vdae

python package for variational diffusion autoencoders


License
MIT
Install
pip install vdae==0.0.1

Documentation

Variational Diffusion Autoencoder (VDAE)

diagram

VDAE is a python library that performs spectral clustering with deep neural networks. See https://arxiv.org/abs/1905.12724.

requirements

To run our package, you'll need Python 3.x and the following python packages:

  • scikit-learn
  • tensorflow==1.15.2
  • annoy
  • h5py
  • POT

installing the package

For the most painless installation process, you can run:

pip install vdae

However, to install the most up-to-date version, instead run:

git clone https://github.com/lihenryhfl/SpectralVAEGAN.git; cd SpectralVAEGAN; python setup.py install

usage

For example usage, please see the example script vdae_examples.ipynb.