Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Practical Deep Learning with Bayesian Principles
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 VOGN instead of Adam | ResNet-18 · ImageNet | validation accuracy (%): 66.39 -> 67.38 | No clear effect |
| Use VOGN instead of Adam | ResNet-18 · CIFAR-10 | validation NLL: 0.55 -> 0.477 | Helped |
| Add momentum and batch normalisation (ablation) | VOGN · CIFAR-10, ResNet-18 | accuracy | Helped |
| Scale effective dataset size by the augmentation factor (ablation) | VOGN with data augmentation · image datasets | performance | Helped |
| Apply data augmentation without scaling effective dataset size (ablation) | VOGN · image datasets | performance | Hurt |
| Increase prior variance (ablation) | VOGN · ImageNet, ResNet-18 | validation accuracy | Helped |
| Increase prior variance (ablation) | VOGN · ImageNet, ResNet-18 | train-test accuracy gap: 2.29 -> 6.38 | Hurt |
| Increase training MC samples per process from 1 to 2 (ablation) | VOGN · ImageNet, ResNet-18 | validation accuracy | Helped |
| Increase training MC samples per process from 2 to 3 (ablation) | VOGN · ImageNet, ResNet-18 | test performance | No clear effect |
| Increase testing MC samples from 10 to 100 (ablation) | VOGN · ImageNet, ResNet-18 | validation accuracy | Helped |
| Replace posterior-sampled VOGN with deterministic OGN (ablation) | VOGN · ImageNet, ResNet-18 | uncertainty estimates | Hurt |
Closest papers by meaning
- Fast and Scalable Bayesian Deep Learning by Weight-Perturbation in Adam (ICML 2018 · similarity 0.89)
- Bayesian Online Natural Gradient (BONG) (NeurIPS 2024 · similarity 0.89)
- On the Expressiveness of Approximate Inference in Bayesian Neural Networks (NeurIPS 2020 · similarity 0.89)
- Position: Bayesian Deep Learning is Needed in the Age of Large-Scale AI (ICML 2024 · similarity 0.88)
- VIKING: Deep variational inference with stochastic projections (NeurIPS 2025 · similarity 0.88)
- Variational Bayesian Pseudo-Coreset (ICLR 2025 · similarity 0.87)
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.