CBP CBP Biometric Performance Evaluation and Testing 2 — Questions and Answers
Question 1: What is the primary purpose of a 'scenario evaluation' in biometric testing?
- To test biometric chips in e-passports under lab conditions
- To evaluate a complete biometric system in conditions that simulate a specific target application (Correct answer)
- To compare algorithms from different vendors using the same dataset
- To assess template aging effects over time
Correct answer: To evaluate a complete biometric system in conditions that simulate a specific target application
A scenario evaluation tests a complete biometric system under conditions that closely simulate the intended real-world deployment scenario, including environmental factors.
Question 2: Which statistical concept describes the degradation in biometric matching accuracy over time due to changes in a person's biometric trait?
- Template drift
- Template aging (Correct answer)
- Biometric decay
- Feature entropy reduction
Correct answer: Template aging
Template aging refers to the gradual degradation in biometric matching performance caused by changes in a person's biometric characteristics over time (e.g., aging face, scarred fingerprints).
Question 3: In a 1:N identification system, what performance metric replaces the simple FMR/FNMR used in 1:1 verification?
- FPIR and FNIR (Correct answer)
- Precision and recall
- Sensitivity and specificity
- AUC and EER
Correct answer: FPIR and FNIR
In 1:N identification, FPIR (False Positive Identification Rate) and FNIR (False Negative Identification Rate) replace FMR and FNMR to account for searching across a large enrolled database.
Question 4: What is 'rank-1 identification rate' in a biometric identification system?
- The probability that the top candidate returned by the system is the correct match (Correct answer)
- The accuracy of the best-performing algorithm in a vendor test
- The success rate of first-attempt enrollments
- The percentage of users assigned rank-1 priority access
Correct answer: The probability that the top candidate returned by the system is the correct match
Rank-1 identification rate is the probability that the correct identity is returned as the top (first) candidate in a 1:N biometric search.
Question 5: What does the term 'biometric menagerie' (wolves, lambs, goats, and sheep) describe?
- Animal biometrics used in wildlife tracking
- The variation in how different individuals' biometrics perform in a system (Correct answer)
- The four types of biometric spoofing attacks
- Categories of biometric system vendors
Correct answer: The variation in how different individuals' biometrics perform in a system
The biometric menagerie describes four user categories: sheep (easy to authenticate), goats (hard to authenticate/high FNMR), lambs (easily impostor-matched/high FMR), and wolves (skilled impostors).
Question 6: According to ISO/IEC 19795, what minimum size should a test database have to produce statistically reliable biometric performance estimates?
- At least 25 subjects
- At least 100 subjects and 1,000 genuine match attempts
- Sufficient size so that confidence intervals are acceptably narrow for the metric being reported (Correct answer)
- Exactly 500 subjects as mandated by the standard
Correct answer: Sufficient size so that confidence intervals are acceptably narrow for the metric being reported
ISO/IEC 19795 specifies that test databases should be large enough to produce confidence intervals narrow enough for the intended use, rather than mandating a fixed number.
CBP CBP Biometric Performance Evaluation and Testing 2
What is the primary purpose of a 'scenario evaluation' in biometric testing?