Generalized non-reducible descriptors
Ventzeslav Valev, Bülent Sankur, Petia I. Radeva · 2002
This paper provides a generalization of non-reducible descriptors. Non-reducible descriptors are used in supervised pattern recognition problems when the patterns descriptions consist of Boolean variables. This generalization extends the concept of distance between patterns of different classes. A mathematical model to construct generalized non-reducible descriptors, a computational procedure, and numerical examples are discussed.