Abstract:
There are different computer models to classify groups of data, among which are neural networks, vector support machines, numerical methods, among others. However, in some cases these strategies consume a large amount of computer resources, reducing the speed of operation during their execution in the various electronic development devices. In this work, a partial solution to this limitation is proposed. The algorithm developed is a classifier that incorporates a backward-propagation neural network, which is trained by means of a modified genetic algorithm that is in charge of finding the appropriate set of weights for the neural network to classify a given group of random numbers. The proposed optimization will allow the use of this algorithm in various classification problems, not only in conventional computing units, but also on various technological platforms with reduced properties (embedded systems), maintaining an optimal balance between the use of resources and the speed of response of the device used.
Page(s):
2606-2615
DOI:
DOI not available
Published:
Journal: Journal of Theoretical and Applied Information Technology, Volume: 98, Issue: 13, Year: 2020