Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Flat Seeking Bayesian Neural Networks
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 |
|---|---|---|---|
| make SWAG sharpness-aware (F-SWAG) | SWAG · CIFAR-100 PreResNet-164 | accuracy: 79.90 -> 80.93 | Helped |
| make Deep-ens sharpness-aware | Deep-ens · CIFAR-100 WideResNet28x10 | accuracy: 83.04 -> 84.52 | Helped |
| make MC-Dropout sharpness-aware | MC-Dropout · CIFAR-100 PreResNet-164 | NLL: 0.9162 -> 0.7027 | Helped |
| make MC-Dropout sharpness-aware | MC-Dropout · CIFAR-100 PreResNet-164 | ECE: 0.0993 -> 0.0514 | Helped |
| make SGLD sharpness-aware | SGLD · CIFAR-10 WideResNet28x10 | accuracy: 95.87 -> 96.43 | Helped |
| make SWAG sharpness-aware | SWAG · CIFAR-100 PreResNet-164 | NLL: 0.6595 -> 0.6704 | Hurt |
| make SWAG sharpness-aware | SWAG · CIFAR-10 PreResNet-164 | ECE: 0.0053 -> 0.0056 | Hurt |
| add geometry term to sharpness-aware posterior (ablation) | F-SGVB · CIFAR-10 Resnet10 | accuracy: 80.60 -> 82.05 | Helped |
| add geometry term to sharpness-aware posterior (ablation) | F-SGVB · CIFAR-100 Resnet18 | accuracy: 61.54 -> 62.58 | Helped |
| add geometry term to sharpness-aware posterior (ablation) | F-SGVB-LRT · CIFAR-100 Resnet18 | ECE: 0.022 -> 0.036 | Hurt |
| make SWAG sharpness-aware | SWAG · CIFAR-10-C average | ECE: 0.0454 -> 0.0200 | Helped |
| make SWAG sharpness-aware | SWAG · CIFAR-10-C average | accuracy: 85.35 -> 85.58 | No clear effect |
Closest papers by meaning
- Collapsed Inference for Bayesian Deep Learning (NeurIPS 2023 · similarity 0.89)
- A Symmetry-Aware Exploration of Bayesian Neural Network Posteriors (ICLR 2024 · similarity 0.87)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.87)
- Make Me a BNN: A Simple Strategy for Estimating Bayesian Uncertainty from Pre-trained Models (CVPR 2024 · similarity 0.87)
- Entropy-MCMC: Sampling from Flat Basins with Ease (ICLR 2024 · similarity 0.87)
- Training Bayesian Neural Networks with Sparse Subspace Variational Inference (ICLR 2024 · similarity 0.86)
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.