Python code (tested on 2.7) for phase and fourier neural network in distrbution regression, regression or classification problems implemented in TensorFlow. Some part of this code was adapted from https://github.com/oxmlcs/kerpy.
The method is described in:
Testing and Learning on Distributions with Symmetric Noise Invariance (Link to be updated)
The code tutorial_dr_aerosol.py demonstrate the use of the phase and fourier neural network on distributional regression with the MISR1 aerosol dataset. The code tutotial_toy.py demonstrates the use of the fourier neural network on a toy classification problem.
The MISR1 dataset can be found in MISR1.mat here or at http://www.dabi.temple.edu/~vucetic/MIR.html.