Question about experiment setting in the paper

AntreasAntoniou / DAGAN

DAGAN: Data Augmentation Generative Adversarial Networks

https://arxiv.org/abs/1711.04340

MIT License

415 stars 103 forks source link

Question about experiment setting in the paper #18

Closed YusukeO closed 5 years ago

YusukeO commented 5 years ago

Hi Antoniou !

I cannot understand the following statement in your paper.

The real or fake label was also passed to the network

It means that adding one extra class "fake" to final layer of CNN?

Thank you!

AntreasAntoniou commented 5 years ago

Nope. It means that we condition our classifier on both a data-point X, and a one-hot vector that indicates whether X came from the real or fake distribution. This way the classifier can create separate boundaries for each type of data-point.

YusukeO commented 5 years ago

Thank you for quick reply! Now everything is clear!