Data Mining based on CMAC Neural Networks
Francisco Palácios, Xiaoou Li, Luís Freitas Rocha · 2006
Neural networks are a widely used data mining technique, but for high-dimensional datasets, the training process of the normal neural networks, such as multilayer perceptron (MLP), is very slow. It is an important drawback for using them in real-time data mining applications where the main requirement is to have an answer within a short time. In this research work we propose a CMAC neural network adaptation for data mining, which most provides fast training time and guaranteed convergence. This paper describes how we built a CMAC adaptation for data mining, obtaining a classification model that can be applied to real-life datasets. Experimental results show that CMAC may be an alternative model for high-dimensional data classification in data mining