Design of Convolutional Fuzzy Neural Network Classifiers

In this paper, we propose a convolutional fuzzy neural network classification to alleviate the problems of processing high-dimensional data and low computational efficiency in traditional convolutional neural networks. The model proposes a convolution fuzzy C-means algorithm, in the meanwhile uses the L2-norm regularization method to estimate parameters, so that it has better generalization ability. The experimental results indicate that the proposed CFNNCs have excellent performance in classification accuracy than classic classification models such as SVM, RVM, KNN, and the experimental accuracy can be maintained above 90%.