This repository provides a reference implementation for learning uncertainty autoencoders as described in the paper:
Uncertainty Autoencoders: Learning Compressed Representations via Variational Information Maximization Aditya Grover, Stefano Ermon
International Conference on Artificial Intelligence and Statistics (AISTATS), 2019 Paper: https://arxiv.org/abs/1807.01442
The codebase is implemented in Python 3.6 and Tensorflow. To install the necessary requirements, run the following commands:
pip install -r requirements.txt
** NOTE: ** An experimental reimplementation is available in pytorch_src/ folder. Use at own risk.
If you find Uncertainty Autoencoders useful in your research, please consider citing the following paper:
@inproceedings{grover2019uncertainty,
title={Uncertainty Autoencoders: Learning Compressed Representations via Variational Information Maximization},
author={Grover, Aditya and Ermon, Stefano},
booktitle={International Conference on Artificial Intelligence and Statistics},
year={2019}}