Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Uncertainty Estimation Using a Single Deep Deterministic Neural Network
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 a two-sided gradient penalty to DUQ (ablation) | DUQ · CIFAR-10 vs SVHN | AUROC: 0.861 ± 0.032 -> 0.927 ± 0.013 | Helped |
| Add a two-sided gradient penalty to DUQ (ablation) | DUQ · FashionMNIST vs MNIST | AUROC: 0.948 ± .004 -> 0.955 ± .007 | Mixed |
| Increase gradient penalty weight from λ = 0 to λ = 0.5 (ablation) | DUQ · CIFAR-10 | accuracy: 94.2% ± 0.2 -> 93.2% ± 0.4 | Hurt |
| Remove the gradient penalty (ablation) | DUQ · Two Moons | out-of-distribution uncertainty quality | Hurt |
| Replace the two-sided gradient penalty with a one-sided penalty (ablation) | DUQ · Two Moons | out-of-distribution uncertainty quality | Hurt |
| Use an invertible feature extractor instead of the gradient penalty (ablation) | DUQ · Two Moons | uncertainty quality | Helped |
| Use an invertible feature extractor (ablation) | DUQ · higher-dimensional datasets | accuracy and uncertainty | Hurt |
| Apply gradient penalty to the feature vector or individual kernel distances rather than the sum of kernels (ablation) | DUQ · larger-scale datasets | uncertainty quality | Hurt |
| Increase gradient penalty weight from λ = 0 to λ = 0.05 (ablation) | DUQ · FashionMNIST | accuracy: 92.4% ± .2 -> 92.4% ± .2 | No clear effect |
| Increase gradient penalty weight from λ = 0 to λ = 0.05 (ablation) | DUQ · FashionMNIST vs NotMNIST | AUROC: 0.933 ± .009 -> 0.946 ± .018 | Helped |
Closest papers by meaning
- Deep Deterministic Uncertainty: A New Simple Baseline (CVPR 2023 · similarity 0.91)
- Density-Softmax: Efficient Test-time Model for Uncertainty Estimation and Robustness under Distribution Shifts (ICML 2024 · similarity 0.88)
- Simple and Principled Uncertainty Estimation with Deterministic Deep Learning via Distance Awareness (NeurIPS 2020 · similarity 0.87)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.87)
- One Step Closer to Unbiased Aleatoric Uncertainty Estimation (AAAI 2024 · similarity 0.86)
- Uncertainty Quantification for Deep Regression using Contextualised Normalizing Flows (NeurIPS 2025 · 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.