Extensiones al ambiente de aprendizaje automatizado Weka
Cargando...
Fecha
Autores
Matías González, Héctor
Araujo Pérez, Liana Isabel
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas. Facultad de Matemática, Física y Computación. Departamento de Ciencias de la Computación
Resumen
Este trabajo presenta extensiones realizadas a la herramienta de Aprendizaje AutomatizadoWeka. Se incorporan los algoritmos que utiliza un nuevo modelo híbrido de Razonamiento Basado en Casos. Específicamente se añade un nuevo tipo de dato para manejar un atributo numérico como variable lingüística y un nuevo filtro con esta finalidad, y se implementan dos algoritmos de clasificación. El diseño de Weka hace que la incorporación de nuevos
modelos no sea una tarea tan compleja, sin embargo en el trabajo queda definida una metodología para adicionar nuevos filtros y clasificadores que facilita aún más esta tarea.
Finalmente, se validan los algoritmos implementados utilizando conjuntos de datos internacionales, mostrando la factibilidad de utilizar esta herramienta y los nuevos algoritmos en el campo de la Inteligencia Artificial.
This work presents the extensions made to the “Weka” Machine Learning tool in order to incorporate the algorithms used in a new Case-Based Reasoning hybrid model. Specifically, a new data type is added to handle a numerical attribute as a linguisticvariable. A new filter and two classification algorithms are purposefully implemented by using Weka. Although its environment was designed with the goal of reducing the complexity when adding new models, additionally a methodology for incorporating new filters and classification algorithms is outlined. Finally, the implemented algorithms are validated by means of well-known international datasets, therefore concluding the feasibility of using this tool and the new algorithms in the Artificial intelligence field.
This work presents the extensions made to the “Weka” Machine Learning tool in order to incorporate the algorithms used in a new Case-Based Reasoning hybrid model. Specifically, a new data type is added to handle a numerical attribute as a linguisticvariable. A new filter and two classification algorithms are purposefully implemented by using Weka. Although its environment was designed with the goal of reducing the complexity when adding new models, additionally a methodology for incorporating new filters and classification algorithms is outlined. Finally, the implemented algorithms are validated by means of well-known international datasets, therefore concluding the feasibility of using this tool and the new algorithms in the Artificial intelligence field.