search

mp2893 / medgan

Generative adversarial network for generating electronic health records.
BSD 3-Clause "New" or "Revised" License
270 stars 89 forks source link

keep_prob always 1.0? #7

Open rcamino opened 6 years ago

rcamino commented 6 years ago

Hello,

Why is keep_prob always 1.0? Does that mean that the dropout in the discriminator is always disabled?

Thank you.

mp2893 commented 6 years ago

Hi Ramiro,

Yes, I initially played around with the dropout value a bit. But eventually I'd decided not to use dropout at all since the model worked fine without it. (I should have deleted the parameter "keep_prob:1.0", but I thought I might want to use dropout again in the future) But you are welcome to try tuning the dropout value for extra performance boost.

Thanks, Ed