Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Scalable Training of Inference Networks for Gaussian-Process Models
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 GPNet instead of SVGP with 500 inducing points | GP regression inference · Kin8nm | RMSE: 0.074$\pm$0.000 -> 0.067$\pm$0.000 | Helped |
| Use GPNet instead of SVGP with 500 inducing points | GP regression inference · Energy | Test log likelihood: -0.600$\pm$0.033 -> -0.847$\pm$0.013 | Hurt |
| Use stochastic functional mirror descent for minibatch training instead of the FBNN heuristic objective | RFE inference network · Kin8nm | RMSE: 0.072$\pm$0.000 -> 0.067$\pm$0.000 | Helped |
| Use GPNet instead of FBNN minibatch training | RFE inference network · Airline Delay | RMSE: 24.114 -> 23.675 | Helped |
| Use GPNet with a CNN-GP prior instead of Conv GP | CNN-GP classification · CIFAR10 | Test error rate: 35.4\% -> 24.63\% | Helped |
| Update GP hyperparameters during training (ablation) | FBNN, M=100 · Airline Delay | RMSE: 27.186 | Hurt |
| Increase measurement-point count from M=2 or 5 to M=20 (ablation) | GPNet · Snelson synthetic dataset | prediction accuracy | Mixed |
| Increase inducing-point count from M=2 or 5 to M=20 (ablation) | SVGP · Snelson synthetic dataset | prediction accuracy | Helped |
| Use the heuristic minibatch objective instead of the shared global objective (ablation) | FBNN · Snelson synthetic dataset | predictive uncertainty calibration | Hurt |
Closest papers by meaning
- Variational Inference for Gaussian Process Models with Linear Complexity (NeurIPS 2017 · similarity 0.91)
- Parametric Gaussian Process Regressors (ICML 2020 · similarity 0.90)
- Orthogonally Decoupled Variational Gaussian Processes (NeurIPS 2018 · similarity 0.89)
- Turbocharging Gaussian Process Inference with Approximate Sketch-and-Project (NeurIPS 2025 · similarity 0.88)
- Amortized Structured Stochastic Variational Inference for Gaussian Process Latent Variable Models (NeurIPS 2026 · similarity 0.88)
- Neighbour-Driven Gaussian Process Variational Autoencoders for Scalable Structured Latent Modelling (ICML 2025 · 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.