@JzCh 谢谢
简单GAN网络 matlab实现说明 此代码在matlab上搭建了简单的生成对抗性网络，用来生成手写数字图像。网络中生成器和鉴别器的隐藏层均为2层，且都是全连接层，是一个比较简单的网络结构。主要用来说明怎么...
发简信
IP属地：河北
-
-
Each time you go to one of those wonderful places ,exotic or otherwise,remember to simplify, remember to take care of yourself,remember to exercise,remember to assert yourself and say no when appropriate,remember to cultivate your relationships,remember what is really really important for you in your life.
-
@JzCh 请问楼主如何通过g_loss和d_loss的变化情况来大致推测GAN是否收敛？或者说g_loss和d_loss是变化越来越小比较好，还是越接近0.5或1比较好。
这是变化的时候。。
c_loss:"0.703654",d_loss:"0.693202"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\832_t_20.png
c_loss:"0.705559",d_loss:"0.693223"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\833_t_20.png
c_loss:"0.704232",d_loss:"0.693208"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\834_t_20.png
c_loss:"0.702664",d_loss:"0.693192"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\835_t_20.png
c_loss:"0.701254",d_loss:"0.693180"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\836_t_20.png
后边就不变了。。
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\891_t_20.png
c_loss:"0.693147",d_loss:"0.693147"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\892_t_20.png
c_loss:"0.693147",d_loss:"0.693147"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\893_t_20.png
c_loss:"0.693147",d_loss:"0.693147"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\894_t_20.png
c_loss:"0.693147",d_loss:"0.693147"
save_sample:C:\Users\TD314\Documents\MATLAB\GAN\program\fake\895_t_20.png
这样合理嘛简单GAN网络 matlab实现
说明 此代码在matlab上搭建了简单的生成对抗性网络，用来生成手写数字图像。网络中生成器和鉴别器的隐藏层均为2层，且都是全连接层，是一个比较简单的网络结构。主要用来说明怎么...
-
-
-
