Z-Score based Fuzzification process for pattern Classification
Abstract— Fuzzification is the process of converting the crisp value into fuzzy value by using membership function. In the proposed classification model, Fuzzification is done by using Z-score method. Membership matrix is formed as the result of Fuzzification process and it contains the membership of each feature value to the given classes. This membership matrix is given as an input to the Multilayer Feedforward Neural Network (MFNN) uses a Backpropagation learning algorithm for training. The MFNN classifies the pattern based on the fuzzy value computed by Z-score method and it assigns the degree of membership for each pattern to the given class. Then, the target class for each pattern is predicted using the MAX defuzzification method. The proposed classification model is applied to the Statlog Heart dataset. It was obtained from the University of California at Irvine (UCI) machine learning repository. The performance measures used by the Neuro fuzzy classifier are Accuracy and Training Speed. The experimental results show that the neuro fuzzy approach consistently outperforms and gives higher accuracy than the artificial neural network
Index Terms— Artificial Neural Network, Accuracy, Back propagation, Classification, Fuzzification, Membership Function, Multilayer Feedforward Neural Network, Defuzzification, Training Speed, Z-Score.
Click Here
International Journal for Trends in Technology & Engineering © 2015 IJTET JOURNAL