Novel feature selection method based on Stochastic Methods Coupled to Support Vector Machines using H- NMR data (data of olive and hazelnut oils)

• Gualdron, Oscar Eduardo ^[1] ; Isaza, Claudia ^[1] ; Duran, Cristhian Manuel
  1. [1] Universidad de Pamplona

     Universidad de Pamplona

     Colombia

     
• Localización: Revista ION: Investigación, Optimización y Nuevos Procesos en Ingeniería, ISSN-e 2145-8480, ISSN 0120-100X, Vol. 27, Nº. 2, 2014 (Ejemplar dedicado a: Revista ION), págs. 17-28
• Idioma: español
• Enlaces
  • Texto completo
• Resumen

  • Abstract One of the principal inconveniences that analysis and information processing presents is that of the representation of dataset. Normally, one encounters a high number of samples, each one with thousands of variables, and in many cases with irrelevant information and noise. Therefore, in order to represent findings in a clearer way, it is necessary to reduce the amount of variables. In this paper, a novel variable selection technique for multivariable data analysis, inspired on stochastic methods and designed to work with support vector machines (SVM), is described. The approach is demonstrated in a food application involving the detection of adulteration of olive oil (more expensive) with hazelnut oil (cheaper). Fingerprinting by H NMR spectroscopy was used to analyze the different samples. Results show that it is possible to reduce the number of variables without affecting classification results. Keywords: feature selection, H-NMR, simulated annealing, support vector machine, olive oil, hazelnut oil. Método de selección de variables basados en métodos estocásticos acoplados a Maquinas de soporte vectorial usando datos de H-NMR (data de aceite de oliva y avellana)