An interval type-2 fuzzy system with hybrid intelligent learning

In this paper, an alternative approach for automatically generation of interval type-2 fuzzy inference systems is proposed. The proposed method comprises of two phases: 1) Structure initialization and parameters fine tuning. In the first phase, a one-pass clustering method is carried out to find both a suitable number of rules and a suitable number of fuzzy sets of each variable in which inputs and targets are used as training data. In the second phase, the genetic algorithm is then employed to fine tune the membership function parameters to increase the performance of the system. The evaluation of the proposed method is then conducted for pattern classification. The results show satisfactory achievement in pattern classification applications and comparable to existing techniques.