The Wasserstein-Fourrier is a distance between time series. The 'toy_data.ipynb' illustrate basic properties of the distance on synthetic datasets.
Machine Learning models could benefit from the rich geometry induced by the WF distance. 'softmax.py' is an implementation of Softmax regression equipped with the WF distance.
The notebooks '2_experiment_UrbanSound.ipynb' and '3_Activity_Recognition.ipynb' show how to operate the softmax regression algorithm on two real world datasets.