Powered by OpenAIRE graph
Found an issue? Give us feedback
addClaim

Ataques adversarios a redes neuronales artificiales con FGSM

Authors: Gómez Domingo, David;

Ataques adversarios a redes neuronales artificiales con FGSM

Abstract

En este documento se presenta la memoria de la estancia en prácticas realizada en el Instituto de Tecnología Cerámica y el Trabajo de Fin de Grado, en el que se profundiza sobre los aspectos teóricos y prácticos presentes en la generación de ejemplos adversarios y los ataques y defensas que se pueden llevar a cabo con ellos. Las labores realizadas durante la estancia en prácticas, se pueden diferenciar en dos grandes bloques. En el primero, se desempeñaron distintas tareas propias del campo de la minería de datos aplicadas al sector de la cerámica, mientras que en el segundo bloque, se centró la atención en un proyecto cuyo objetivo es crear una herramienta de visualización y análisis de datos relevantes para las empresas de este sector. Finalmente, para el TFG se ha realizado un análisis teórico y práctico sobre los ejemplos adversarios. La parte teórica se centra en la demostración matemática que prueba que el Fast Gradient Sign Method (FGSM) generalizado tiene efecto de regularización. Por lo que respecta a la parte práctica, en ella se exponen los resultados obtenidos durante la realización de ataques adversarios a dos populares arquitecturas de redes neuronales previamente entrenadas como son Inception V3 y MobileNet V2.

Treball Final de Grau en Matemàtica Computacional. Codi: MT1030. Curs: 2018/2019

Country
Spain
Related Organizations
Keywords

Grau en Matemàtica Computacional, efecto de regularización, regularization effect, redes neuronales artificiales, adversarial examples, machine learning, ejemplos adversarios, FGSM, aprendizaje automático, Bachelor's Degree in Computational Mathematics, Grado en Matemática Computacional, artificial neural networks

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    0
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Average
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Average
    OpenAIRE UsageCounts
    Usage byUsageCounts
    visibility views 55
  • 55
    views
    Powered byOpenAIRE UsageCounts
Powered by OpenAIRE graph
Found an issue? Give us feedback
visibility
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
views
OpenAIRE UsageCountsViews provided by UsageCounts
0
Average
Average
Average
55
Upload OA version
Are you the author of this publication? Upload your Open Access version to Zenodo!
It’s fast and easy, just two clicks!