Attribute Reduction of Fuzzy Data Using a Pseudometric Based Resemblance Relation

The problem of attribute reduction has been proved to be NP-hard problem. We present an extension of attribute reduction algorithm to a general type of fuzzy data, in which a pseudometric based resemblance relation is introduced to model approximate equality between objects in the universe instead of fuzzy equivalence relation. We also give an extended quality of classification, which is used as a denominator in the reduction process. An example is given to illustrate the presented algorithm.