NLNL-Negative-Learning-for-Noisy-Labels icon indicating copy to clipboard operation
NLNL-Negative-Learning-for-Noisy-Labels copied to clipboard

Published 20 hours ago • verified publisher icon ydkim1293

How to select the samples after NL?

Open monsterlyg opened this issue 3 years ago • 1 comments

After NL, the next step is SelNL, but i have problems with "py > 1/c". According to my understanding, the ideal condition is that the network will output a low probability corresponding to a complementary label after NL. If we select the samples with output probability over 1/c, did we select the data that can not be effectively splitted after NL? So what exactly the 'py' means? Or what the confidence 'py' represents?

monsterlyg avatar Aug 11 '20 02:08 monsterlyg

After NL, the next step is SelNL, but i have problems with "py > 1/c". According to my understanding, the ideal condition is that the network will output a low probability corresponding to a complementary label after NL. If we select the samples with output probability over 1/c, did we select the data that can not be effectively splitted after NL? So what exactly the 'py' means? Or what the confidence 'py' represents?

I guess the y is the noisy label in trainset， not the complementary label.

Codeczh avatar Apr 29 '21 08:04 Codeczh