RyanJDick / DRAW-hard

Implementation of the Deep Recurrent Attentive Writer (DRAW) using 'hard' attention.

DRAW-hard

TensorFlow implementation of DRAW: A Recurrent Neural Network For Image Generation with 'hard' attention.

Useful Resources

This project used the following base implmentation of DRAW and corresponding blog post as a starting point:

• https://github.com/ericjang/draw
• http://blog.evjang.com/2016/06/understanding-and-implementing.html

Citations

@article{DBLP:journals/corr/GregorDGW15,
author    = {Karol Gregor and
             Ivo Danihelka and
             Alex Graves and
             Daan Wierstra},
title     = {{DRAW:} {A} Recurrent Neural Network For Image Generation},
journal   = {CoRR},
volume    = {abs/1502.04623},
year      = {2015},
url       = {http://arxiv.org/abs/1502.04623},
archivePrefix = {arXiv},
eprint    = {1502.04623},
timestamp = {Wed, 07 Jun 2017 14:40:10 +0200},
biburl    = {https://dblp.org/rec/bib/journals/corr/GregorDGW15},
bibsource = {dblp computer science bibliography, https://dblp.org}
}