Updated and corrected: Academics say too much uncertainty in biometrics to be dependable
24 May, 2011
category: Biometrics
The Internet can be a funny thing and sometimes mistakes are made. This is one of those times.
This Web post, “Academics say too much uncertainty in biometrics to be dependable,” is based on a news article that misconstrued the original research. The research states that lab tests don’t always translate to the real world but in no way says that country’s are dropping biometric programs.
The full NIST report can be downloaded here.
ThirdFactor apologizes for the error.
James L. Wayman, Antonio Possolo and Anthony J. Mansfield, co-authors of a study from San Jose State University, the statistical engineering division of the U.S. National Institute of Standards and Technology and the UK National Physics Laboratory respectively, have released a paper arguing that biometric test results cannot be translated to the real world.
The trio argued in a paper titled “Fundamental issues in biometric performance testing: A modern statistical and philosophical framework for uncertainty assessment” that the level of uncertainty in biometric technology is high enough that lab tests to predict real world performance are not accurate, according to a Money Life article.
The findings in the paper were summed up by its authors as follows: test repeatability and reproducibility from lab-based tests cannot be applied to the real world scenario testing due to a loss of statistical control and, subsequently, scenario data cannot be applied to mathematical models that require high levels of certainty.
Read the full story here.