https://github.com/devkihyun/fcnae-tensorflow

Tensorflow implement of Stacked Autoencoder using Fully Convlolutional Network(FCN)
https://github.com/devkihyun/fcnae-tensorflow

Last synced: 6 days ago
JSON representation

Tensorflow implement of Stacked Autoencoder using Fully Convlolutional Network(FCN)

• Host: GitHub
• URL: https://github.com/devkihyun/fcnae-tensorflow
• Owner: DevKiHyun
• Created: 2018-09-21T15:52:42.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-11-12T00:20:59.000Z (about 6 years ago)
• Last Synced: 2024-07-25T11:10:34.663Z (5 months ago)
• Language: Python
• Size: 16.6 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# FCNAE-Tensorflow (2018/09/21)

## Introduction

I implement a tensorflow model of a Staked Autoencoder using Fully Convolutional Network. 

- I use mnist dataset as training dataset.

## Environment

- Ubuntu 16.04

- Python 3.5

## Depenency

- Numpy

- matplotlib

## Files

- fcnae.py : Model definition.

- main.py : Train the model and pass the default value.

## How to use

### Training

```shell

python main.py

# Default args: training_epoch = 200, batch_size = 128, n_layers = 3, stride = 2, learning_rate = 0.0001

# You can change args: training_epoch = 300, batch_size = 64, n_layers = 2, stride = 4, learning_rate = 0.0005

python main.py --training_epoch 300 --batch_size 64 --n_layers 2 --stride 4 --learning_rate 0.0005

```