ABSTRACT
The combination of Artificial Neural Networks (ANN) and Fuzzy Inference Systems (FIS) has drawn the attention of several researchers in various scientific and engineering fields due to the rising needs of intelligent systems to solve the real world complex problems. ANN learns the presented inputs from the base by updating the interconnections between layers. FIS is a most common computation model based on the concept of fuzzy set theory, fuzzy if-then rules, and fuzzy reasoning. There are several advantages in the fusion of ANN and FIS. ANN and FIS can be integrated in several methods depending upon the application. The integration of ANN and FIS can be classified into three categories namely concurrent model, cooperative model, and fully fused model. This chapter begins with a discussion of the features of each model
and generalizes the advantages and deficiencies of each model. The chapter further focuses on the different types of fused neuro-fuzzy systems such as such as FALCON, ANFIS, GARIC, NEFCON, FINEST, FUN, EFuNN, and SONFIN and citing the advantages and disadvantages of each model. A detailed description of ANFIS including its architecture and learning algorithm are discussed. The implementation detail of hybrid neuro-fuzzy model is also delineated. An explanation on Classification and Regression Trees with its computational issues, computational details, computational formulas, advantages, and examples is given in this chapter. The data clustering algorithms such as hard c-means, Fuzzy c-means, and subtractive clustering are also described. The combination of Neuro and Fuzzy for computing applications is a
popular model for solving complex problems. Whenever there is a knowledge expressed in linguistic rules, an FIS can be modelled, and if information is available, or if the parameters can be learned from a simulation (training) then ANNs can be used. While building a FIS, the fuzzy sets,
implement an ANN for a specific application the architecture and learning algorithm are required. The drawbacks in these approaches appear complementary and consequently it is natural to consider implementing an integrated system combining the neuro-fuzzy concepts.
