louis-pujol / isde-paper Goto Github PK

This folder contains code to reproduce experiments presented in the paper ISDE : Independence Structure Density Estimation. It contains three notebooks :


- Synthetic data.ipynb: to reproduce expermients of section 4.1
- Real data - loglikelihood.ipynb : to reproduce experiments of section 4.2.1
- Qualitative analysis.ipynb : to reproduce experiments of section 4.2.2

In the folder includes you can find the source code for the execution of density estimation algorithms FDE, BDCS, ISDE and GM in separate files
and a file containing useful functions for qualitative analysis.

To be executed it requires the following packages :
- numpy
- scipy
- scikit-learn
- matplotlib
- seaborn
- PuLP (https://pypi.org/project/PuLP/)
- pyKeOps (https://www.kernel-operations.io/keops/python/installation.html)
- more-itertools (https://pypi.org/project/more-itertools/)