FoKL

Karhunen LoÃ¨ve decomposed Gaussian processes with forward variable selection. Use this package for scalable GP regression and fast inference on static and dynamic datasets.

Quick Start

Install

pip install FoKL

Usage

For instructions on how to use the package, please see documentation in 'emulator.py'

Citing FoKL

If you use FoKL, please cite the following paper:

K. Hayes, M.W. Fouts, A. Baheri and D.S. Mebane, "Forward variable selection enables fast and accurate dynamic system identification with Karhunen-LoÃ¨ve decomposed Gaussian processes", arXiv:2205.13676

@misc{hayes2023forward,
      title={Forward variable selection enables fast and accurate dynamic system identification with Karhunen-Lo\`eve decomposed Gaussian processes}, 
      author={Kyle Hayes and Michael W. Fouts and Ali Baheri and David S. Mebane},
      year={2023},
      eprint={2205.13676},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

Contributors

David Mebane (ideas and original code)
Kyle Hayes (integrator)
Derek Slack (Python porting)

Funding

Funding provided by National Science Foundation, Award No. 2119688