Non-standard biometric tech or project? iBeta suggests ad hoc testing

While independent testing is important for biometrics security, iBeta notes that the standards underpinning the most common tests do not apply to all technologies or projects. That does not mean that independent testing is impossible for those biometric systems; instead, the testing lab proposes, they should consider Ad Hoc testing.
The position is explained a new white paper: “Biometric Ad Hoc Testing: When the Standards Don’t Meet Your Goals.” The four-page white paper begins with a review of popular biometrics standards and the situations in which they do not apply. Standardized test protocols often do not apply to experimental technologies and specialized projects, iBeta says.
The limitations of standards-based testing are explained with reference to the presentation attack detection testing the lab is best known for and performance testing. The ISO/IEC standard for PAD testing concludes when the pass or fail error rate is achieved. Ad hoc testing can provide more comprehensive understanding of how effective the system is by continuing until the available samples have all been tested. Additionally, ad hoc PAD testing can evaluate effectiveness with a mix of the attack methods included in Level 1 and Level 2 tests. PAD and performance testing can also be combined, yielding valuable practical insight into the effectiveness of the software.
Independent laboratories can help businesses design the kind of ad hoc testing that yields more real-world relevance, flexibility, timeliness and customization, according to iBeta. The white paper delves into each potential benefit.
The paper also suggests possible drawbacks to ad hoc biometrics testing, and provides an example of how the tests can be set up and carried out.
iBeta has referred in the past to some of its ad hoc tests as “Bake-Offs” and “Face-Offs.”
Biometric bake-offs: Identify solutions through fair competition
Article Topics
biometric testing | biometrics | iBeta | research and development | standards
Comments