search

mingyuliutw / UNIT

Unsupervised Image-to-Image Translation
Other
1.98k stars 360 forks source link

Discriminator share #49

Closed taki0112 closed 6 years ago

taki0112 commented 6 years ago

Why did not the weight of discriminators be shared? Or maybe you tried, but the results were not good?

mingyuliutw commented 6 years ago

@taki0112 Thanks for your question. I found for some tasks, discriminator weight sharing is quite useful. For example, for the SVHN to MNIST domain adaptation, the two adversarial discriminators share weights for several layers. I also found that for the face image translation, discriminator weight-sharing is helpful too (The yaml file I released actually use this setting.). But when the domains are quite different and a patch-based discriminator is used, which often only have few layers. Discriminator weight sharing could hurt the performance. I am a deep believer of no free lunch theorem and am perfectly fine with using different models for different tasks. But some people prefer one model that rules all.