Analysis and Application of Data Mining Based on Clustering Algorithm

摘要

As the ART2 neural network clustering occurs normalization in the data inputting mode by vector and nonlinear transformation pretreatment process is easy to be filtered as a substrate for an important, but a minor component of the noise, while there are still phenomenon of the drift mode in the learning process due to the correction of the value of weight, this paper proposes an improved method of ART2 neural network. On the one hand, as the inputting mode enters into the network to start learning, in the learning process of the input mode to enter the network, it saves their amplitude information, relaxes negative real nonlinear conversion, and considers the shortest distance of the input pattern to each cluster center.; on the other hand, there is also a need to make the corresponding treatment on the non-linear transfer function, so that it can properly handle the input of negative and retains its negative form after the stable F1 layer, not causing loss of information in the inputting mode; in another aspect, in order to eliminate outliers' influence on clustering results, this paper also carried out on the input mode to determine outliers.