https://github.com/musyoku/adgm

Chainer implementation of Auxiliary Deep Generative Models (ADGM) and Skip Deep Generative Model (SDGM)
https://github.com/musyoku/adgm

Last synced: 9 days ago
JSON representation

Chainer implementation of Auxiliary Deep Generative Models (ADGM) and Skip Deep Generative Model (SDGM)

• Host: GitHub
• URL: https://github.com/musyoku/adgm
• Owner: musyoku
• Created: 2016-08-10T10:02:46.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2016-12-26T04:16:14.000Z (almost 8 years ago)
• Last Synced: 2024-08-02T12:21:59.386Z (3 months ago)
• Language: Python
• Homepage:
• Size: 103 KB
• Stars: 6
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
## Auxiliary Deep Generative Models

Chainer implementation of the papers:

- [Auxiliary Deep Generative Models](http://arxiv.org/abs/1602.05473)

- [Improving Stochastic Gradient Descent with Feedback](https://arxiv.org/abs/1611.01505)

- [Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks](https://arxiv.org/abs/1602.07868)

- [Categorical Reparameterization with Gumbel-Softmax](https://arxiv.org/abs/1611.01144)

[この記事](http://musyoku.github.io/2016/09/10/Auxiliary-Deep-Generative-Models/)で実装したコードです。

See also:

- [VAE](https://github.com/musyoku/variational-autoencoder)

- [AAE](https://github.com/musyoku/adversarial-autoencoder)

### Requirements

- Chainer 1.17

- Pylab

- pandas

this code contains following repos:

- [chainer-sequential](https://github.com/musyoku/chainer-sequential)

## Runnning

### ADGM

run `train_adgm/train.py`

### SDGM

run `train_sdgm/train.py`

## Validation Accuracy

![acuracy](http://musyoku.github.io/images/post/2016-09-10/adgm_graph.png)

## Analogies

### ADGM

![analogy](http://musyoku.github.io/images/post/2016-09-10/analogy_adgm.png)

### SDGM

![analogy](http://musyoku.github.io/images/post/2016-09-10/analogy_sdgm.png)