Suppose I have a neural network $N$ that produces the output probabilities $[0.3, 0.8]$. Normally, I would specify a threshold of 0.5 for the argmax of the prediction, let's say, second arg > 0.5 means that the image is attractive, and if both probabilities are lesser than 0.5 we don't have that good of a prediction.

My question is, can we plot this threshold on a ROC curve so we can figure out the best value?

  • $\begingroup$ When you say "can we plot this threshold" do you mean "can we plot different thresholds"? $\endgroup$ – nbro Jul 8 '20 at 14:10
  • $\begingroup$ Well the ROC as far as I understand it is plotting points based on the specificity and sensitivity that’s coming at a specific threshold. So, yes, i’m asking if we can see what threshold is the best $\endgroup$ – iustin Jul 8 '20 at 16:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.