An Entangled Mixture of Variational Autoencoders Approach to Deep Clustering

This repository includes the accompanying code for the paper "An Entangled Mixture of Variational Autoencoders Approach to Deep Clustering". Avi Caciularu and Jacob Goldberger.

We heavily relied on the code from the K-Autoencoders Deep Clustering repository. The instructions are based on the K-Autoencoders Deep Clustering repository, and are modified to support our version.

How to Use?

• Clone this repository.
• cd to the cloned dir
• run conda create -n k_dae python=3.6
• run conda activate k_dvae
• run pip install -r path\to\requirements.txt

Run Example:

• python main.py -dn mnist

Optional Args:

• --dataset_name - The name of dataset [mnist / fashion / usps] default: mnist
• --save_dir - path to output folder. (contains logs and model.)

Contact

For additional data or code please contact Avi Caciularu.

Citation

If you find our work useful, please cite the paper as:

@article{caciularu2022entangledmixture,
      title={An Entangled Mixture of Variational Autoencoders Approach to Deep Clustering},
      author={Caciularu, Avi and Goldberger, Jacob},
      journal={Neurocomputing},
      year={2023},
}