Un método de aproximación de funciones basado en el enfoque de los prototipos más cercanos utilizando relaciones de similaridad
Fecha
2012-07-04
Autores
Bello García, Marilyn
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
En el trabajo se estudia el problema de la aproximación de funciones usando el paradigma de los prototipos más cercanos.
Se propone un método para construir prototipos usando relaciones de similaridad y un método de aproximación de funciones que utiliza del conjunto prototipos. Para cada clase de similaridad se construye un prototipo. Los resultados experimentales muestran que el método propuesto para construir prototipos logra una reducción significativa de la cantidad de instancias mientras la precisión se preserva.
El método aproximación de funciones que usa el conjunto prototipos y el constructor de prototipos se incorporaron a la plataforma WEKA. También se incorporó una adaptación del algoritmo de los k vecinos más cercanos para dar solución a problemas de aproximación de funciones usando relaciones de similaridad.
La incorporación de estos dos métodos de aproximación de funciones es de vital importancia para los investigadores del campo del Aprendizaje Automatizado al contar con nuevos métodos para dar solución a problemas de regresión.
El método de aproximación de funciones basado en prototipos más cercano que es propuesto en este trabajo se utilizó para resolver dos problemas reales de la Ingeniería Civil, específicamente en el pronóstico de la capacidad resistente en muros de mampostería y la capacidad resistente al deslizamiento longitudinal en las losas compuestas con lámina colaborante.
In this work the problem of the approximation of functions is studied using the paradigm of the nearest prototype. It is proposed a method for building prototypes using relations of similarity and a method of approximation of function that makes use of the set of prototypes. A prototype is built for each class of similarity. The experimental results show that with the proposed method a relevant reduction of the quantity de instances is obtained while the precision is preserved. The approximation method of function that makes use of the set of prototype and the prototype constructor are incorporated to the platform WEKA. Also an adaptation of the algorithm of the k nearest neighbors is incorporated for giving solution to the problem of approximation using relation of similarity. The incorporation of these two methods of approximation of function has a high importance for the investigators of the field of the machine learning because this provides new methods for giving solutions to regression problems. The method of approximation of function based in nearest prototypes proposed in this work was utilized to solve two real problems pertaining to the Civil Engineering, specifically in the prognostic of the resistant capacity in structural masonry and the resistant capacity to the glide longitudinal in the composite slabs with steel desk.
In this work the problem of the approximation of functions is studied using the paradigm of the nearest prototype. It is proposed a method for building prototypes using relations of similarity and a method of approximation of function that makes use of the set of prototypes. A prototype is built for each class of similarity. The experimental results show that with the proposed method a relevant reduction of the quantity de instances is obtained while the precision is preserved. The approximation method of function that makes use of the set of prototype and the prototype constructor are incorporated to the platform WEKA. Also an adaptation of the algorithm of the k nearest neighbors is incorporated for giving solution to the problem of approximation using relation of similarity. The incorporation of these two methods of approximation of function has a high importance for the investigators of the field of the machine learning because this provides new methods for giving solutions to regression problems. The method of approximation of function based in nearest prototypes proposed in this work was utilized to solve two real problems pertaining to the Civil Engineering, specifically in the prognostic of the resistant capacity in structural masonry and the resistant capacity to the glide longitudinal in the composite slabs with steel desk.
Descripción
Palabras clave
Aproximación de Funciones, Prototipos más Cercanos, Relaciones de Similaridad, Weka, Ingeniería Civil