search

tatsy / keras-generative

Deep generative networks, coded with Keras.
122 stars 28 forks source link

reconstruct exact input #13

Closed tsly123 closed 3 years ago

tsly123 commented 6 years ago

Hi, Thank you for your work. Currently, I am working with the CVAEGAN and reconstruct face with emotion, and I want to exactly reconstruct the input as normal autoencoder does and keep the discriminator as the same time. My thought is to change the discriminator binary_crossentropy (here) loss by mse? Is it right?

Thank you. tsly

tatsy commented 3 years ago

Sorry, but your question is not due to the issue of my code. Let me close this issue.