Research map / Adversarial robustness and security / Black-box and transfer attacks
KOEnsAttack: Towards Efficient Data-Free Black-Box Adversarial Attacks via Knowledge-Orthogonalized Substitute Ensembles
Chaoyong Yang, Jia-Li Yin, Bin Chen, Zhaozhe Hu, Xiaolei Liu, Wei Lin
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 |
|---|---|---|---|
| add sample hardness enhancement (SHE) to baseline (ablation) | KOEnsAttack baseline · CIFAR-10 VGG-19 targeted | attack success rate (%): 49.00 -> 93.74 | Improved |
| add knowledge orthogonalization module (KOM) to baseline (ablation) | KOEnsAttack baseline · CIFAR-100 ResNet-34 targeted | attack success rate (%): 29.82 -> 36.60 | Improved |
| add KOM on top of SHE (ablation) | KOEnsAttack · Tiny ImageNet ResNet-50 untargeted | attack success rate (%): 64.54 -> 78.23 | Improved |
| add SHE on top of KOM (ablation) | KOEnsAttack · SVHN ResNet-18 untargeted | attack success rate (%): 94.94 -> 97.41 | Improved |
| use KOEnsAttack instead of DisGUIDE | data-free black-box attack · CIFAR-100 ResNet-34 untargeted BIM | attack success rate (%): 97.10 -> 98.12 | Improved |
| increase SHE iteration steps T (ablation) | KOEnsAttack · CIFAR-10 | attack success rate (%) | Improved |
| vary KOM weight lambda (ablation) | KOEnsAttack · CIFAR-10 | attack success rate (%) | No clear effect |
| use hard-label KOEnsAttack against DFME which uses soft probabilities | KOEnsAttack · Tiny ImageNet ResNet-50 untargeted PGD | attack success rate (%): 57.58 -> 78.23 | 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.
- DisGUIDE — DisGUIDE: Disagreement-Guided Data-Free Model Extraction (AAAI 2023) · better on 2 of 2 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
- Delving into Data: Effectively Substitute Training for Black-box Attack (CVPR 2021 · similarity 0.88)
- DaST: Data-Free Substitute Training for Adversarial Attacks (CVPR 2020 · similarity 0.86)
- AGS: Affordable and Generalizable Substitute Training for Transferable Adversarial Attack (AAAI 2024 · similarity 0.85)
- AUTE: Peer-Alignment and Self-Unlearning Boost Adversarial Robustness for Training Ensemble Models (AAAI 2025 · similarity 0.84)
- Unlearn and Burn: Adversarial Machine Unlearning Requests Destroy Model Accuracy (ICLR 2025 · similarity 0.84)
- Adversarial Machine Unlearning (ICLR 2025 · similarity 0.84)
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.
