Research map / Adversarial robustness and security / Certified and empirical robustness
Adversarial Robustness of Implicit Neural Representation-Based Classifiers
Jayoung Kim, Kookjin Lee, Noseong Park, Sanghyun Hong
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 |
|---|---|---|---|
| use FD INR surrogate instead of Naive INR surrogate (ablation) | PGD attack on DWS · MNIST | perturbed accuracy: 10.12% -> 0.01% | Improved |
| use FD INR surrogate instead of Naive INR surrogate (ablation) | PGD attack on NFT · MNIST | perturbed accuracy: 20.57% -> 0.00% | Improved |
| use FD INR surrogate instead of Naive INR surrogate (ablation) | PGD attack on Inr2Vec · F-MNIST | perturbed accuracy: 17.91% -> 13.89% | Improved |
| use FD INR surrogate instead of Naive INR surrogate (ablation) | PGD attack on Inr2Vec · CIFAR-10 | perturbed accuracy: 0.46% -> 5.22% | Hurt |
| use FD embedding surrogate instead of Naive embedding surrogate (ablation) | PGD attack on Inr2Vec · CIFAR-10 | perturbed accuracy: 4.58% -> 13.59% | Hurt |
| perturb INR weights (function space) instead of input space via surrogate (ablation) | attack on NFT · CIFAR-10 | perturbed accuracy: 0.15% -> 39.82% | Hurt |
| perturb INR weights (function space) instead of input space via surrogate (ablation) | attack on DWS · MNIST | perturbed accuracy: 0.01% -> 47.74% | Hurt |
| add diffusion denoising defense | FD INR attack on NFN · F-MNIST | perturbed accuracy: 0.27% -> 14.22% | Improved |
| add diffusion denoising defense | FD INR attack on NFT · MNIST | perturbed accuracy: 0.00% -> 0.00% | No clear effect |
| add diffusion denoising defense | FD INR attack on ScaleGMN · CIFAR-10 | perturbed accuracy: 0.25% -> 4.84% | Improved |
| add function-space adversarial training | NFN · MNIST | perturbed accuracy: 25.8% -> 81.1% | Improved |
| add function-space adversarial training | NFT · CIFAR-10 | perturbed accuracy: 39.8% -> 24.1% | Hurt |
Closest papers by meaning
- Adversarial Robustness of Supervised Sparse Coding (NeurIPS 2020 · similarity 0.89)
- Adversarial Robustness Across Representation Spaces (CVPR 2021 · similarity 0.88)
- Adversarial vulnerability for any classifier (NeurIPS 2018 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.88)
- Rethinking Invariance Regularization in Adversarial Training to Improve Robustness-Accuracy Trade-off (ICLR 2025 · similarity 0.87)
- First Line of Defense: A Robust First Layer Mitigates Adversarial Attacks (AAAI 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.
