Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Kernel Packet: An Exact and Scalable Algorithm for Gaussian Process Regression with Matérn Correlations
Haoyuan Chen, Liang Ding, Rui Tuo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 KP Matérn-3/2 instead of inducing points | GP regression · CO2 interpolation | training time (sec): 0.28 -> 0.18 | Improved |
| use KP Matérn-3/2 instead of RFF | GP regression · CO2 interpolation | prediction time (1e-3 sec): 3.34 -> 2.84 | Improved |
| use KP Matérn-3/2 instead of inducing points | GP regression · Stock price regression | prediction time (1e-3 sec): 9.88 -> 2.83 | Improved |
| use KP Matérn-3/2 instead of RFF | GP regression · Stock price regression | training time (sec): 0.50 -> 0.37 | Improved |
| use KP instead of inducing points and RFF | GP regression · CO2 interpolation | interpolation quality | Improved |
| use KP instead of inducing points and RFF | GP regression · Stock price regression | trend reconstruction | Improved |
| use KP instead of laGP, inducing points and RFF on full grids | 2D sin function full grid GP · Full grid design | MSE and computational time: KP up to 2^26 grid points; laGP breaks down above 2^16 | Improved |
| use KP instead of the direct method on sparse grids | GP MLE-predictor, Griewank d=10 and 20 · Sparse grid design | computational time: at least twice faster above 10^5 training points | Improved |
| use KP instead of the direct method on sparse grids | GP MLE-predictor, Griewank d=10 and 20 · Sparse grid design | MSE: differences within ±10^{-10} | No clear effect |
| use Matérn-5/2 instead of Matérn-3/2 in KP | KP · Stock price regression | training time (sec): 0.37 -> 0.44 | Hurt |
Closest papers by meaning
- Gaussian process regression: Optimality, robustness, and relationship with kernel ridge regression (JMLR 2022 · similarity 0.89)
- Log-Linear-Time Gaussian Processes Using Binary Tree Kernels (NeurIPS 2022 · similarity 0.88)
- KernelMatmul: Scaling Gaussian Processes to Large Time Series (AAAI 2025 · similarity 0.87)
- gp2Scale: A Class of Compactly Supported Non-Stationary Kernels and Distributed Computing for Exact Gaussian Processes on 10 Million Data Points (ICML 2026 · similarity 0.87)
- Multiresolution Kernel Approximation for Gaussian Process Regression (NeurIPS 2017 · similarity 0.87)
- Solving High Frequency and Multi-Scale PDEs with Gaussian Processes (ICLR 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.
