search

RockySJ / WDGRL

124 stars 35 forks source link

mnist code #4

Closed ltnn closed 6 years ago

ltnn commented 6 years ago

Hi, I think you made a mistake in in mnist code. You put both source and target to train the classifier (use y_ instead of ys_true) so the result is 99.1%, which is too good to be true.

RockySJ commented 6 years ago

Thank you for your reminding. I have made a mistake here. I will refine this part of code and rerun the experiments.

ltnn commented 6 years ago

Thanks, btw, there is another interesting dataset: imageclef domain adaptation. You may like to test it, too. http://www.imageclef.org/2014/adaptation#test