Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
What Are Bayesian Neural Network Posteriors Really Like?
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 full-batch HMC instead of standard training | ResNet-20-FRN · CIFAR-10 | accuracy: 83.44 -> 89.64 | Helped |
| Use HMC instead of deep ensembles | ResNet-20-FRN · CIFAR-10 | accuracy: 88.49 -> 89.64 | Helped |
| Use HMC rather than Mahalanobis scoring | ResNet-20-FRN · CIFAR-100 OOD detection | AUC-ROC: 0.882 -> 0.857 | Hurt |
| Use HMC rather than deep ensembles | ResNet-20-FRN · CIFAR-10-C | accuracy: 76.99 -> 70.91 | Hurt |
| Use a logistic rather than Gaussian prior | CNN-LSTM HMC · IMDB | Log likelihood: -0.311 -> -0.304 | Helped |
| Replace Gaussian prior with a mixture of Gaussians | CNN-LSTM HMC · IMDB | accuracy: 0.866 -> 0.863 | No clear effect |
| Use SGHMC-CLR instead of HMC | ResNet-20-FRN · CIFAR-10 | agreement with HMC: 94.01 -> 92.67 | Hurt |
| Use deep ensembles instead of MFVI | ResNet-20-FRN · CIFAR-10 | total variation from HMC: 0.136 -> 0.115 | Helped |
| Add data augmentation while using a cold posterior (T=0.1) (ablation) | ResNet-20 · CIFAR-10 | accuracy: 87.46 -> 91.12 | Helped |
| Use a cold posterior (T=0.1) without augmentation (ablation) | ResNet-20-FRN · CIFAR-10 | accuracy: 84.27 -> 80.84 | Hurt |
| Increase the HMC step size (ablation) | ResNet-20-FRN HMC · CIFAR-10 | BMA log-likelihood: -0.331 -> -0.3406 -> -0.3407 -> -0.895 | Hurt |
| Increase HMC trajectory length (ablation) | ResNet-20-FRN HMC · CIFAR-10 | accuracy and log-likelihood | Helped |
Closest papers by meaning
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.90)
- How Good is the Bayes Posterior in Deep Neural Networks Really? (ICML 2020 · similarity 0.90)
- Parameter Expanded Stochastic Gradient Markov Chain Monte Carlo (ICLR 2025 · similarity 0.89)
- A Symmetry-Aware Exploration of Bayesian Neural Network Posteriors (ICLR 2024 · similarity 0.88)
- Variational Bayesian Pseudo-Coreset (ICLR 2025 · similarity 0.88)
- Scalable Bayesian Learning with posteriors (ICLR 2025 · similarity 0.88)
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.