Reduction of the entries number of the training set for ANN through formal concept analysis and its application to solar energy systems

Abstract

The artificial intelligence has been developed in order to represent human knowledge in computers systems. It has two main fields: the symbolic field that works with symbolic data; and the connectionist field whose main example is artificial neural network and whose main characteristic is the capacity of learning by data samples. To obtain a high accuracy with generalization capacity net, the data set should cover all the problem possibilities. This situation can increase the time spent by the training process. Then, techniques for reducing the number of training sets preserving the representative characteristic are necessary. As formal concept analysis has been proposed as a powerful tool for data analysis, it has been used in this work as a way to reduce the training set elements number.