Skip to content

Low-variance and unbiased gradient for backpropagation through categorical random variables, with application in variational auto-encoder and reinforcement learning. ICML 2019

ARM-gradient/ARSM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

37 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Code to show the simulation results in [ARSM: Augment-REINFORCE-Swap-Merge Estimator for Gradient Backpropagation Through Categorical Variables]

Data sets

The MNIST data is self-contained and the Omniglot data is in the repository.
RL data is from OpenAI Gym

Citations

Below are the paper to cite if you find the algorithms in this repository useful in your own research:

@inproceedings{ARSM_ICML2019,
title={{ARSM}: {A}ugment-{REINFORCE}-swap-merge estimator for gradient backpropagation through categorical variables},
author={Mingzhang Yin and Yuguang Yue and Mingyuan Zhou}, booktitle={ICML}, year={2019} }

License Info

This code is offered under the MIT License.

About

Low-variance and unbiased gradient for backpropagation through categorical random variables, with application in variational auto-encoder and reinforcement learning. ICML 2019

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages