Published: April 4, 2022
Author(s)
Richard Kuhn (NIST), M S Raunak (NIST), Charles Prado (Inmetro), Vinay Patil (University of Massachusetts), Raghu Kacker (NIST)
Conference
Name: IEEE International Conference on Software Testing Verification and Validation Workshop (ICSTW 2022)
Dates: 04/04/2022 - 04/13/2022
Location: [Virtual]
Citation: 2022 IEEE International Conference on Software Testing, Verification and Validation Workshop (ICSTW), pp. 110-117
Combinatorial coverage measures have been defined and applied to a wide range of problems. These methods have been developed using measures that depend on the inclusion or absence of t-tuples of values in inputs and test cases. We extend these coverage measures to include the frequency of occurrence of combinations, in an approach that we refer to as combination frequency differencing (CFD). This method is particularly suited to artificial intelligence and machine learning (AI/ML) applications, where training data sets used in learning systems are dependent on the prevalence of various attributes of elements of class and non-class sets. We illustrate the use of this method by applying it to analyzing the susceptibility of physical unclonable functions (PUFs) to machine learning attacks. Preliminary results suggest that the method may be useful for identifying bit combinations that have a disproportionately strong influence on PUF response bit values.
Combinatorial coverage measures have been defined and applied to a wide range of problems. These methods have been developed using measures that depend on the inclusion or absence of t-tuples of values in inputs and test cases. We extend these coverage measures to include the frequency of occurrence...
See full abstract
Combinatorial coverage measures have been defined and applied to a wide range of problems. These methods have been developed using measures that depend on the inclusion or absence of t-tuples of values in inputs and test cases. We extend these coverage measures to include the frequency of occurrence of combinations, in an approach that we refer to as combination frequency differencing (CFD). This method is particularly suited to artificial intelligence and machine learning (AI/ML) applications, where training data sets used in learning systems are dependent on the prevalence of various attributes of elements of class and non-class sets. We illustrate the use of this method by applying it to analyzing the susceptibility of physical unclonable functions (PUFs) to machine learning attacks. Preliminary results suggest that the method may be useful for identifying bit combinations that have a disproportionately strong influence on PUF response bit values.
Hide full abstract
Keywords
combinatorial frequency; combinatorial testing; Physical(ly) Unclonable Functions; PUF
Control Families
None selected
