Please use this identifier to cite or link to this item: http://hdl.handle.net/2445/176952
Title: Introducción y optimización estocástica de redes neuronales profundas MLP
Author: Deulofeu Gómez, Rubén
Director/Tutor: Pujol Vila, Oriol
Keywords: Aprenentatge automàtic
Treballs de fi de grau
Xarxes neuronals (Informàtica)
Resolució de problemes
Representació del coneixement (Teoria de la informació)
Machine learning
Bachelor's thesis
Neural networks (Computer science)
Problem solving
Knowledge representation (Information theory)
Issue Date: 21-Jun-2020
Abstract: [en] Since technology has stressed so much our society, human beings have always dreamed about to achieve the most. Among all those dreams, some more reachables than others, there is the ability to create machines that think by themselves. This chimera, despite of being quite different from how our ancestors ever imagined, is nowadays at the summit of its history. The massive increase of data, through digitalization, and the constant technological improvements, in this case, in the form of progress in high performance hardware production, have been the main drivers of this change. This grade project covers a specific area that is, sometimes, a part of the aforementioned creation process, known as Artificial Intelligence. This specific area is called Deep Learning. Deep learning techniques can be regarded as the algorithms that exploit data using models based on non-linear function compositions. In this respect, optimization plays a crucial role as it is the driver that transform the information in data into the model parameters. The project aims at understanding the mathematical basis of optimization, namely stochastic optimization theory and its application to deep learning.
Note: Treballs Finals de Grau de Matemàtiques, Facultat de Matemàtiques, Universitat de Barcelona, Any: 2020, Director: Oriol Pujol Vila
URI: http://hdl.handle.net/2445/176952
Appears in Collections:Treballs Finals de Grau (TFG) - Matemàtiques

Files in This Item:
File Description SizeFormat 
176952.pdfMemòria1.15 MBAdobe PDFView/Open


This item is licensed under a Creative Commons License Creative Commons