Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Deep Ensembles via the Neural Tangent Kernel
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 untrainable random function delta to each ensemble member (NTKGP) instead of standard deep ensembles/RP | MLP ensemble · MNIST vs NotMNIST error vs confidence | error at confidence 0.6: over 15% lower error | Helped |
| use NTKGP instead of standard deep ensembles | MLP ensemble · MNIST test (in-distribution) | classification error: around 0.2% increased error for ensemble size 10 | Hurt |
| use NTKGP instead of standard deep ensembles | heteroscedastic MLP ensemble · Flight Delays under dataset shift | NLL | Mixed |
| use NTKGP instead of deep ensembles/RP | heteroscedastic MLP ensemble · Flight Delays combined test | RMSE vs confidence | Helped |
| use NTKGP instead of deep ensembles/RP | Myrtle-10 CNN ensemble · CIFAR-10 vs SVHN error vs confidence | error vs confidence | No clear effect |
| use NTKGP instead of deep ensembles | Myrtle-10 CNN ensemble · CIFAR-10 test (in-distribution) | classification error | Helped |
| use NTKGP-param instead of RP-param at small ensemble size (ablation) | toy 1D regression ensemble · toy 1D regression | mean predictor approximation | Hurt |
Closest papers by meaning
- Exploring the Uncertainty Properties of Neural Networks’ Implicit Priors in the Infinite-Width Limit (ICLR 2021 · similarity 0.89)
- Disentangling the Predictive Variance of Deep Ensembles through the Neural Tangent Kernel (NeurIPS 2022 · similarity 0.89)
- Critical feature learning in deep neural networks (ICML 2024 · similarity 0.88)
- Theoretical Limitations of Ensembles in the Age of Overparameterization (ICML 2025 · similarity 0.87)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.86)
- Continual learning with the neural tangent ensemble (NeurIPS 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.