Clasificación Automática de los Trazos del Test de Bender basada en modelos de aprendizaje profundo.

Ruíz Vázquez, Dionisio (2018) Clasificación Automática de los Trazos del Test de Bender basada en modelos de aprendizaje profundo. Maestría thesis, Universidad Autónoma de Chihuahua.

[img]
Vista Previa
Text
Final Tesis Dionisio.pdf

Download (10Mb) | Vista Previa

Resumen

In this work is presented a methodology to automatically classify the nine strokes on the Bender Gestalt Test using computational models. The aim of the Test is to measure the visomotor skill by analyzing the nine strokes. In total nine different computational models based on artificial neural networks, specifically convolutional neural networks were tested. In order to expand the database and have data with different problematics, such as rotation, translation, scale and elastic distortion were implemented as a data augmentation strategy. In the same way, transfer learning strategies were implemented using different databases with the purpose to boost performance. The results achieved overcome 90% accuracy in all folds, demonstrating that the use of computational models based con convolutional neural networks are able to automatically classify the nine Bender strokes.

Tipo de Documento: Tesis (Maestría)
Palabras Clave: Bender Gestalt Test, computational model, convolutional neural network, data augmentation, transfer learning
Clasificación temática: Q Science > QA Mathematics > QA76 Computer software
Usuario Remitente: Admin Administrador del Respositorio
Depositado: 13 Ago 2020 16:14
Ultima Modificación: 13 Ago 2020 16:14
URI: http://repositorio.uach.mx/id/eprint/291

Actions (login required)

Ver Objeto Ver Objeto

Universidad Autonoma de Chihuahua