Research map / Adversarial robustness and security / Certified and empirical robustness
Improving Bayesian Neural Networks by Adversarial Sampling
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Adversarial Sampling during training | ResNet20 · CIFAR-10 | ensembled accuracy: 87.01 ± 0.65 -> 88.76 ± 0.73 | Helped |
| Add Adversarial Sampling during training | ResNet56 · CIFAR-10 | ensembled accuracy: 88.22 ± 0.41 -> 89.61 ± 0.93 | Helped |
| Add Adversarial Sampling during training | VGG · CIFAR-10 | ensembled accuracy: 89.80 ± 0.12 -> 90.86 ± 0.32 | Helped |
| Add Adversarial Sampling during training | ResNet20 · CIFAR-100 | ensembled accuracy: 56.56 ± 1.07 -> 57.62 ± 0.91 | Helped |
| Add Adversarial Sampling during training | ResNet56 · CIFAR-100 | ensembled accuracy: 53.21 ± 2.40 -> 58.63 ± 1.58 | Helped |
| Add Adversarial Sampling during training | VGG · CIFAR-100 | ensembled accuracy: 47.60 ± 1.01 -> 56.11 ± 0.66 | Helped |
| Use Adversarial Sampling with λ increased from 0 to 0.8 (ablation) | ResNet20 · CIFAR-10 | error rate | Helped |
| Use completely Adversarial Sampling (λ = 1.0) instead of a mixture (ablation) | ResNet20 · CIFAR-10 | error rate | Hurt |
| Add Adversarial Sampling during training | ResNet20 · CIFAR-10 | lowest sampled-model accuracy: 82.73 ± 0.88 -> 86.33 ± 0.45 | Helped |
| Add Adversarial Sampling during training | ResNet56 · CIFAR-10 | lowest sampled-model accuracy: 82.71 ± 0.55 -> 87.30 ± 0.32 | Helped |
| Add Adversarial Sampling during training | VGG · CIFAR-10 | lowest sampled-model accuracy: 85.04 ± 0.44 -> 88.68 ± 0.53 | Helped |
| Add Adversarial Sampling during training | ResNet20 · CIFAR-100 | lowest sampled-model accuracy: 52.54 ± 1.54 -> 54.83 ± 0.95 | Helped |
Closest papers by meaning
- Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network (ICLR 2019 · similarity 0.91)
- Adversarial Distillation of Bayesian Neural Network Posteriors (ICML 2018 · similarity 0.89)
- Bayesian Adversarial Learning (NeurIPS 2018 · similarity 0.89)
- Bayesian Learning with Information Gain Provably Bounds Risk for a Robust Adversarial Defense (ICML 2022 · similarity 0.88)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.87)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (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.