Research map / Adversarial robustness and security / Black-box and transfer attacks
TRS: Transferability Reduced Ensemble via Promoting Gradient Diversity and Model Smoothness
Zhuolin Yang, Linyi Li, Xiaojun Xu, Shiliang Zuo, Qian Chen, Pan Zhou, Benjamin I. P. Rubinstein, Ce Zhang, Bo Li
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 |
|---|---|---|---|
| Jointly reduce inter-model loss-gradient similarity and enforce neighborhood smoothness | TRS ensemble · MNIST, PGD (50), ε=0.15 | robust accuracy (%): 47.7 -> 85.1 | Improved |
| Jointly reduce inter-model loss-gradient similarity and enforce neighborhood smoothness | TRS ensemble · CIFAR-10, PGD (50), ε=0.01 | robust accuracy (%): 37.1 -> 50.5 | Improved |
| Jointly reduce inter-model loss-gradient similarity and enforce neighborhood smoothness | TRS ensemble · MNIST, FGSM, ε=0.1 | robust accuracy (%): 95.8 -> 95.6 | Mixed |
| Jointly reduce inter-model loss-gradient similarity and enforce neighborhood smoothness | TRS ensemble · CIFAR-10, FGSM, ε=0.02 | robust accuracy (%): 63.8 -> 44.2 | Hurt |
| Add the model-smoothness loss to cosine-similarity regularization (ablation) | TRS ensemble · MNIST, PGD (50), ε=0.1 | robust accuracy (%): 0.0 -> 93.0 | Improved |
| Add the model-smoothness loss to cosine-similarity regularization (ablation) | TRS ensemble · CIFAR-10, PGD (50), ε=0.01 | robust accuracy (%): 0.0 -> 50.5 | Improved |
| Use min-max neighborhood smoothing instead of pointwise gradient ℓ2 regularization (ablation) | TRS ensemble · MNIST, PGD (50), ε=0.1 | robust accuracy (%): 73.4 -> 93.0 | Improved |
| Use min-max neighborhood smoothing instead of pointwise gradient ℓ2 regularization (ablation) | TRS ensemble · CIFAR-10, PGD (50), ε=0.01 | robust accuracy (%): 15.7 -> 50.5 | Improved |
| Enforce model smoothness in addition to gradient dissimilarity | Ensemble base models · MNIST and CIFAR-10, PGD transferability | attack success rate | Improved |
Closest papers by meaning
- Understanding Model Ensemble in Transferable Adversarial Attack (ICML 2025 · similarity 0.93)
- Rethinking Model Ensemble in Transfer-based Adversarial Attacks (ICLR 2024 · similarity 0.92)
- An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial Transferability (ICCV 2023 · similarity 0.91)
- Transferability Bound Theory: Exploring Relationship between Adversarial Transferability and Flatness (NeurIPS 2024 · similarity 0.90)
- Enhancing Adversarial Transferability with Adversarial Weight Tuning (AAAI 2025 · similarity 0.89)
- LRS: Enhancing Adversarial Transferability through Lipschitz Regularized Surrogate (AAAI 2024 · similarity 0.88)
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.
