Please use this identifier to cite or link to this item: http://hdl.handle.net/10553/23943
Title: Experimentación y comparativa de diferentes modelos de redes neuronales artificiales para el procesamiento del lenguaje natural
Authors: Jiménez Perera, Gabriel
Director: Guerra Artal, Cayetano Nicolás 
UNESCO Clasification: 120317 Informática
Keywords: Procesamiento
Lenguaje natural
DNC
MemN2N
LSTM
Memoria
Neuronal
Issue Date: 2017
Abstract: El procesamiento del lenguaje natural ha sido tradicionalmente una tarea compleja y poco trivial en el diseño de algoritmos. Gracias a la inteligencia artificial, se han logrado grandes progresos en este entorno y número de modelos que hacen frente a estos problemas normalmente poco tratables ha ido incrementando. Este proyecto propone experimentar y comparar tres modelos de redes neuronales artificiales que han tenido bastante éxito en el procesamiento del lenguaje natural: LSTM (Long Short-Term Memory), MemN2N (modelo propuesto por Facebook) y DNC (modelo propuesto por Google). Para esta tarea, estos modelos optimizados han sido adaptados a un ámbito concreto, con el objetivo de comparar los resultados de cada uno.
Natural language processing has traditionally been a complex, hardly-trivial task in algorithm design. Thanks to artificial intelligence, great progress has been made in this environment and the number of models that face these usually hardly treatable problems has increasingly grown.This project proposes experimenting and comparing three artificial neural network models that have had quite accomplishment in natural language processing: LSTM (Long Short-Term Memory), MemN2N (model proposed by Facebook) and DNC (model proposed by Google). For this task, these optimized models have been adapted to a concrete scope, with the objective of comparing the results of each.
URI: http://hdl.handle.net/10553/23943
Rights: by-nc-nd
Appears in Collections:Trabajo final de grado
Thumbnail
Memoria
Adobe PDF (2,81 MB)
Show full item record

Page view(s)

15
checked on Feb 22, 2021

Download(s)

67
checked on Feb 22, 2021

Google ScholarTM

Check


Share



Export metadata



Items in accedaCRIS are protected by copyright, with all rights reserved, unless otherwise indicated.