dc.contributorMaziero, Jonas
dc.contributorhttp://lattes.cnpq.br/1270437648097538
dc.contributorDuzzioni, Eduardo Inacio
dc.contributorMarchi, Jerusa
dc.contributorRibeiro, Alexandre Dias
dc.contributorMombach, José Carlos Merino
dc.creatorFarias, Tiago de Souza
dc.date.accessioned2023-03-17T15:27:13Z
dc.date.accessioned2023-09-04T19:48:04Z
dc.date.available2023-03-17T15:27:13Z
dc.date.available2023-09-04T19:48:04Z
dc.date.created2023-03-17T15:27:13Z
dc.date.issued2023-02-01
dc.identifierhttp://repositorio.ufsm.br/handle/1/28266
dc.identifier.urihttps://repositorioslatinoamericanos.uchile.cl/handle/2250/8627866
dc.description.abstractReversible neural networks are a type of neural network where you can recover the input values knowing only the output values of the network. This thesis presents a method to approximate the reversibility of neural networks, where a neural network is trained to approximate the input values through the gradient of a cost function that depends on the output values. Applied in generative processes, reversibility allows generating data statistically similar to the training set. With a change in the proposed reversibility technique, it is possible to make local training of a neural network, saving computational memory resources, which can be applied to arbitrary problems such as classification. Differentiable programming is a computing paradigm where a program is built from differentiable blocks, offering the advantages of differentiability, which can be used to modify the program according to a data set and an objective function, as well as scalability, where a program can be run on hardware that offers high parallelism capability, such as GPU and TPU. This thesis presents the use of differentiable programming to approximate the solution of differential equations, demonstrating its ability to help solve physical problems that can be represented by this type of equation. Another developed differentiable programming application in spin models, which can be used to simulate a variety of phenomena such as magnetic materials, graphs and biological cells, offering advantages in scalability and execution time.
dc.publisherUniversidade Federal de Santa Maria
dc.publisherBrasil
dc.publisherFísica
dc.publisherUFSM
dc.publisherPrograma de Pós-Graduação em Física
dc.publisherCentro de Ciências Naturais e Exatas
dc.rightshttp://creativecommons.org/licenses/by-nc-nd/4.0/
dc.rightsAttribution-NonCommercial-NoDerivatives 4.0 International
dc.subjectAprendizado de máquina
dc.subjectRede neural
dc.subjectReversibilidade
dc.subjectProcessos gerativos
dc.subjectTreinamento local
dc.subjectProgramação diferenciável
dc.subjectEquações diferenciais
dc.subjectModelos de spin
dc.subjectMachine learning
dc.subjectNeural network
dc.subjectReversibility
dc.subjectGenerative processes
dc.subjectLocal training
dc.subjectDifferentiable programming
dc.subjectDifferential equations
dc.subjectSpin models
dc.titleRedes neurais reversíveis e caracterização de problemas físicos através de programação diferenciável
dc.typeTese


Este ítem pertenece a la siguiente institución