Research map / Adversarial robustness and security / Black-box and transfer attacks
Adversarial Defence by Diversified Simultaneous Training of Deep Ensembles
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 |
|---|---|---|---|
| Apply PDD+DEG simultaneous training instead of baseline independent training | ResNet-18 ensemble · CIFAR-100, white-box C&W (L2=0.01) | recognition accuracy (%): 0.38 -> 53.63 | Helped |
| Apply PDD+DEG simultaneous training instead of ADP | ResNet-18 ensemble · Tiny-ImageNet, white-box PGD (ε=0.02) | recognition accuracy (%): 3.81 -> 17.49 | Helped |
| Apply PDD+DEG simultaneous training instead of ADP | ResNet-18 ensemble · CIFAR-100, adaptive black-box PGD (ε=0.02) | recognition accuracy (%): 27.75 -> 77.60 | Helped |
| Add PDD diversified dropout (ablation) | Deep ensemble training · CIFAR-100, white-box PGD (ε=0.02) | recognition accuracy (%): 22.43 | Helped |
| Add DEG gradient dispersion without PDD (ablation) | Deep ensemble training · CIFAR-100, white-box PGD (ε=0.02) | recognition accuracy (%): 2.45 | No clear effect |
| Combine PDD with DEG rather than using either alone (ablation) | Deep ensemble training · CIFAR-100, white-box BIM, PGD, and JSMA | recognition accuracy (%): 21.76; 2.45; 30.90; 22.43; 2.45; 29.88; 34.73; 18.41; 45.74 | Helped |
| Add PDD to adversarial training (ablation) | AdvTFGSM ensemble · CIFAR-100, white-box PGD (ε=0.04) | recognition accuracy (%): 3.62 -> 19.5 | Helped |
| Add PDD to adversarial training (ablation) | AdvTPGD ensemble · CIFAR-100, white-box PGD (ε=0.04) | recognition accuracy (%): 15.69 -> 26.13 | Helped |
Closest papers by meaning
- Improving Adversarial Robustness via Promoting Ensemble Diversity (ICML 2019 · similarity 0.92)
- Ensemble Adversarial Training: Attacks and Defenses (ICLR 2018 · similarity 0.91)
- On the Diversity of Adversarial Ensemble Learning (ICML 2025 · similarity 0.90)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.88)
- Enhancing Adversarial Transferability with Checkpoints of a Single Model's Training (CVPR 2025 · similarity 0.88)
- Two is Better than One: Efficient Ensemble Defense for Robust and Compact Models (CVPR 2025 · 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.