Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
A general linear-time inference method for Gaussian Processes on one dimension
Jackson Loper, David Blei, John P. Cunningham, Liam Paninski
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 LEG kernel fitted by gradient descent instead of Pade / scale-mixture state-space proposals | state-space approximation of popular kernels · kernel approximation | approximation error: most cases lower error; rank-12 no better than Pade approximant | Mixed |
| Increase LEG rank from 7 to 13 for triangle kernel | LEG kernel approximation · triangle kernel | uniform error within .01: rank 7 -> rank 13 needed for .01 | Improved |
| Use LEG instead of Celerite family for kernel estimation | maximum-likelihood kernel estimator · simulated GP data | kernel estimation accuracy | Improved |
| Use LEG instead of spectral mixture kernels | maximum-likelihood kernel estimator · simulated GP data | kernel estimation accuracy | No clear effect |
| Run on GPU instead of many CPU cores | leggps likelihood and posterior variance computation · speed benchmark (Q=5, D=4) | runtime | Improved |
| Use cyclic reduction (leggps) instead of associative scan (Pyro GaussianHMM) | state-space gradient computation · speed benchmark | runtime | No clear effect |
| Use single-threaded pylds instead of leggps with fewer than 200 observations | posterior variance computation · speed benchmark | runtime | Improved |
Closest papers by meaning
- State Space Gaussian Processes with Non-Gaussian Likelihood (ICML 2018 · similarity 0.92)
- Scalable Exact Inference in Multi-Output Gaussian Processes (ICML 2020 · similarity 0.91)
- Infinite-Horizon Gaussian Processes (NeurIPS 2018 · similarity 0.91)
- Computation-Aware Gaussian Processes: Model Selection And Linear-Time Inference (NeurIPS 2024 · similarity 0.89)
- Turbocharging Gaussian Process Inference with Approximate Sketch-and-Project (NeurIPS 2025 · similarity 0.89)
- KernelMatmul: Scaling Gaussian Processes to Large Time Series (AAAI 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.
