Research map / Adversarial robustness and security / Attacks on text and graphs
Open-set Adversarial Defense
Rui Shao, Pramuditha Perera, Pong C. Yuen, Vishal M. Patel
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 feature-denoising layers to the encoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | AUC-ROC: 66.1 -> 68.5 | Improved |
| add feature-denoising layers to the encoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | closed-set accuracy: 69.9 -> 70.4 | Improved |
| add decoder to the encoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | AUC-ROC: 66.1 -> 67.3 | Improved |
| add decoder to the encoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | closed-set accuracy: 69.9 -> 68.8 | Hurt |
| add denoising layers to encoder+decoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | AUC-ROC: 67.3 -> 68.2 | Improved |
| add denoising layers to encoder+decoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | closed-set accuracy: 68.8 -> 70.6 | Improved |
| add self-supervised denoising to encoder+decoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | AUC-ROC: 67.3 -> 68.5 | Improved |
| add self-supervised denoising to encoder+decoder (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | closed-set accuracy: 68.8 -> 71.9 | Improved |
| add self-supervised denoising on top of encoder+decoder+denoising layers (full OSDN) (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | AUC-ROC: 68.2 -> 69.6 | Improved |
| add self-supervised denoising on top of encoder+decoder+denoising layers (full OSDN) (ablation) | OSDN encoder · CIFAR-10 ablation (open-set) | closed-set accuracy: 70.6 -> 72.8 | Improved |
| use OSDN instead of feature denoising | OSDN encoder · SVHN (FGSM) | AUC-ROC: 64.5 -> 71.4 | Improved |
| use OSDN instead of adversarial training | OSDN encoder · TinyImageNet (PGD) | closed-set accuracy: 40.3 -> 41.6 | Improved |
Closest papers by meaning
- Open-Set Adversarial Defense with Clean-Adversarial Mutual Learning (IJCV 2022 · similarity 0.97)
- Learning Network Architecture for Open-Set Recognition (AAAI 2022 · similarity 0.88)
- OpenGAN: Open-Set Recognition via Open Data Generation (ICCV 2021 · similarity 0.87)
- Deep Defense: Training DNNs with Improved Adversarial Robustness (NeurIPS 2018 · similarity 0.86)
- Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models (ICLR 2018 · similarity 0.86)
- Operational Open-Set Recognition and PostMax Refinement (ECCV 2024 · similarity 0.86)
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.
