Weird Validation behavior

samet-akcay / skip-ganomaly

Source code for Skip-GANomaly paper

MIT License

192 stars 64 forks source link

Weird Validation behavior #11

Open Shumpei-Kikuta opened 4 years ago

Shumpei-Kikuta commented 4 years ago

Thank you for your great work!

I made an experiment with shuffle True when validation to make sure it never affects performance. But the result I got was significantly lower than shuffle False. I don't think that AUC is never affected by the order in the validation datasets.

Would you have any ideas?

24hours commented 4 years ago

this issue is related to https://github.com/samet-akcay/ganomaly/issues/47