Research map / Adversarial robustness and security / Certified and empirical robustness
Reverse KL-Divergence Training of Prior Networks: Improved Uncertainty and Adversarial Robustness
Andrey Malinin, Mark Gales
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 |
|---|---|---|---|
| Train Prior Networks with the proposed reverse-KL criterion instead of standard adversarial training | Prior Networks · Adversarial training | adversarial robustness | Improved |
| Set the reverse-KL loss gamma to 10.0 rather than 1.0 (ablation) | Reverse-KL Prior Networks · Other datasets | reverse-KL loss performance | Improved |
| Vary beta (ablation) | Reverse-KL Prior Networks · Not specified | model performance | No clear effect |
| Set beta to 0 for OOD training data (ablation) | Reverse-KL Prior Networks · OOD detection | separation of in-domain and OOD uncertainty | Improved |
| Set beta to 1 for adversarial training data (ablation) | Reverse-KL Prior Networks · Adversarial training | joint target-class prediction and uncertainty | Improved |
Closest papers by meaning
- A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks (NeurIPS 2018 · similarity 0.86)
- Decoupled Kullback-Leibler Divergence Loss (NeurIPS 2024 · similarity 0.86)
- DIBS: Diversity Inducing Information Bottleneck in Model Ensembles (AAAI 2021 · similarity 0.86)
- Using Pre-Training Can Improve Model Robustness and Uncertainty (ICML 2019 · similarity 0.85)
- LiBRe: A Practical Bayesian Approach to Adversarial Detection (CVPR 2021 · similarity 0.85)
- Uncertainty-Aware Deep Classifiers Using Generative Models (AAAI 2020 · similarity 0.85)
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.
