Research map / Adversarial robustness and security / Black-box and transfer attacks
Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization
Read the paperCheck 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 |
|---|---|---|---|
| discrete surrogate (set-function local search with lazy greedy, no gradient estimation) replacing gradient-estimation black-box attacks | black-box ℓ∞ attack · Cifar-10 adversarially trained WRN, untargeted | success rate: 48.0% vs 38.6% for Bandits | Helped |
| discrete surrogate with accelerated local search | black-box ℓ∞ attack · Cifar-10 adversarially trained WRN, untargeted | avg queries: 1261 vs 1877 for Bandits | Helped |
| discrete surrogate with accelerated local search | black-box ℓ∞ attack · ImageNet untargeted eps=0.05 | success rate: 98.5% vs 94.9% for Bandits | Helped |
| discrete surrogate with accelerated local search | black-box ℓ∞ attack · ImageNet untargeted eps=0.05 | avg queries: 722 vs 1030 for Bandits | Helped |
| discrete surrogate with accelerated local search | black-box ℓ∞ attack · ImageNet targeted eps=0.05 | avg queries: 7485 vs 16284 for NES (about 55% fewer) | Helped |
| discrete surrogate with accelerated local search | black-box ℓ∞ attack · ImageNet targeted eps=0.05 | success rate: 99.9% vs 99.7% for NES | Helped |
| discrete surrogate with accelerated local search | black-box ℓ∞ attack · ImageNet untargeted eps=0.01 and 0.03 | success rate: margin over Bandits up to 10% at smaller eps (81.3% vs 72.4% at eps=0.01) | Helped |
| replacing gradient-estimation hyperparameter tuning with the discrete surrogate (only block size k) (ablation) | black-box ℓ∞ attack · ImageNet untargeted hyperparameter sensitivity | success rate variance | Helped |
Closest papers by meaning
- Efficient Black-box Adversarial Attacks via Bayesian Optimization Guided by a Function Prior (ICML 2024 · similarity 0.90)
- Blackbox Attacks via Surrogate Ensemble Search (NeurIPS 2022 · similarity 0.89)
- Projection & Probability-Driven Black-Box Attack (CVPR 2020 · similarity 0.89)
- Simple Black-box Adversarial Attacks (ICML 2019 · similarity 0.89)
- Understanding the Robustness of Randomized Feature Defense Against Query-Based Adversarial Attacks (ICLR 2024 · similarity 0.88)
- BRUSLEATTACK: A QUERY-EFFICIENT SCORE- BASED BLACK-BOX SPARSE ADVERSARIAL ATTACK (ICLR 2024 · similarity 0.86)
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.