Repository from our work in a variational auto-encoder with discrete (binary) variables.

• Using Gumbel-Max trick (a soft version)
• Binary VAE: Hashing (Done)
  • Good results show that Binary VAE outperforms Traditional VAE on Hashing
• Categorical VAE: Classes

• Preliminary work in a conference
  • 🔓 Final published version
  • 🔓 Preprint
  • Presentation
• Extended work in a journal
  • 🔒 Final published version
  • 🔓 Preprint

Mena, Francisco and Ñanculef, Ricardo. "A binary variational autoencoder for hashing". Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications: 24th Iberoamerican Congress, CIARP 2019, Havana, Cuba, October 28-31, 2019, Proceedings 24, 2019

@inproceedings{mena2019binary,
  title={A binary variational autoencoder for hashing},
  author={Mena, Francisco and {\~N}anculef, Ricardo},
  booktitle={Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications: 24th Iberoamerican Congress, CIARP 2019, Havana, Cuba, October 28-31, 2019, Proceedings 24},
  pages={131--141},
  year={2019},
  organization={Springer},
  doi={10.1007/978-3-030-33904-3_12}
}

this is a reference of our initial work with text data

Mena, Francisco, Ñanculef, Ricardo, and Valle, Carlos. "Interpretable and effective hashing via Bernoulli variational auto-encoders". Intelligent Data Analysis, 24(S1), 141-166. 2020

@article{mena2020interpretable,
  title={Interpretable and effective hashing via Bernoulli variational auto-encoders},
  author={Mena, Francisco and {\~N}anculef, Ricardo and Valle, Carlos},
  journal={Intelligent Data Analysis},
  volume={24},
  number={S1},
  pages={141--166},
  year={2020},
  publisher={IOS Press},
  doi={10.3233/IDA-200013}
}

this is a reference of our extended work with text+image data