Examining fault-tolerance using unlikely inputs: turning the test distribution up-side down
Jeffrey Voas, Keith Miller · 2002
This paper turns the concept of input distributions on its head to exploit "inverse input distributions". Although such distributions are not true mathematical inverse functions, they capture the intuitive property of members that have high frequencies in the original distribution have low frequencies in the "inverse" distribution, and vice versa. These new distributions have uses in testing for reliability estimates, and more importantly, fault-tolerance analysis.