Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Dual Parameterization of Sparse Variational Gaussian Processes
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 the dual parameterization instead of q-SVGP | SVGP · MNIST, E LR 0.040, M LR 0.05, 1 E step and 1 M step | NLPD: 0.304 ± 0.015 -> 0.304 ± 0.006 | No clear effect |
| Use the dual parameterization instead of q-SVGP | SVGP · MNIST, E LR 0.035, M LR 0.10, 2 E steps and 1 M step | NLPD: 0.289 ± 0.010 -> 0.283 ± 0.007 | Helped |
| Use the dual parameterization instead of q-SVGP | SVGP · MNIST, E LR 0.030, M LR 0.10, 3 E steps and 1 M step | NLPD: 0.293 ± 0.020 -> 0.281 ± 0.010 | Helped |
| Use the dual parameterization instead of q-SVGP | SVGP · MNIST, E LR 0.025, M LR 0.03, 4 E steps and 2 M steps | NLPD: 0.259 ± 0.010 -> 0.255 ± 0.006 | Helped |
| Use the dual parameterization instead of q-SVGP | SVGP · MNIST, E LR 0.050, M LR 0.03, 4 E steps and 2 M steps | NLPD: 0.282 ± 0.007 -> 0.283 ± 0.006 | Hurt |
| Use the dual parameterization instead of q-SVGP | SVGP · MNIST, E LR 0.030, M LR 0.03, 4 E steps and 1 M step | NLPD: 0.243 ± 0.003 -> 0.230 ± 0.009 | Helped |
| Use the dual parameterization's direct natural-gradient updates | SVGP · Standard benchmark | wall-clock time: roughly 5 times faster | Helped |
| Use natural-gradient training instead of non-natural-gradient q-SVGP | SVGP · UCI regression and classification tasks | ELBO | Helped |
| Use t-SVGP instead of whitened q-SVGP | SVGP · UCI regression and classification tasks | NLPD | Mixed |
Closest papers by meaning
- Orthogonally Decoupled Variational Gaussian Processes (NeurIPS 2018 · similarity 0.91)
- Variational Inference for Gaussian Process Models with Linear Complexity (NeurIPS 2017 · similarity 0.90)
- A Generalized Stochastic Variational Bayesian Hyperparameter Learning Framework for Sparse Spectrum Gaussian Process Regression (AAAI 2017 · similarity 0.90)
- New Bounds for Sparse Variational Gaussian Processes (ICML 2025 · similarity 0.90)
- SigGPDE: Scaling Sparse Gaussian Processes on Sequential Data (ICML 2021 · similarity 0.90)
- Sparse Gaussian Processes: Structured Approximations and Power-EP Revisited (NeurIPS 2025 · similarity 0.90)
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.