\# Exponential-wrapped normal distributions on symmetric positive definite matrices and Grassmannians
This code performs the classification experiments described in the paper:
The files Grassmannian_2_4.py and SPD_2.py perform respectively a classification experiment using exponential-wrapped normal distributions on :
\- the Grassmannian of planes on R^4 (Grassmannian_2_4.py)
\- 2 by 2 symmetric positive definite matrices (SPD_2.py)
In both cases, points are sampled from 4 equiprobable classes with exponential-wrapped normal distributions. For each class, we estimate an exponential-wrapped normal distributions using a moment estimator. The classification is then performed using maximum a posteriori probabilities.
The code uses the python package Geomstats: