Using threshold and bias at the same time in NN

Question

I'm using NN with sigmoid binary activation. And for threshold I using 0,5. So if output < 0,5, it classified as 0. And if output >= 0,5 it classified as 1. But I'm using bias too at the same time. Is it ok? Cause from what I know bias change threshold into 0.

Answer 1

The bias is added before the activation, i.e. $y=\sigma(wx+b)$, so that's not directly associated with what you choose as threshold.

In addition, instead of thresholding your output, you might also consider probabilistic scoring rules like ROC AUC. A starting point might be this thread.