Research map / Adversarial robustness and security / Black-box and transfer attacks
Prior Convictions: Black-box Adversarial Attacks with Bandits and Priors
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 |
|---|---|---|---|
| add the time prior to NES-style gradient estimation (ablation) | Bandits_T · ImageNet Inception v3 ℓ∞ | average queries on successful attacks: 1735 -> 1781 | Hurt |
| add the time prior to NES-style gradient estimation (ablation) | Bandits_T · ImageNet Inception v3 ℓ∞ | failure rate: 22.2% -> 11.6% | Helped |
| add the time prior to NES-style gradient estimation (ablation) | Bandits_T · ImageNet Inception v3 ℓ2 | average queries on successful attacks: 2938 -> 2690 | Helped |
| add the time prior to NES-style gradient estimation (ablation) | Bandits_T · ImageNet Inception v3 ℓ2 | failure rate: 34.4% -> 30.4% | Helped |
| add the time and data-dependent priors (ablation) | Bandits_TD · ImageNet Inception v3 ℓ∞ | average queries on successful attacks: 1735 -> 1117 | Helped |
| add the time and data-dependent priors (ablation) | Bandits_TD · ImageNet Inception v3 ℓ∞ | failure rate: 22.2% -> 4.6% | Helped |
| add the time and data-dependent priors (ablation) | Bandits_TD · ImageNet Inception v3 ℓ2 | average queries on successful attacks: 2938 -> 1858 | Helped |
| add the time and data-dependent priors (ablation) | Bandits_TD · ImageNet Inception v3 ℓ2 | failure rate: 34.4% -> 15.5% | Helped |
| carry gradient information forward across attack steps (ablation) | Bandits_T · ImageNet Inception v3 ℓ∞ | average queries on successful attacks: 1735 -> 1781 | Mixed |
Closest papers by meaning
- Improving Black-box Adversarial Attacks with a Transfer-based Prior (NeurIPS 2019 · similarity 0.92)
- Efficient Black-box Adversarial Attacks via Bayesian Optimization Guided by a Function Prior (ICML 2024 · similarity 0.91)
- BayesOpt Adversarial Attack (ICLR 2020 · similarity 0.90)
- Understanding the Robustness of Randomized Feature Defense Against Query-Based Adversarial Attacks (ICLR 2024 · similarity 0.89)
- Mind the Gap: Detecting Black-box Adversarial Attacks in the Making through Query Update Analysis (CVPR 2025 · similarity 0.88)
- BRUSLEATTACK: A QUERY-EFFICIENT SCORE- BASED BLACK-BOX SPARSE ADVERSARIAL ATTACK (ICLR 2024 · similarity 0.87)
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.