Nguyen, Cong Long. Hybrid interval type-2 fuzzy logic system optimization. Doctoral Degree(Information Technology). King Mongkut's University of Technology North Bangkok. Central Library. : King Mongkut's University of Technology North Bangkok, 2014.
Hybrid interval type-2 fuzzy logic system optimization
Abstract:
Design type-2 fuzzy logic systems is very complex tasks, which needs to determine the best number of rules, the number of membership functions, suitable parameters, and structure of the fuzzy system. This study proposes an optimal a design of interval type 2 fuzzy logic system. First of all, rough sets based attribute reduction using chaos firefly algorithmmisproposedto find optimal subsets of input variables. Consequently, the fuzzy c-mean clustering algorithm is applied to build the structure of fuzzy rule base and identify the number of fuzzy rules. Finally, a chaos firefly and genetic hybrid algorithm is proposed and then applied to optimize parameters of the fuzzy logic system.The proposed fuzzy system was evaluated in three applications: sea water level prediction, stock price prediction, and heart disease diagnosis. For sea level prediction and stock market prices prediction, comparisons of the proposed fuzzy system to genetic algorithms based and chaos firefly algorithm based interval type-2 fuzzy system as well as ANFIS were investigated. A lower prediction error measured by using root mean square error and mean absolute percentage error from the proposed method was proven. In addition, for the heart disease diagnosis, comparisons of the proposed fuzzy system to NaiveBayers, support vector machines, and artificial neural network were also presented with the result of classification 7.5% increase. Furthermore, convergence speed of the proposed attribute reduction is approximately 92% faster than rough sets based attribute reduction using particle swarm optimization.