New approaches to incremental learning good classification tests

The paper is devoted to incremental inferring of a special kind of logical classification rules called good tests. They are "good" because they cover the largest possible number of objects w.r.t. inclusion relation on the set of all subsets of objects. Moreover we are interested in such good tests which are maximally redundant (GMRTs), i.e. their subsets of attributes are closed. Incremental learning allows to have more flexible control of GMRTs inferring than a usual (batch) case of learning. We develop two new generic approaches to infer GMRTs. First approach provides learning with a use of pattern recognition-like processes. Second approach implements an object taxonomic organisation. All considerations are supplied with running examples.