FORMAL ASPECTS OF A MULTIPLE-RULE CLASSIFIER
Josef Tkadlec, Ivan Brůha · International Journal of Pattern Recognition and Artificial Intelligence · 2003
This paper deals with the multiple-rule problem which arises when several decision rules (of different classes) match ("fire" for) an input to-be-classified (unseen) object. The paper focuses on formal aspects and theoretical methodology for the above problem. The general definitions of the notions of a Designer, Learner and Classifier are presented in a formal matter, including parameters that are usually attached to the above concepts such as rule consistency, completeness, quality, matching rate, etc. We thus provide the minimum-requirement definitions as necessary conditions for these concepts. Any designer (decision-system builder) of a new multiple-rule system may start with these minimum requirements. We only expect that the Classifier makes its decisions according to its decision scheme induced as a knowledge base (theory, model, concept description). Also, two case studies are discussed. We conclude with a general flow chart for a decision-system builder. He/she can just pursue it and select parameters of a Learner and Classifier, following the minimum requirements provided.