# Is MC Dropout Bayesian?

arXiv: https://arxiv.org/abs/2110.04286 · HTML: https://mlsanity.com/paper/2110.04286/

## Key facts

- Cited by 35 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 26 references that are in the map.
- First posted on arXiv 8 Oct 2021.

## Builds on (most cited first, 11 of 11)

- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [Weight Uncertainty in Neural Networks](https://mlsanity.com/paper/1505.05424/) (2015)
- [Advances in Variational Inference](https://mlsanity.com/paper/1711.05597/) (2017)
- [Expectation Propagation for approximate Bayesian inference](https://mlsanity.com/paper/1301.2294/) (2013)
- [Multiplicative Normalizing Flows for Variational Bayesian Neural Networks](https://mlsanity.com/paper/1703.01961/) (2017)
- [A Unified Particle-Optimization Framework for Scalable Bayesian Sampling](https://mlsanity.com/paper/1805.11659/) (2018)
- [Variational Inference using Implicit Distributions](https://mlsanity.com/paper/1702.08235/) (2017)
- [Variational Boosting: Iteratively Refining Posterior Approximations](https://mlsanity.com/paper/1611.06585/) (2016)
- [Bridging the Gap between Stochastic Gradient MCMC and Stochastic Optimization](https://mlsanity.com/paper/1512.07962/) (2015)
- [Meta-Learning for Stochastic Gradient MCMC](https://mlsanity.com/paper/1806.04522/) (2018)
- [Hierarchical Implicit Models and Likelihood-Free Variational Inference](https://mlsanity.com/paper/1702.08896/) (2017)

## Built on by (most cited first, 30 of 35)

- [A Primer on Bayesian Neural Networks: Review and Debates](https://mlsanity.com/paper/2309.16314/) (2023)
- [GFlowOut: Dropout with Generative Flow Networks](https://mlsanity.com/paper/2210.12928/) (2022)
- [Efficient Bayesian Uncertainty Estimation for nnU-Net](https://mlsanity.com/paper/2212.06278/) (2022)
- [Conformal Prediction for Causal Effects of Continuous Treatments](https://mlsanity.com/paper/2407.03094/) (2024)
- [DiffPO: A causal diffusion model for learning distributions of potential outcomes](https://mlsanity.com/paper/2410.08924/) (2024)
- [That Label's Got Style: Handling Label Style Bias for Uncertain Image Segmentation](https://mlsanity.com/paper/2303.15850/) (2023)
- [Data-Driven Physics-Informed Neural Networks: A Digital Twin Perspective](https://mlsanity.com/paper/2401.08667/) (2024)
- [Uncertainty Quantification with the Empirical Neural Tangent Kernel](https://mlsanity.com/paper/2502.02870/) (2025)
- [Body Knowledge and Uncertainty Modeling for Monocular 3D Human Body Reconstruction](https://mlsanity.com/paper/2308.00799/) (2023)
- [A Review of Bayesian Uncertainty Quantification in Deep Probabilistic Image Segmentation](https://mlsanity.com/paper/2411.16370/) (2024)
- [Bayesian posterior approximation with stochastic ensembles](https://mlsanity.com/paper/2212.08123/) (2022)
- [An uncertainty-aware Digital Shadow for underground multimodal CO2 storage monitoring](https://mlsanity.com/paper/2410.01218/) (2024)
- [Split Conformal Prediction in the Function Space with Neural Operators](https://mlsanity.com/paper/2509.04623/) (2025)
- [Towards Reliable Uncertainty Quantification via Deep Ensembles in Multi-output Regression Task](https://mlsanity.com/paper/2303.16210/) (2023)
- [Inter-Rater Uncertainty Quantification in Medical Image Segmentation via Rater-Specific Bayesian Neural Networks](https://mlsanity.com/paper/2306.16556/) (2023)
- [Neural Network Prediction of Strong Lensing Systems with Domain Adaptation and Uncertainty Quantification](https://mlsanity.com/paper/2411.03334/) (2024)
- [Why you don't overfit, and don't need Bayes if you only train for one epoch](https://mlsanity.com/paper/2411.14478/) (2024)
- [Generation Space Size: Understanding and Calibrating Open-Endedness of LLM Generations](https://mlsanity.com/paper/2510.12699/) (2025)
- [Unreliable Uncertainty Estimates with Monte Carlo Dropout](https://mlsanity.com/paper/2512.14851/) (2025)
- [TRIE: An Evaluation Framework for Stochastic PDE Surrogates](https://mlsanity.com/paper/2607.00196/) (2026)
- [Propagating Variational Model Uncertainty for Bioacoustic Call Label Smoothing](https://mlsanity.com/paper/2210.10526/) (2022)
- [Distance Preserving Machine Learning for Uncertainty Aware Accelerator Capacitance Predictions](https://mlsanity.com/paper/2307.02367/) (2023)
- [Out-of-distribution Object Detection through Bayesian Uncertainty Estimation](https://mlsanity.com/paper/2310.19119/) (2023)
- [Enhance Eye Disease Detection using Learnable Probabilistic Discrete Latents in Machine Learning Architectures](https://mlsanity.com/paper/2402.16865/) (2024)
- [Bayesian Uncertainty Estimation by Hamiltonian Monte Carlo: Applications to Cardiac MRI Segmentation](https://mlsanity.com/paper/2403.02311/) (2024)
- [Bayesian Neural Networks for 2D MRI Segmentation](https://mlsanity.com/paper/2311.14875/) (2023)
- [BayesJudge: Bayesian Kernel Language Modelling with Confidence Uncertainty in Legal Judgment Prediction](https://mlsanity.com/paper/2404.10481/) (2024)
- [On Uniform, Bayesian, and PAC-Bayesian Deep Ensembles](https://mlsanity.com/paper/2406.05469/) (2024)
- [Taylor-Sensus Network: Embracing Noise to Enlighten Uncertainty for Scientific Data](https://mlsanity.com/paper/2409.07942/) (2024)
- [Fine-Grained Uncertainty Quantification via Collisions](https://mlsanity.com/paper/2411.12127/) (2024)

## Abstract (arXiv)

MC Dropout is a mainstream "free lunch" method in medical imaging for approximate Bayesian computations (ABC). Its appeal is to solve out-of-the-box the daunting task of ABC and uncertainty quantification in Neural Networks (NNs); to fall within the variational inference (VI) framework; and to propose a highly multimodal, faithful predictive posterior. We question the properties of MC Dropout for approximate inference, as in fact MC Dropout changes the Bayesian model; its predictive posterior assigns 0 probability to the true model on closed-form benchmarks; the multimodality of its predictive posterior is not a property of the true predictive posterior but a design artefact. To address the need for VI on arbitrary models, we share a generic VI engine within the pytorch framework. The code includes a carefully designed implementation of structured (diagonal plus low-rank) multivariate normal variational families, and mixtures thereof. It is intended as a go-to no-free-lunch approach, addressing shortcomings of mean-field VI with an adjustable trade-off between expressivity and computational complexity.