Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Simple and Principled Uncertainty Estimation with Deterministic Deep Learning via Distance Awareness
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 |
|---|---|---|---|
| Use SNGP instead of a dense-output deep ensemble | Wide ResNet-28-10 · CIFAR-10, SVHN OOD | OOD AUPR: 0.964 ± 0.01 -> 0.990 ± 0.01 | Helped |
| Use SNGP instead of a dense-output deep ensemble | Wide ResNet-28-10 · CIFAR-100, SVHN OOD | OOD AUPR: 0.888 ± 0.01 -> 0.923 ± 0.01 | Helped |
| Use SNGP instead of a dense-output deep ensemble | BERT Base · CLINC OOS | OOD AUPR: 0.862 ± 0.01 -> 0.880 ± 0.01 | Helped |
| Add the GP output layer to the spectrally normalized network (ablation) | DNN-SN · CIFAR-10, SVHN OOD | OOD AUPR: 0.974 ± 0.01 -> 0.990 ± 0.01 | Helped |
| Add spectral normalization to the GP-output network (ablation) | DNN-GP · CIFAR-10, SVHN OOD | OOD AUPR: 0.976 ± 0.01 -> 0.990 ± 0.01 | Helped |
| Add spectral normalization to the GP-output network (ablation) | DNN-GP · CLINC OOS | OOD AUROC: 0.941 ± 0.01 -> 0.969 ± 0.01 | Helped |
| Apply spectral normalization to BERT transformer hidden layers (ablation) | BERT SNGP · CLINC OOS | predictive and calibration performance | Hurt |
| Increase the GP hidden dimension from 1024 to 2048 (ablation) | SNGP · Vision and language experiments | model performance | No clear effect |
| Vary input layer normalization values (ablation) | SNGP · Vision and language experiments | model performance | No clear effect |
Closest papers by meaning
- Deep Deterministic Uncertainty: A New Simple Baseline (CVPR 2023 · similarity 0.90)
- Make Me a BNN: A Simple Strategy for Estimating Bayesian Uncertainty from Pre-trained Models (CVPR 2024 · similarity 0.89)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.88)
- SURE: SUrvey REcipes for building reliable and robust deep networks (CVPR 2024 · similarity 0.87)
- A Rate-Distortion View of Uncertainty Quantification (ICML 2024 · similarity 0.87)
- Distance-informed Neural Processes (NeurIPS 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.