CPE407/ECG607: Biometrics
Spring 2008, 3 credits: Test 2
Dr. Pushkin Kachroo http://www.ece.vt.edu/pushkin Email: pushkin@unlv.edu
PROBLEM 1 : (25 points) Table 1 summarizes data obtained for non-matches and matches
with the corresponding scores.
Table 1: Score Distribution
Scores Non-Matches Matches
0-5 10 0
5-10 15 0
10-15 15 2
15-20 20 5
20-25 10 15
25-30 2 10
30-35 0 15
35-40 0 5
1. Plot the probability density functions for non-match scores pn(s) and match scores pm(s).
2. Write the formula for FMR and then compute it from the given data for threshold of 20.
3. Write the formula for FNMR and then compute it from the given data for threshold of 20.
4. Compute and draw the ROC curve from the data.
5. Consider another data set which is similar to the one from table 1 except the line with
scores from 15 −20, the values have been replaced by 10 and 10 for match and non-match
scores. The same is also done for the line with scores from 20 −25. Using the equal error
rate method, find out which set is better.
PROBLEM 2 : (20 points) Consider a database with mentries.
1. Find the probability of a false accept from this database.
2. Find the probability of correct reject.
3. Find the probability of an ambiguous imposter.
4. Find the probability of correct identification.
PROBLEM 3 : (5 points) Explain in a few sentences how rank order statistics are used to
compute the list length.
PROBLEM 4 : (5 points) Draw and label an example of a Zephyr chart and explain in one
sentence what it is used for.