Research map / Adversarial robustness and security / Certified and empirical robustness
Improving Adversarial Robustness via Guided Complement Entropy
Hao-Yun Chen, Jhao-Hong Liang, Shih-Chieh Chang, Jia-Yu Pan, Yu-Ting Chen, Wei Wei, Da-Cheng Juan
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 |
|---|---|---|---|
| replace cross-entropy with GCE (alpha=1/3) | ResNet-50 · Tiny ImageNet | error rate (%): 39.54 -> 38.56 | Improved |
| replace cross-entropy with GCE (alpha=1/3) | LeNet-5 · MNIST | error rate (%): 0.8 -> 0.67 | Improved |
| replace cross-entropy with GCE (alpha=1/3) | ResNet-56 · CIFAR10 | error rate (%): 7.99 -> 7.18 | Improved |
| replace cross-entropy with GCE (alpha=1/3) | ResNet-56 · CIFAR100 | error rate (%): 31.9 -> 31.75 | No clear effect |
| use guided factor alpha=1/2 instead of smaller alpha (ablation) | GCE · CIFAR100 | error rate (%): 31.75 -> 40.59 | Hurt |
| use guided factor alpha=1/2 instead of smaller alpha (ablation) | GCE · CIFAR10 | error rate (%): 7.18 -> 9.18 | Hurt |
| train with GCE instead of XE, FGSM attack | LeNet-5 · MNIST FGSM eps=0.2 | accuracy under attack (%): 38.88 -> 62.74 | Improved |
| train with GCE instead of XE, BIM attack | LeNet-5 · MNIST BIM eps=0.2 | accuracy under attack (%): 2.15 -> 34.49 | Improved |
| train with GCE instead of XE, MIM attack | LeNet-5 · MNIST MIM eps=0.2 | accuracy under attack (%): 2.29 -> 39.81 | Improved |
| train with GCE instead of XE, PGD attack | ResNet-56 · CIFAR10 PGD eps=0.04 | accuracy under attack (%): 0.0 -> 5.91 | Improved |
| train with GCE instead of XE, PGD attack | ResNet-50 · Tiny ImageNet PGD eps=0.04 | accuracy under attack (%): 0.0 -> 9.52 | Improved |
| use GCE instead of COT, BIM attack | ResNet-56 · CIFAR10 BIM eps=0.04 | accuracy under attack (%): 7.49 -> 19.59 | Improved |
Closest papers by meaning
- Certified Adversarial Robustness with Additive Noise (NeurIPS 2019 · similarity 0.89)
- Improving Generalization of Adversarial Training via Robust Critical Fine-Tuning (ICCV 2023 · similarity 0.89)
- A Combination of Noise and Bilateral Filters Achieve Supralinear and Scalable Adversarial Robustness in CNNs (CVPR 2026 · similarity 0.89)
- l∞-Robustness and Beyond: Unleashing Efficient Adversarial Training (ECCV 2022 · similarity 0.89)
- Learning Better Certified Models from Empirically-Robust Teachers (NeurIPS 2026 · similarity 0.88)
- Out of Thin Air: Exploring Data-Free Adversarial Robustness Distillation (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.
