Python codes for the Structured Recognition non-linear Gaussian Process Factor Analysis model in our NeurIPS 2022 paper: Structured Recognition for Generative Models with Explaining Away.
For any question regarding the paper or the code, please contact changmin.yu98[at]gmail.com and maneesh[at]gatsby.ucl.ac.uk
For executing the codes, try running:
- SR-nlGPFA
python experiments/place_cell.py --model sr-nlgpfa
- treeSRVAE
python treeSRVAE/treeSRVAE.py --dataset bar_test --tree-structured-gen True
- gmmSRVAE
python gmmSRVAE/models/SRVAE.py --dataset pinwheel --full-dependency True --seed 0
If you find the paper or the code helpful for your research, please consider citing us with the following format:
@inproceedings{yustructured,
title={Structured Recognition for Generative Models with Explaining Away},
author={Yu, Changmin and Soulat, Hugo and Burgess, Neil and Sahani, Maneesh},
booktitle={Advances in Neural Information Processing Systems}
}