Data Mining Based on Compensation Fuzzy Neural Networks

The purpose: Solve the existing problems of fuzzy neural network are processing speed and low accuracy. Methods: This paper presents a data mining based on compensation fuzzy neural networks. Through research and analysis of the traditional compensation neural network, we are optimized it according to the characteristics of data mining. And optimize the calculation and training of the data mining. The result: In the first group can be seen that using this algorithm can be get a higher convergence. In the case of less data, the accuracy of the proposed algorithm is almost the same with literature algorithm. In the case of a large amount of data, the proposed method is better than other algorithms. conclusion: The experimental results basically consistent with the expected results. It can maintain convergence effect, while ensuring the accuracy of the network.