In this paper, we proposed Type-1/Type-2 polynomial Radial Basis Function Neural Networks. RBF neural networks based on Interval Type-2 fuzzy set consist of four layers such as Input layer, hidden layer, Karnik and Mendel (KM) layer, and output layer....
In this paper, we proposed Type-1/Type-2 polynomial Radial Basis Function Neural Networks. RBF neural networks based on Interval Type-2 fuzzy set consist of four layers such as Input layer, hidden layer, Karnik and Mendel (KM) layer, and output layer. In the receptive field of hidden layer, Interval Type-2 fuzzy set is used. The characteristic of Interval Type-2 fuzzy has Footprint of Uncertainly (FOU) which denotes a certain level of robustness in the presence of unknown information when compared with the Type-1 fuzzy set.
The prototype of the receptive field is obtained by three kinds of methods such as Min-Max, K-means, FCM algorithm. Also, the values of the standard deviation of input variables are used as distribution of coefficients receptive field. We use the linear polynomial function as connection weight of network. Type-2 fuzzy set needs type-reduction process in order to estimate model output. In KM layer, Karnik and Mendel (KM) algorithm is used for type-reduction. The parameters such as center values of receptive field, constant deviation, and connection weight between hidden layer and output layer are optimized by Conjugate Gradient Method (CGM) and Space Search Evolutionary Algorithm (SSEA). To handle proposed modeling as well as The pattern recognition problems, proposed modeling and pattern recognition methods are applied to modeling dataset (Gas furnace, Automobile Miles per Gallon, and Boston Housing) and pattern classification dataset (Synthetic, Iris, and Pima) and their results are compared with those reported in the previous studies.