Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Epistemic versus aleatoric decomposition
Critically analyzes how to define and separate epistemic and aleatoric uncertainty.
75 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2017–2026) · an approach in Uncertainty and Gaussian processes, keyword cluster “uncertainty estimation · epistemic · quantification”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
25 papers in 2023–24 and 31 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 9.2% to 10.1%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2017 · What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? (NeurIPS 2017 · 2,921 citations)
- 2019 · CXPlain: Causal Explanations for Model Interpretation under Uncertainty (NeurIPS 2019 · 73 citations)
- 2020 · Crowd Counting with Decomposed Uncertainty (AAAI 2020 · 100 citations)
- 2021 · Learning Probabilistic Ordinal Embeddings for Uncertainty-Aware Regression (CVPR 2021 · 55 citations)
- 2022 · Improving Evidential Deep Learning via Multi-Task Learning (AAAI 2022 · 28 citations)
- 2023 · The Unreasonable Effectiveness of Deep Evidential Regression (AAAI 2023 · 37 citations)
- 2024 · Estimating Epistemic and Aleatoric Uncertainty with a Single Model (NeurIPS 2024 · 17 citations)
- 2025 · Adaptive Sampling to Reduce Epistemic Uncertainty Using Prediction Interval-Generation Neural Networks (AAAI 2025 · 2 citations)
- 2026 · Cooperative Variance Estimation and Bayesian Neural Networks for Disentangling Aleatoric and Epistemic Uncertainties (ICML 2026 · 2 citations)
Earliest papers here
- What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision? (NeurIPS 2017 · 2,921 citations)
- Estimating Uncertainty Online Against an Adversary (AAAI 2017 · 11 citations)
- CXPlain: Causal Explanations for Model Interpretation under Uncertainty (NeurIPS 2019 · 73 citations)
What has been tried
53 papers here report 460 measured changes (a component added, removed or swapped, compared with a baseline): helped 58%, no clear effect 11%, hurt 23%, mixed 8%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| increase the number of sampled hyper-network weights M | OOD temperature forecasting · epistemic uncertainty localization | M={2, 4, 8, 16} | Helped | Estimating Epistemic and Aleatoric Uncertainty with a Single Model (NeurIPS 2024 · 17 citations) |
| Use random forest alone rather than an ensemble of logistic regression, support vector machine, and random forest | UQ estimation · Brier score | Helped | Using AI Uncertainty Quantification to Improve Human Decision-Making (ICML 2024 · 6 citations) | |
| decrease λ across {1e-4,1e-3,1e-2,1e-1} | CIFAR10 OOD detection · average AUROC | Helped | Are Uncertainty Quantification Capabilities of Evidential Deep Learning a Mirage? (NeurIPS 2024 · 5 citations) | |
| Log score as loss/plug-in G instead of Brier or Spherical scores | CIFAR10 OOD · AUROC | Log Bayes (1) 91.36 vs Brier 90.44 and Spherical 90.46 on CIFAR100 | Helped | From Risk to Uncertainty: Generating Predictive Uncertainty Measures via Bayesian Estimation (ICLR 2025 · 2 citations) |
| Use single-sample action selection rather than double sampling | Tabular MDP · value estimation bias | Hurt | Epistemic Bellman Operators (AAAI 2025 · 1 citation) | |
| add entropy regularization to outer loss minimization | synthetic binary classification · faithfulness to reference distribution | Mixed | Is Epistemic Uncertainty Faithfully Represented by Evidential Deep Learning Methods? (ICML 2024 · 1 citation) | |
| use Kernel SHAP instead of RELAX as the base stochastic R-XAI method | PASCAL-VOC and MS-COCO vs EuroSAT, ResNet50 and ViT · AUROC | 0.999 vs 1.000; 0.999 vs 0.973; 1.000 vs 1.000; 1.000 vs 0.939 | No clear effect | REPEAT: Improving Uncertainty Estimation in Representation Learning Explainability (AAAI 2025 · 0 citations) |
| add the mask token to the MLP input | COCO · model uncertainty rel. AUC (%) | 61.19% -> 63.66% | Helped | UncertainSAM: Fast and Efficient Uncertainty Quantification of the Segment Anything Model (ICML 2025 · 0 citations) |
| Remove more uncertainty-region samples from the reference dataset | All datasets · detection performance | Hurt | Confidential Guardian: Cryptographically Prohibiting the Abuse of Model Abstention (ICML 2025 · 0 citations) | |
| Increase the in-context dataset size | Logistic regression, in-distribution and out-of-distribution test inputs · Epistemic uncertainty | Hurt | Variational Uncertainty Decomposition for In-Context Learning (NeurIPS 2025 · 0 citations) | |
| Learn both allocation probabilities p and dispersion τ instead of fixing them | DMNIST noisy ID · UQ performance | Helped | Uncertainty Estimation by Flexible Evidential Deep Learning (NeurIPS 2025 · 0 citations) | |
| increase the k-snapshot size used for post-hoc calibration | CIFAR-10H · aleatoric uncertainty estimation error | Helped | Provable Uncertainty Decomposition via Higher-Order Calibration (ICLR 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 17 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CIFAR-100 · 5 of 17 papers
- CIFAR-10 · 5 of 17 papers
- SVHN · 2 of 17 papers
- TinyImageNet · 2 of 17 papers
- CIFAR-10-C · 2 of 17 papers
Baselines
- BALD · 2 of 17 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 56 papers; from OpenAlex and the papers' affiliations).
- Ludwig-Maximilians-Universität München · 4 papers
- Google · 4 papers
- McGill University · 2 papers
- German Aerospace Center · 2 papers
- Massachusetts Institute of Technology · 2 papers
- Harvard University · 2 papers
- University of California, San Diego · 2 papers
- IBM · 2 papers
- Ghent University · 2 papers
- University of Oxford · 2 papers
Use it
Download BibTeX of 36 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Priors and functional variational uncertainty (112 papers)
- Single-model predictive uncertainty (74 papers)
- Bayesian decision-making under uncertainty (57 papers)
Elsewhere in the topic
- Nonparametric and PAC-Bayes Bayesian learning (130 papers)
- Simulation-based inference (118 papers)
- Variational Bayesian deep learning (105 papers)
- Approximate inference in Bayesian neural networks (93 papers)
- Sparse variational Gaussian processes (92 papers)
- PAC-Bayes and decoupled GP learning (83 papers)
- Bayesian view of meta-learning (63 papers)
- Infinite-width networks as Gaussian processes (49 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).