No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
data
notebooks
.gitignore
LICENSE
README.md
SymInvBagKernel.py
__init__.py
aux_fct.py
fourier_nn.py
kernel_layer.py
phase_fourier_dr_nn.py

README.md

phase_learn

Python code (tested on 2.7) for distribution regression using features based on empirical phase functions.

The method is described in: H. C. L. Law, C. Yau, and D. Sejdinovic, Testing and Learning on Distributions with Symmetric Noise Invariance, in Advances in Neural Information Processing Systems (NIPS), 2017. arxiv

The phase features with ridge regression are incorporated as a part of the kerpy package.

Notebooks

  • The notebook tutorial_toy.ipynb demonstrates the use of the neural network implementations for learning Fourier features on a toy classification problem.
  • The notebook Phase_Fourier_ridge_tutorial.ipynb demonstrates the use of the phase ridge regression (using kerpy) with the MISR1 aerosol dataset.
  • Notebook Phase_Neural_Network_Aerosol.ipynb demonstrates neural network implementations for learning both phase and Fourier features on the MISR1 aerosol dataset.

The MISR1 dataset in MISR1.mat is originally from http://www.dabi.temple.edu/~vucetic/MIR.html.