This is a PyTorch implementation of ClusterGAN, an approach to unsupervised clustering using generative adversarial networks.

The package as well as the necessary requirements can be installed by running make -f Makefile or via

virtualenv -p /usr/local/bin/python3 venv
source venv/bin/activate
python setup.py install

To run ClusterGAN on the MNIST dataset, ensure the package is setup and then run

python train.py -r test_run -s mnist -b 64 -n 300

where a directory runs/mnist/test_run will be made and contain the generated output (models, example generated instances, training figures) from the training run. The -r option denotes the run name, -s the dataset (currently MNIST and Fashion-MNIST), -b the batch size, and -n the number of training epochs.

Below is an example set of training curves for 200 epochs, batch size of 64 on the MNIST dataset.

To generate examples from randomly sampled latent space variables,

python gen-examples.py -r test_run -s mnist -b 100

Here are some example generated images by specified class (each row) of the learned labels in latent space.

To produce a TSNE figure depicting the clustering of the latent space encoding of real images,

python tsne-cluster.py -r test_run -s mnist

Below is the tSNE clustering figure of the latent space vectors of true MNIST images fed into the encoder.

MIT License

Copyright (c) 2018 Zigfried Hampel-Arias