Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Tackling covariate shift with node-based 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 |
|---|---|---|---|
| add entropy regularization (gamma-ELBO) to node-based BNN latent variables, larger gamma (ablation) | node-based BNN (VGG16, CIFAR100) · CIFAR100 corrupted test (OOD) | OOD performance | Mixed |
| increase number of Gaussian mixture components K from 1 to 4 in variational posterior (ablation) | node-based BNN (VGG16, CIFAR100) · CIFAR100 corrupted test (OOD) | OOD performance | Helped |
| use only output latent variables instead of input latent variables (ablation) | node-based BNN (VGG16, CIFAR100) · CIFAR100 corrupted test (OOD) | OOD performance | Helped |
| add entropy regularization (high gamma) to node-based BNN (ablation) | node-based BNN (ResNet18, CIFAR10 with noisy labels) · CIFAR10 with 20/40/80% noisy training labels | NLL on noisy vs clean labels | Helped |
| node-based BNN vs SWAG on CIFAR10 | node-based BNN (ResNet18) · CIFAR10 corrupted test | NLL/error | Hurt |
| node-based BNN vs baselines on CIFAR100 ID data ECE | node-based BNN (ResNet18) · CIFAR100 clean (ID) | ECE | Hurt |
| node-based BNN (vs baselines) under increasing corruption | node-based BNN (ResNet18, CIFAR100) · CIFAR100 corrupted test (OOD) | ECE | Helped |
| node-based BNN training cost vs SWAG | node-based BNN · runtime | training time | Hurt |
Closest papers by meaning
- Unlabelled Data Improves Bayesian Uncertainty Calibration under Covariate Shift (ICML 2020 · similarity 0.90)
- Quantifying Uncertainty in the Presence of Distribution Shifts (NeurIPS 2025 · similarity 0.89)
- Dangers of Bayesian Model Averaging under Covariate Shift (NeurIPS 2021 · similarity 0.89)
- Bayesian Adaptation for Covariate Shift (NeurIPS 2021 · similarity 0.88)
- Make Me a BNN: A Simple Strategy for Estimating Bayesian Uncertainty from Pre-trained Models (CVPR 2024 · similarity 0.88)
- PAC Prediction Sets Under Covariate Shift (ICLR 2022 · 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.