarial Perturbations (ICML 2024 · similarity 0.89)
- Adversarial Perturbations Are Formed by Iteratively Learning Linear Combinations of the Right Singular Vectors of the Adversarial Jacobian (ICML 2025 · similarity 0.89)
- BRUSLEATTACK: A QUERY-EFFICIENT SCORE- BASED BLACK-BOX SPARSE ADVERSARIAL ATTACK (ICLR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
