• 设为首页
  • 点击收藏
  • 手机版
    手机扫一扫访问
    迪恩网络手机版
  • 关注官方公众号
    微信扫一扫关注
    迪恩网络公众号

jonzhaocn/GAN-Base-on-Matlab: A implementation of GAN in matlab

原作者: [db:作者] 来自: 网络 收藏 邀请

开源软件名称(OpenSource Name):

jonzhaocn/GAN-Base-on-Matlab

开源软件地址(OpenSource Url):

https://github.com/jonzhaocn/GAN-Base-on-Matlab

开源编程语言(OpenSource Language):

MATLAB 97.6%

开源软件介绍(OpenSource Introduction):

GAN-Base-on-Matlab

data

mnist_uint8

download link

Example

example_1

  • network structure:
generator.layers = {
    struct('type', 'input', 'output_shape', [100, batch_size]) 
    struct('type', 'fully_connect', 'output_shape', [3136, batch_size], 'activation', 'leaky_relu')
    struct('type', 'reshape', 'output_shape', [7,7,64, batch_size])
    struct('type', 'conv2d_transpose', 'output_shape', [14, 14, 32, batch_size], 'kernel_size', 5, 'stride', 2, 'padding', 'same', 'activation', 'leaky_relu')
    struct('type', 'conv2d_transpose', 'output_shape', [28, 28, 1, batch_size], 'kernel_size', 5, 'stride', 2, 'padding', 'same', 'activation', 'sigmoid')
};
discriminator.layers = {
    struct('type', 'input', 'output_shape', [28, 28, 1, batch_size])
    struct('type', 'conv2d', 'output_maps', 32, 'kernel_size', 5, 'padding', 'same', 'activation', 'leaky_relu')
    struct('type', 'sub_sampling', 'scale', 2)
    struct('type', 'conv2d', 'output_maps', 64, 'kernel_size', 5, 'padding', 'same', 'activation', 'leaky_relu')
    struct('type', 'sub_sampling', 'scale', 2)
    struct('type', 'reshape', 'output_shape', [3136, batch_size])
    struct('type', 'fully_connect', 'output_shape', [1, batch_size], 'activation', 'sigmoid')
};
  • result:

example_2

  • network structure:
generator.layers = {
    struct('type', 'input', 'output_shape', [100, batch_size]) 
    struct('type', 'fully_connect', 'output_shape', [1024, batch_size], 'activation', 'relu')
    struct('type', 'fully_connect', 'output_shape', [28*28, batch_size], 'activation', 'sigmoid') 
    struct('type', 'reshape', 'output_shape', [28, 28, 1, batch_size])
};
discriminator.layers = {
    struct('type', 'input', 'output_shape', [28,28,1, batch_size])
    struct('type', 'reshape', 'output_shape', [28*28, batch_size]) 
    struct('type', 'fully_connect', 'output_shape', [1024, batch_size], 'activation', 'relu')
    struct('type', 'fully_connect', 'output_shape', [1, batch_size], 'activation', 'sigmoid') 
};
  • result:

Reference

  1. https://grzegorzgwardys.wordpress.com/2016/04/22/8/
  2. Dumoulin V, Visin F. A guide to convolution arithmetic for deep learning[J]. 2016.
  3. https://github.com/rasmusbergpalm/DeepLearnToolbox/tree/master/CNN
  4. http://neuralnetworksanddeeplearning.com/index.html



鲜花

握手

雷人

路过

鸡蛋
该文章已有0人参与评论

请发表评论

全部评论

专题导读
热门推荐
阅读排行榜

扫描微信二维码

查看手机版网站

随时了解更新最新资讯

139-2527-9053

在线客服(服务时间 9:00~18:00)

在线QQ客服
地址:深圳市南山区西丽大学城创智工业园
电邮:jeky_zhao#qq.com
移动电话:139-2527-9053

Powered by 互联科技 X3.4© 2001-2213 极客世界.|Sitemap