Research map / Learning theory and algorithms / Sample complexity
Tighter Bounds on the Log Marginal Likelihood of Gaussian Process Regression Using Conjugate Gradients
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 CGLB for hyperparameter selection instead of Iterative GP | Gaussian process regression · bike, elevators, poletele, kin40k, protein | predictive RMSE and NLPD: bike RMSE 0.014 -> 0.004, NLPD -2.016 -> -4.216; elevators RMSE 0.353 -> 0.354, NLPD 0.407 -> 0.386; poletele RMSE 0.079 | Mixed |
| use CGLB for hyperparameter selection instead of SGPR | Gaussian process regression · bike, elevators, poletele, kin40k, protein | predictive RMSE and NLPD: bike RMSE 0.010 -> 0.004, NLPD -3.280 -> -4.216; elevators RMSE 0.354 -> 0.354, NLPD 0.386 -> 0.386; poletele RMSE 0.078 | Helped |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · bike | RMSE: 0.004 -> 0.003 | Helped |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · bike | NLPD: -4.216 -> -3.811 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · elevators | NLPD: 0.386 -> 0.387 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · elevators | RMSE: 0.354 -> 0.355 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · poletele | NLPD: -1.201 -> -1.126 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · poletele | RMSE: 0.077 -> 0.080 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · kin40k | NLPD: -0.919 -> -0.826 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · kin40k | RMSE: 0.086 -> 0.088 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · protein | NLPD: 0.749 -> 0.771 | Hurt |
| use 2048 instead of 4096 inducing points (ablation) | CGLB · protein | RMSE: 0.522 -> 0.529 | Hurt |
Closest papers by meaning
- New Bounds for Sparse Variational Gaussian Processes (ICML 2025 · similarity 0.90)
- Preconditioning for Scalable Gaussian Process Hyperparameter Optimization (ICML 2022 · similarity 0.90)
- Variational Inference for Gaussian Process Models with Linear Complexity (NeurIPS 2017 · similarity 0.89)
- Sparse Variational Student-t Processes (AAAI 2024 · similarity 0.87)
- Robust and Conjugate Gaussian Process Regression (ICML 2024 · similarity 0.87)
- Robust Gaussian Processes via Relevance Pursuit (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.