Confident_classifier icon indicating copy to clipboard operation
Confident_classifier copied to clipboard
verified publisher icon alinlab

Why the KL Divergence Loss needs to be multiplied by number of classes?

Open tim5go opened this issue 5 years ago • 0 comments

https://github.com/alinlab/Confident_classifier/blob/462db01967f8a96374f2ab6a534b7c81fd872d2f/src/run_joint_confidence.py#L156 I'm bit confused about why you needa multiply the KL divergence loss by the number of classes. I can't find it from the definition of your paper, could you briefly explain it?

tim5go avatar Aug 20 '20 08:08 tim5go