A Comparative And Analysis Research Of Neural Network In The Classification Algorithm Of Data Mining

With the development of information technology, people’s ability of collecting and producing data is increasing, but the ability of knowledge acquirement and data analysis remains comparatively slow. From data collection, databases creation, data management, data analysis, data mining technology consequently producing and developing.Data Mining (DM) is an interdisciplinary subject, which is involving a number of fields, including databases, machine learning and artificial intelligence. Data mining is also known as knowledge discovery in databases, which is to obtain valid data from a large, novel, useful, non-trivial process understandable patterns,and also is to extract knowledge from large amounts of data. Classification is a very important research topic in data mining. The use of classification techniques can extract the same model of describing the data class or function from the data set, and can classify the unknown category of each object in the dataset to the right category. Currently, the classification algorithm is mainly used, such as statistical classification method, decision trees, artificial neural network methods. Different algorithms will produce different classifiers, the classifier is good or bad will directly affect the accuracy and efficiency of data mining. Therefore, when large-scale massive data classification, select the appropriate classification algorithm is very important.Artificial Neural Network (ANN) is one of the important methods of data mining, by simulating the human brain biological neural networks, nodes with certain neuronal processing functions according to certain network structure connected, it has to deal with imprecise data, fuzzy data, or the ability to map complex nonlinear problems. Artificial neural network model is able to identify the connection weights of the network topology and neuron threshold decision. By optimizing the weights of the network topology and network, the threshold can be optimized artificial neural network model.In view of the classification problems in practical application, this paper introduces three kinds of network structure and algorithm of artificial neural network algorithm, and the advantages and disadvantages of three kinds of algorithms, expounds the theory basis of extreme learning machine. Limit of machine learning algorithm was applied to six real data set, and implement classification application test, and support vector machine (SVM) and the results of the experiment and comparing the experimental results in BP algorithm. Through the experiment results show that extreme learning machine in classification accuracy and time etc. Opposite, have obvious advantages.