Modelos de máxima entropía y su resolución numérica mediante métodos de segundo orden con aplicación en la predicción de presencia de especies en áreas naturales

Abstract

In this project we calculate accurate simulations of the maximum entropy model, using second order algorithms applied in the prediction of the distribution of species in natural areas. In particular, we use the OESOM method, which uses orthant directions and second-order information of the l1 regularization. It is considered a database collected for the species Bradypus variegatus with which comparisons are made for the method used by the MaxEnt software and the OESOM method. This second order method can make more accurate predictions. In addition, it efficiently recognizes the set of null variables. A deeper study of the benefits of second-order algorithms applied to these types of problems is proposed.