Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
CUPID: A Plug-in Framework for Joint Aleatoric and Epistemic Uncertainty Estimation with a Single Model
Xinran Xu, Xiuyi Fan
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 |
|---|---|---|---|
| joint training of both branches (vs aleatoric alone) (ablation) | CUPID Aleatoric · GLV2 misclassification detection | AUC: 0.863 -> 0.870 | Improved |
| joint training of both branches (vs epistemic alone) (ablation) | CUPID Epistemic · GLV2 misclassification detection | AUC: 0.744 -> 0.769 | Improved |
| joint training of both branches (vs epistemic alone) (ablation) | CUPID Epistemic · PAPILA OOD detection | AUC: 0.771 -> 0.877 | Improved |
| joint training of both branches (vs aleatoric alone) (ablation) | CUPID Aleatoric · PAPILA OOD detection | AUC: 0.508 -> 0.379 | Hurt |
| add differential feature loss (ablation) | CUPID Epistemic · PAPILA OOD detection | AUC: 0.839 -> 0.877 | Improved |
| add differential feature loss (ablation) | CUPID Aleatoric · CIFAR10 OOD detection | AUC: 0.988 -> 0.983 | Hurt |
| add differential feature loss (ablation) | CUPID Aleatoric · HAM10000 misclassification detection | Spearman: +0.004 | Improved |
| move CUPID closer to the output layer (ablation) | CUPID Aleatoric · medical image misclassification detection | AUC | Improved |
| move CUPID to earlier layers (ablation) | CUPID Epistemic · medical image misclassification detection | AUC | Improved |
| insert CUPID after the upsampling module in ESRGAN (ablation) | CUPID Aleatoric · super-resolution uncertainty | Pearson correlation | Improved |
| insert CUPID before the upsampling module in ESRGAN (ablation) | CUPID Epistemic · super-resolution uncertainty | Pearson correlation | Improved |
| use CUPID instead of MC Dropout on tabular MLP | MLP classifier · Covertype misclassification detection | AUC: 0.563 -> 0.769 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Rate-in — Rate-In: Information-Driven Adaptive Dropout Rates for Improved Inference-Time Uncertainty Estimation (CVPR 2025) · better on 2 of 2 shared results
- BayesCap — BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks (ECCV 2022) · better on 2 of 2 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
- Vicinal Label Supervision for Reliable Aleatoric and Epistemic Uncertainty Estimation (NeurIPS 2025 · similarity 0.87)
- Uncertainty Quantification for Deep Regression using Contextualised Normalizing Flows (NeurIPS 2025 · similarity 0.86)
- Enhancing Uncertainty Estimation and Interpretability with Bayesian Non-negative Decision Layer (ICLR 2025 · similarity 0.86)
- Are Uncertainty Quantification Capabilities of Evidential Deep Learning a Mirage? (NeurIPS 2024 · similarity 0.86)
- Epistemic Uncertainty Quantification For Pre-Trained Neural Networks (CVPR 2024 · similarity 0.86)
- Evaluating Epistemic Uncertainty: Beyond OOD Detection and Active Learning (NeurIPS 2026 · 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.
