Resumen de Generalized metrics in classifiers

Carlos A. Hernández Nava, Pedro Lara Velázquez, Hérica Sánchez Larios, E. A. Rincón García, Miguel Angel Gutiérrez Andrade, Sergio G de los Cobos Silva, Román Mora Gutiérrez, Ivan O. Cruz Ruiz

• Euclidean distance and squared Euclidean distance are the most commonly used metrics in research using classifier systems, but it does not mean that they are the ones with which the best results are obtained. In this work, a set of well-known instances are classified using the higher-order Minkowski distance, first with integer values, then with values in hundredths around p = 1, and finally, experiments were conducted with a linear combination of metrics, that is, a hybrid, all of the above in order to observe the behavior of two different classifiers with various metrics and the accuracy they achieve for each database.