info presentación miembros líneas investigación publicaciones investigación tesis docencia

[Articulos Revista] [Articulos Congreso] [Reports] [Libros] [Capitulos Libro] [Todo] [Resumen]

Categoria WoS
Titulo: Texture Classification using Kernel-Based Techniques
Tipo: congreso internacional
Congreso: International Work Conference on Artificial Neural Network (IWANN)
Fecha: 12-14/6/2013
Lugar celebracion: Puerto de la Cruz
Volumen: LNCS 7902
Paginas: 427-434
ISSN: 0302-9743
ISBN: 978-3-642-38678-7
Libro: IWANN 2013, Part I
Editorial: Springer Berlin Heidelberg
doi: 10.1007/978-3-642-38679-4_42


In this paper, a high-dimensional textural heterogenous dataset is evaluated. This problem should be studied with specific techniques or a solution for decreasing dimensionality should be applied in order to improve the classi- fication results. Thus, this problem is tackled by means of three differente techniques: an specific technique such as Multiple Kernel Learning, and two different feature selection techniques such as Support Vector Machines- Recursive Feature Elimination and a Genetic Algorithm-based approaches. We found that the best technique is Support Vector Machines-Recursive Feature Elimination, with a AUROC score of 92,45%

    .: SABIA :.  Sistemas Adaptativos y Bioinspirados en Inteligencia Artificial