Data Stream Mining: an Evolutionary Approach
Cargando...
Autores
Tipo de contenido
Document language:
Español
Fecha
Título de la revista
ISSN de la revista
Título del volumen
Documentos PDF
Resumen
Este trabajo presenta un algoritmo para agrupar flujos de datos, llamado ESCALIER. Este algoritmo es una extensión del algoritmo de agrupamiento evolutivo ECSAGO Evolutionary Clustering with Self Adaptive Genetic Operators. ESCALIER toma el proceso evolutivo propuesto por ECSAGO para encontrar grupos en los flujos de datos, los cuales son definidos por la técnica Sliding Window. Para el mantenimiento y olvido de los grupos detectados a través de la evolución de los datos, ESCALIER incluye un mecanismo de memoria inspirado en la teoría de redes inmunológicas artificiales. Para probar la efectividad del algoritmo, se realizaron experimentos utilizando datos sintéticos simulando un ambiente de flujos de datos, y un conjunto de datos reales.
Abstract. This work presents a data stream clustering algorithm called ESCALIER. This algorithm is an extension of the evolutionary clustering ECSAGO - Evolutionary Clustering with Self Adaptive Genetic Operators. ESCALIER takes the advantage of the evolutionary process proposed by ECSAGO to find the clusters in the data streams. They are defined by sliding window technique. To maintain and forget clusters through the evolution of the data, ESCALIER includes a memory mechanism inspired by the artificial immune network theory. To test the performance of the algorithm, experiments using synthetic data, simulating the data stream environment, and a real dataset are carried out.
Abstract. This work presents a data stream clustering algorithm called ESCALIER. This algorithm is an extension of the evolutionary clustering ECSAGO - Evolutionary Clustering with Self Adaptive Genetic Operators. ESCALIER takes the advantage of the evolutionary process proposed by ECSAGO to find the clusters in the data streams. They are defined by sliding window technique. To maintain and forget clusters through the evolution of the data, ESCALIER includes a memory mechanism inspired by the artificial immune network theory. To test the performance of the algorithm, experiments using synthetic data, simulating the data stream environment, and a real dataset are carried out.