Generalised-Variogram-Method

This repository contains the code used in

Felipe Tobar, Elsa Cazelles and Taco de Wolff. Computationally-efficient initialisation of GPs: The generalised variogram method. Transactions of Machine Learning Research, in press, 2023. OpenReview & arXiv

This repository contains the following files:

The files waflgp.py and gpinit.py defines the class, constructor and methods for the spectral and temporal distances respectively.
The file utils.py, which contains simple auxiliary functions
The Jupyer Notebook Exp0_minimal_ex, a minimal working example of our toolbox
Notebooks Exp1, Exp2, Exp3, Exp4, Exp5, Exp6, Exp7 replicate experiments in the paper.
The data used for the experiments (heart-rate and audio)

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
Data		Data
free-spoken-digit-dataset-1.0.9/recordings		free-spoken-digit-dataset-1.0.9/recordings
revised Exp4		revised Exp4
revised Exp7		revised Exp7
Exp0 _minimal_ ex.ipynb		Exp0 _minimal_ ex.ipynb
Exp1a_reduction_in_time fixed-seed.ipynb		Exp1a_reduction_in_time fixed-seed.ipynb
Exp1b_reduction_in_time free-seed.ipynb		Exp1b_reduction_in_time free-seed.ipynb
Exp2 - Effect of periodogram.ipynb		Exp2 - Effect of periodogram.ipynb
Exp3 - computational complexity.ipynb		Exp3 - computational complexity.ipynb
Exp5 - audio signal.ipynb		Exp5 - audio signal.ipynb
Exp6 - fitting mixtures.ipynb		Exp6 - fitting mixtures.ipynb
LICENSE		LICENSE
README.md		README.md
utils.py		utils.py
waflgp.py		waflgp.py

Provide feedback