Research map / Adversarial robustness and security / Certified and empirical robustness
GradDiv: Adversarial Robustness of Randomized Neural Networks via Gradient Diversity Regularization
Sungyoon Lee, Hoki Kim, Jaewook Lee
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 GradDiv regularization | Adv-BNN · CIFAR10 | combined robust accuracy (%): 42.95 -> 45.77 | Improved |
| add GradDiv regularization | Adv-BNN · CIFAR10 | clean accuracy (%): 75.78 -> 77.51 | Improved |
| add GradDiv regularization | Adv-BNN · CIFAR10 | EOT-PGD accuracy (%): 43.39 -> 45.88 | Improved |
| add GradDiv regularization | Adv-BNN · CIFAR10 | EOT-APGDCE accuracy (%): 59.74 -> 68.91 | Improved |
| add GradDiv regularization | Adv-BNN · CIFAR10 | PGD accuracy (%): 66.32 -> 75.72 | Improved |
| add GradDiv regularization | Adv-BNN · CIFAR10 | B&B accuracy (%): 75.09 -> 75.38 | No clear effect |
| add GradDiv regularization | Adv-BNN · CIFAR10 | FAB accuracy (%): 74.94 -> 77.04 | No clear effect |
| add GradDiv regularization | Adv-BNN · CIFAR10 | FAB accuracy vs RSE (%): 83.37 -> 77.04 | Hurt |
| add GradDiv regularization | Adv-BNN · MNIST | sample MRL of gradients: 0.64 -> 0.15 | Improved |
| add GradDiv regularization | Adv-BNN · MNIST | EOT-PGD accuracy gain at eps 0.36, 0.38, 0.4: +37.52%p, +37.72%p, +15.55%p | Improved |
| add GradDiv regularization | Adv-BNN · STL10 | EOT-PGD accuracy | Improved |
| add GradDiv-DPP regularization | Adv-BNN · MNIST | transfer attack accuracy among sample models (%): 24 -> 75 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Adv-BNN — Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network (ICLR 2019) · better on 12 of 12 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Theoretical evidence for adversarial robustness through randomization (NeurIPS 2019 · similarity 0.90)
- Scaling up the Randomized Gradient-Free Adversarial Attack Reveals Overestimation of Robustness Using Established Attacks (IJCV 2019 · similarity 0.89)
- Adversarial Robustness via Deformable Convolution with Stochasticity (ICML 2025 · similarity 0.89)
- Adversarial Robustness through Random Weight Sampling (NeurIPS 2023 · similarity 0.88)
- Learning Randomized Algorithms with Transformers (ICLR 2025 · similarity 0.87)
- Understanding the Robustness of Randomized Feature Defense Against Query-Based Adversarial Attacks (ICLR 2024 · 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.
