Use of GT labels to calculate classificaiton threshold at test time

ys-zong / MEDFAIR

[ICLR 2023 spotlight] MEDFAIR: Benchmarking Fairness for Medical Imaging

https://ys-zong.github.io/MEDFAIR/

60 stars 11 forks source link

Use of GT labels to calculate classificaiton threshold at test time #16

Open anissa218 opened 4 days ago

anissa218 commented 4 days ago

Hi, it seems like you use the ground truth test labels (tol_target) to find the optimal classification threshold, and subsequently determine accuracy, FNR, FPR etc.