mlsanityData to 7 Oct 2026

Generative Uncertainty in Diffusion Models

arXiv 2502.20946 ↗First posted 28 Feb 2025cs.LG cs.AIcited by 20

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Diffusion models have recently driven significant breakthroughs in generative modeling. While state-of-the-art models produce high-quality samples on average, individual samples can still be low quality. Detecting such samples without human inspection remains a challenging task. To address this, we propose a Bayesian framework for estimating generative uncertainty of synthetic samples. We outline how to make Bayesian inference practical for large, modern generative models and introduce a new semantic likelihood (evaluated in the latent space of a feature extractor) to address the challenges posed by high-dimensional sample spaces. Through our experiments, we demonstrate that the proposed generative uncertainty effectively identifies poor-quality samples and significantly outperforms existing uncertainty-based methods. Notably, our Bayesian framework can be applied post-hoc to any pretrained diffusion or flow matching model (via the Laplace approximation), and we propose simple yet effective techniques to minimize its computational overhead during sampling.

Key facts. Cited by 20 papers in the map (data through 7 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 68 of its 88 references that are in the map. First posted on arXiv 28 Feb 2025.

Builds on

68 of 88 references.

Closest ancestors · 18

Same direction, newest first

  1. 2024Diffusion Model Guided Sampling with Pixel-Wise Aleatoric Uncertainty Estimation29 Nov
  2. 2024Shedding Light on Large Generative Networks: Estimating Epistemic Uncertainty in Diffusion Models5 Jun
  3. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesis5 Mar
  4. 2024Estimating Epistemic and Aleatoric Uncertainty with a Single Model5 Feb
  5. 2023BayesDiff: Estimating Pixel-wise Uncertainty in Diffusion via Bayesian Inference17 Oct
  6. 2023How to Trust Your Diffusion Model: A Convex Optimization Approach to Conformal Risk Control7 Feb
  7. 2022Scalable Diffusion Models with Transformers19 Dec
  8. 2022Diffusion Models already have a Semantic Latent Space20 Oct
  9. 2022All are Worth Words: A ViT Backbone for Diffusion Models25 Sep
  10. 2022Classifier-Free Diffusion Guidance26 Jul
  11. 2022DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps2 Jun
  12. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
6 more
  1. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  2. 2021Diffusion Models Beat GANs on Image Synthesis11 May
  3. 2021Maximum Likelihood Training of Score-Based Diffusion Models22 Jan
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov
  5. 2020Denoising Diffusion Implicit Models6 Oct
  6. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 50

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  3. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,451
  4. 2022Flow Matching for Generative Modelingcited by 5,426
  5. 2013Intriguing properties of neural networkscited by 5,114
  6. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
  7. 2018Neural Ordinary Differential Equationscited by 3,733
  8. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,550
  9. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,335
  10. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 925
  11. 2019Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methodscited by 903
  12. 2023Stochastic Interpolants: A Unifying Framework for Flows and Diffusionscited by 897
38 more
  1. 2017Deep Bayesian Active Learning with Image Datacited by 812
  2. 2022Equivariant Diffusion for Molecule Generation in 3Dcited by 666
  3. 2019A Simple Baseline for Bayesian Uncertainty in Deep Learningcited by 555
  4. 2020Bayesian Deep Learning and a Probabilistic Perspective of Generalizationcited by 480
  5. 2018Do Deep Generative Models Know What They Don't Know?cited by 443
  6. 2022DiffuSeq: Sequence to Sequence Text Generation with Diffusion Modelscited by 439
  7. 2018Assessing Generative Models via Precision and Recallcited by 401
  8. 2021Laplace Redux -- Effortless Bayesian Deep Learningcited by 359
  9. 2020Hands-on Bayesian Neural Networks -- a Tutorial for Deep Learning Userscited by 336
  10. 2017Stochastic Gradient Descent as Approximate Bayesian Inferencecited by 298
  11. 2015A note on the evaluation of generative modelscited by 291
  12. 2017Advances in Variational Inferencecited by 236
  13. 2020Being Bayesian, Even Just a Bit, Fixes Overconfidence in ReLU Networkscited by 227
  14. 2023Diffusion Hyperfeatures: Searching Through Time and Space for Semantic Correspondencecited by 211
  15. 2023Flow Matching in Latent Spacecited by 155
  16. 2022Image-to-Image Regression with Distribution-Free Uncertainty Quantification and Applications in Imagingcited by 107
  17. 2021Scalable Marginal Likelihood Estimation for Model Selection in Deep Learningcited by 105
  18. 2024Understanding Hallucinations in Diffusion Models through Mode Interpolationcited by 87
  19. 2020Bayesian Deep Learning via Subnetwork Inferencecited by 85
  20. 2020The Case for Bayesian Deep Learningcited by 68
  21. 2022Do Bayesian Neural Networks Need To Be Fully Stochastic?cited by 63
  22. 2020Depth Uncertainty in Neural Networkscited by 56
  23. 2020GIQA: Generated Image Quality Assessmentcited by 51
  24. 2024EmerDiff: Emerging Pixel-level Semantic Knowledge in Diffusion Modelscited by 42
  25. 2024Rethinking Aleatoric and Epistemic Uncertaintycited by 40
  26. 2024A-Bench: Are LMMs Masters at Evaluating AI-generated Images?cited by 39
  27. 2019Bayesian Variational Autoencoders for Unsupervised Out-of-Distribution Detectioncited by 35
  28. 2022Rarity Score : A New Metric to Evaluate the Uncommonness of Synthesized Imagescited by 35
  29. 2017Bayesian GANcited by 31
  30. 2023Principal Uncertainty Quantification with Spatial Correlation for Image Restoration Problemscited by 28
  31. 2023A Primer on Bayesian Neural Networks: Review and Debatescited by 26
  32. 2022Semantic uncertainty intervals for disentangled latent spacescited by 22
  33. 2023The Memory Perturbation Equation: Understanding Model's Sensitivity to Datacited by 16
  34. 2024Towards Understanding and Quantifying Uncertainty for Text-to-Image Generationcited by 11
  35. 2024What makes an image realistic?cited by 10
  36. 2023Fully Bayesian Autoencoders with Latent Sparse Gaussian Processescited by 9
  37. 2023Uncertainty and Structure in Neural Ordinary Differential Equationscited by 6
  38. 2023Anomaly Score: Evaluating Generative Models and Individual Generated Images based on Complexity and Vulnerabilitycited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 3; 2026: 17.

20 papers.

Sort:

By area:

  1. 2025Algebraic Language Models for Inverse Design of Metamaterials via Diffusion Transformerscited by 5
  2. 2025Mitigating Diffusion Model Hallucinations with Dynamic Guidancecited by 5
  3. 2026Calibrated Test-Time Guidance for Bayesian Inferencecited by 4
  4. 2026Uncertainty Quantification for Flow-Based Vision-Language-Action Modelscited by 4
  5. 2026Flow Matching with Uncertainty Quantification and Guidancecited by 3
  6. 2026Quantifying Epistemic Uncertainty in Diffusion Modelscited by 2
  7. 2026When Are Two Scores Better Than One? Investigating Ensembles of Diffusion Modelscited by 1
  8. 2026Reducing Memorisation in Generative Models via Riemannian Bayesian Inferencecited by 1
  9. 2026Uncertainty-Aware Distribution-to-Distribution Flow Matching for Scientific Imagingcited by 1
  10. 2026Spatially Adaptive Noise Injectioncited by 1
  11. 2025Control-Augmented Autoregressive Diffusion for Data Assimilationcited by 0
  12. 2026Uncertainty Quantification for Large Language Diffusion Modelscited by 0
8 more
  1. 2026Flowing with Confidencecited by 0
  2. 2026Efficient and Uncertainty-Aware Diffusion Framework for Offline-to-Online Reinforcement Learningcited by 0
  3. 2026Uncertainty Estimation for Molecular Diffusion Modelscited by 0
  4. 2026Test-Time Noise Guided Adaptation for Realistic Autoregressive Video Generationcited by 0
  5. 2026LM-X: Explainable Vision--Language--Action Modeling via Progress, Event, and Uncertainty Predictioncited by 0
  6. 2026Is your uncertainty map wrong, or is its target? Exact diagnostics for the Tweedie diagonal, and a gradient-free alternativecited by 0
  7. 2026Generative Uncertainty as a Self-supervised Signal for Semantic Similarity Learningcited by 0
  8. 2026Matisse: Evidence-Space Reasoning for Active 3D Reconstructioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Generative Modeling with Bayesian Sample Inference
  2. 2024Conditional sampling within generative diffusion models
  3. 2025Diffusion Models: A Mathematical Introduction
  4. 2025Improved probabilistic regression using diffusion models
  5. 2026Leveraging generative models to assist Monte Carlo sampling
  6. 2022Let us Build Bridges: Understanding and Extending Diffusion Generative Models
  7. 2023User-defined Event Sampling and Uncertainty Quantification in Diffusion Models for Physical Dynamical Systems
  8. 2024Generative Diffusion Modeling: A Practical Handbook