A new interval type-2 fuzzy reasoning method for classification systems based on normal forms of a possibility-based fuzzy measure

Abstract

To deal with two sources of uncertainty, cluster centers and reasoning aggregation operators, in the inference engine, this paper represents a new fuzzy reasoning method for an interval type-2 fuzzy classification system including cluster-based rules.In the reasoning, a new Possibility-based fuzzy measure is introduced to consider the uncertainty of cluster centers determined by the iterative clustering algorithms (because of the NP-Hardness of clustering models) in an interval type-2 fuzzy rule-based classification system (IT2 FRBCS). Then, the diversity of aggregation operators which are generally applied in reasoning process to combine the matching degrees of a test or unseen pattern with all clusters of an output class is regarded as the other source of reasoning uncertainty extending the crisp matching degrees into the intervals. The disjunctive and conjunctive normal forms of the OR operator for presented fuzzy measure are used as the lower and upper bounds of this interval, respectively. To evaluate the performance of the proposed system, a comparison against some other systems is conducted for 10 data sets. Further, it is applied to a practical problem of classification of generation units’ bidding behavior in the Iranian wholesale electricity market. Experimental results display the superior performance of the proposed system.