Contrastive-Adaptation-Network-for-Unsupervised-Domain-Adaptation icon indicating copy to clipboard operation
Contrastive-Adaptation-Network-for-Unsupervised-Domain-Adaptation copied to clipboard

Published 20 hours ago • verified publisher icon kgl-prml

Reproduce results on Office-31

Open kuztown opened this issue 4 years ago • 2 comments

Hi! I'm trying to reproduce results on Office-31. I've trained each experiment 10 times. I've selected the best results in each train(not the last one). I've calculated average of 10 samples and it is worse in D->A: in paper result is 78, my result is 77, in D->W: in paper result is 99.1, my is 98.3 and W->A: in paper result is 77, my result is 75.6. Can you help me with this issue?

kuztown avatar May 20 '20 09:05 kuztown

Same problem for me. Actually i found the parameter "FILTER_THRESHOLD" in the config file makes big difference. According to what i found, set it to 0.05 would increase about 2% accuracy on a->d, compared with 1, i.e. no filtering. You may try different values for the parameter to see if it works.

Ledzy avatar May 26 '20 17:05 Ledzy

Thank you for your reply I've trained with different FILTER_THRESHOLD(0.05, 0.5, 0.75) on d->a. Unfortunately, accuracy wasn't increased

kuztown avatar Jun 04 '20 09:06 kuztown