Research map / Adversarial robustness and security / Black-box and transfer attacks
Adversarial Eigen Attack on Black-Box Models
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use Jacobian eigenvector directions from the fixed pre-trained model instead of DCT directions | EigenBA · ImageNet validation, untargeted attack | Avg. queries (all): 1056 -> 518 | Helped |
| use Jacobian eigenvector directions from the fixed pre-trained model instead of DCT directions | EigenBA · ImageNet validation, targeted attack | Avg. queries (all): 5437 -> 4140 | Helped |
| use Jacobian eigenvector directions rather than randomly chosen representation basis directions | EigenBA · ImageNet validation, untargeted attack | Avg. queries (all): 843 -> 518 | Helped |
| use Jacobian eigenvector directions rather than randomly chosen representation basis directions | EigenBA · CIFAR-10 test, targeted attack | Avg. queries (all): 419 -> 244 | Helped |
| reserve 1.0 of the pre-trained model parameters (ablation) | EigenBA · CIFAR-10 test | Avg. queries (all): 440 -> 88 | Helped |
| reserve 0.9 of the pre-trained model parameters (ablation) | EigenBA · CIFAR-10 test | Avg. queries (all): 440 -> 85 | Helped |
| reserve 0.8 of the pre-trained model parameters (ablation) | EigenBA · CIFAR-10 test | Avg. queries (all): 440 -> 130 | Helped |
| reserve 0.7 of the pre-trained model parameters (ablation) | EigenBA · CIFAR-10 test | Avg. queries (all): 440 -> 195 | Helped |
| reserve 0.6 of the pre-trained model parameters (ablation) | EigenBA · CIFAR-10 test | Avg. queries (all): 440 -> 382 | No clear effect |
| reserve 0.5 of the pre-trained model parameters (ablation) | EigenBA · CIFAR-10 test | Avg. queries (all): 440 -> 700 | Hurt |
Closest papers by meaning
- A Frank-Wolfe Framework for Efficient and Effective Adversarial Attacks (AAAI 2020 · similarity 0.90)
- Black-Box Adversarial Attack with Transferable Model-based Embedding (ICLR 2020 · similarity 0.89)
- Efficient Black-box Adversarial Attacks via Bayesian Optimization Guided by a Function Prior (ICML 2024 · similarity 0.87)
- BRUSLEATTACK: A QUERY-EFFICIENT SCORE- BASED BLACK-BOX SPARSE ADVERSARIAL ATTACK (ICLR 2024 · similarity 0.86)
- DifAttack: Query-Efficient Black-Box Adversarial Attack via Disentangled Feature Space (AAAI 2024 · similarity 0.86)
- ADBA: Approximation Decision Boundary Approach for Black-Box Adversarial Attacks (AAAI 2025 · similarity 0.85)
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.