0
$\begingroup$

I have a system that compares two items and produces a match score. Scores below a threshold are manually inspected to determine if they match or don't(imposter). Scores above the threshold are assumed to match with no further testing. Can I use the distribution of imposter scores below threshold to estimate the imposters above threshold that were assumed to match (false accept rate)? If no, how can I estimate the false accept rate without testing those above threshold? During system test where all items were verified the distribution of imposter scores was lognormal.

$\endgroup$
3
  • $\begingroup$ So for all the scores that were below a threshold you manually inspect them, save all these results, get a distribution, and now you want to estimate the imposters above threshold that were assumed to match (false accept rate); can you explain better what you mean by this? $\endgroup$ Jul 17 '19 at 7:41
  • $\begingroup$ That is correct. I know the distribution of scores for the imposters that are below the threshold that the system identifies and are prevent from going forward. I want to estimate how many are above threshold. Basically, I want to know what gets past the system that should not. $\endgroup$
    – Cavalent
    Jul 17 '19 at 15:35
  • $\begingroup$ I don't understand. You say that the imposters are defined as being below the threshold, but then you say you want to estimate how many are above threshold? None of them? Since they are all below the threshold? $\endgroup$ Jul 19 '19 at 12:00

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.