Training of support vector machine with the use of multivariate normalization

摘要: Graphical abstractDisplay Omitted HighlightsWe analyze SVM (support vector machines) techniques.We propose a multivariable normalization of the inputs both during training and classification processes.The applied to real is equivalent use that uses Mahalanobis distance measure.The study confirms improvement achieved in processes. techniques have recently arrived complete wide range methods for complex systems. These systems offer similar performances other classifiers (such as neuronal networks or classic statistical classifiers) they are becoming valuable tool industry resolution problems. One fundamental elements this type classifier metric used determining between samples population be classified. Although Euclidean measure most natural solving problems, it presents certain disadvantages when trying develop can adapted characteristics sample space change. Our proposes means avoiding problem using multivariate (both processes). Using experimental results produced from significant number populations, Lastly, demonstrates measure, non-normalized data.