The National Institute of Standards and Technology (NIST) is seeking public comments on a draft white paper detailing the use of combination frequency differencing in artificial intelligence and machine learning applications.
The paper introduces a new method related to combinatorial testing and measurement that is particularly appropriate for AI and ML applications, the agency said Monday.
NIST said that more recently, methods applying coverage measures have been used in AI and ML applications to explain and analyze the aspects of transfer learning.
“These methods have been developed using measures that depend on the inclusion or absence of t-tuples of values in inputs and test cases,” NIST explained.
In the paper, NIST scientists Richard Kuhn, Raghu Kacker and M S Raunak extend the combinatorial coverage measures to include the frequency of occurrence of combinations.
“We illustrate the use of this method by applying it to analyzing physically unclonable functions (PUFs) for bit combinations that disproportionately influences PUF response values, and in turn provides an indication of the PUF potentially being more vulnerable to model-building attacks. Additionally, it is shown that combination frequency differences provide a simple but effective algorithm for classification problems,” the authors said.
Interested parties may submit feedback on the draft paper until Feb. 7, 2022.