Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Neural Network Priors Revisited
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 heavy-tailed Laplace or Student-t weight priors instead of an isotropic Gaussian | Bayesian FCNN · MNIST | test error | Helped |
| Use heavy-tailed Laplace or Student-t weight priors instead of an isotropic Gaussian | Bayesian FCNN · FashionMNIST | test error | Helped |
| Use heavy-tailed priors instead of an isotropic Gaussian | Bayesian FCNN · MNIST and FashionMNIST | cold-posterior effect on predictive performance | Helped |
| Use a correlated Gaussian prior with spatial covariance within convolutional filters instead of an isotropic Gaussian | Bayesian CNN · MNIST | predictive performance | Helped |
| Use a correlated Gaussian prior with spatial covariance within convolutional filters instead of an isotropic Gaussian | Bayesian CNN · FashionMNIST | predictive performance | Helped |
| Use a correlated Gaussian prior with spatial covariance within convolutional filters instead of an isotropic Gaussian | Bayesian CNN · MNIST and FashionMNIST | cold-posterior effect | No clear effect |
| Use a correlated Gaussian prior with spatial covariance within convolutional filters instead of an isotropic Gaussian | Bayesian ResNet20 · CIFAR-10 | predictive performance | Helped |
| Use a correlated Gaussian prior with spatial covariance within convolutional filters instead of an isotropic Gaussian | Bayesian ResNet20 · CIFAR-10 | cold-posterior effect | Hurt |
| Use a correlated Gaussian prior instead of heavy-tailed priors | Bayesian CNN and ResNet · MNIST, FashionMNIST, and CIFAR-10 | predictive performance | Helped |
| Use a heavy-tailed prior instead of Gaussian or correlated Gaussian priors | Bayesian CNN · MNIST and FashionMNIST | predictive performance | Hurt |
| Use a heavy-tailed prior instead of an isotropic Gaussian | Bayesian FCNN · MNIST and FashionMNIST | calibration | Mixed |
| Use a heavy-tailed prior instead of an isotropic Gaussian | Bayesian FCNN · MNIST and FashionMNIST | OOD detection | Mixed |
Closest papers by meaning
- Understanding Priors in Bayesian Neural Networks at the Unit Level (ICML 2019 · similarity 0.90)
- A statistical theory of cold posteriors in deep neural networks (ICLR 2021 · similarity 0.90)
- Hierarchical Gaussian Process Priors for Bayesian Neural Network Weights (NeurIPS 2020 · similarity 0.90)
- Exact marginal prior distributions of finite Bayesian neural networks (NeurIPS 2021 · similarity 0.89)
- A Symmetry-Aware Exploration of Bayesian Neural Network Posteriors (ICLR 2024 · similarity 0.88)
- Beyond IID weights: sparse and low-rank deep Neural Networks are also Gaussian Processes (ICLR 2024 · 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.