Institute of Information Theories and Applications FOI ITHEA
Abstract:
Logic based Pattern Recognition extends the well known similarity models, where the distance
measure is the base instrument for recognition. Initial part (1) of current publication in iTECH-06 reduces the logic
based recognition models to the reduced disjunctive normal forms of partially defined Boolean functions. This
step appears as a way to alternative pattern recognition instruments through combining metric and logic
hypotheses and features, leading to studies of logic forms, hypotheses, hierarchies of hypotheses and effective
algorithmic solutions. Current part (2) provides probabilistic conclusions on effective recognition by logic means in
a model environment of binary attributes.