Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Learning under Model Misspecification: Applications to Variational and Ensemble methods
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 second-order PAC² variational learning instead of standard variational inference | Bayesian neural networks · Fashion-MNIST and CIFAR-10 prediction tasks | test NLL | Helped |
| Use the tighter PAC²_T variance bound instead of the PAC² variance bound | PAC² variational learning · Fashion-MNIST and CIFAR-10 prediction tasks | test NLL | Helped |
| Use second-order PAC² ensemble learning instead of a single-model ensemble based on a first-order bound | Ensemble models · Fashion-MNIST and CIFAR-10 prediction tasks | test NLL | Helped |
| Use the tighter PAC²_T variance bound instead of the PAC² variance bound | PAC² ensemble learning · Fashion-MNIST and CIFAR-10 prediction tasks | test NLL | Helped |
| Use a variational approach rather than the MAP model | Variational methods · CIFAR-10 classification | test NLL | No clear effect |
| Use an ensemble variational family represented by mixtures of point masses rather than a standard mean-field variational family | Variational methods · Fashion-MNIST and CIFAR-10 prediction tasks | test NLL | Helped |
| Include a diversity variance term in the ensemble objective | PAC² ensemble learning · Generalization under model misspecification | generalization risk | Helped |
| Remove the diversity variance term from the ensemble objective (ablation) | First-order PAC-Bayes ensemble learning · Ensemble model averaging | generalization risk | Hurt |
Closest papers by meaning
- Variational Bayes under Model Misspecification (NeurIPS 2019 · similarity 0.88)
- Dissecting Non-Vacuous Generalization Bounds based on the Mean-Field Approximation (ICML 2020 · similarity 0.88)
- A PAC-Bayesian Generalization Bound for Equivariant Networks (NeurIPS 2022 · similarity 0.88)
- CBMA: Improving Conformal Prediction through Bayesian Model Averaging (ICLR 2025 · similarity 0.86)
- Recursive PAC-Bayes: A Frequentist Approach to Sequential Prior Updates with No Information Loss (NeurIPS 2024 · similarity 0.86)
- Generalization Bounds via Meta-Learned Model Representations: PAC-Bayes and Sample Compression Hypernetworks (ICML 2025 · 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.