This is code for the paper "Convolutional Poisson Gamma Belief Network" published in ICML2019.

Created by Chaojie Wang , Bo Chen , Sucheng Xiao at Xidian University and Mingyuan Zhou at University of Texas at Austin. https://arxiv.org/abs/1905.05394

Tensorflow >= 1.0

PyCUDA >= 0.8

PyCUDA can be download from following address https://mathema.tician.de/software/pycuda/

All data source files can be found in following addresses and have been included in our repository.

• MR: https://www.cs.cornell.edu/people/pabo/movie-review-data/

• TREC: http://cogcomp.cs.illinois.edu/Data/QA/QC/

• SUBJ: http://www.cs.cornell.edu/people/pabo/movie-review-data/

• ELEC: https://github.com/riejohnson/ConText

• IMDB: http://ai.stanford.edu/~amaas/data/sentiment/

• CPFA_Mnist_Demo folder contains 4 different training algorithms for CPFA, including Toeplitz, Element, Element-Parallel and SGMCMC mehthods.

• CPGBN_Text_Demo folder contains Datasets and experiment code to reproduce the results in our paper.

• CPGBN_Derivation_Draft file provides a detailed derivation for the CPGBN.

If you find that the algorithms in this repository are useful for your research, please refer to the following article:

@inproceedings{CPGBN_ICML2019,
title={{C}onvolutional {P}oisson {G}amma {B}elief {N}etwork},
author={Chaojie Wang and Bo Chen and Sucheng Xiao and Mingyuan Zhou}, booktitle={ICML}, year={2019}}

Contact Bo Chen bchen@mail.xidian.edu.cn or Chaojie Wang xd_silly@163.com

Copyright (c), 2018, Chaojie Wang xd_silly@163.com