dc.contributor | Maziero, Jonas | |
dc.contributor | http://lattes.cnpq.br/1270437648097538 | |
dc.contributor | Duzzioni, Eduardo Inacio | |
dc.contributor | Marchi, Jerusa | |
dc.contributor | Ribeiro, Alexandre Dias | |
dc.contributor | Mombach, José Carlos Merino | |
dc.creator | Farias, Tiago de Souza | |
dc.date.accessioned | 2023-03-17T15:27:13Z | |
dc.date.accessioned | 2023-09-04T19:48:04Z | |
dc.date.available | 2023-03-17T15:27:13Z | |
dc.date.available | 2023-09-04T19:48:04Z | |
dc.date.created | 2023-03-17T15:27:13Z | |
dc.date.issued | 2023-02-01 | |
dc.identifier | http://repositorio.ufsm.br/handle/1/28266 | |
dc.identifier.uri | https://repositorioslatinoamericanos.uchile.cl/handle/2250/8627866 | |
dc.description.abstract | Reversible 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.publisher | Universidade Federal de Santa Maria | |
dc.publisher | Brasil | |
dc.publisher | Física | |
dc.publisher | UFSM | |
dc.publisher | Programa de Pós-Graduação em Física | |
dc.publisher | Centro de Ciências Naturais e Exatas | |
dc.rights | http://creativecommons.org/licenses/by-nc-nd/4.0/ | |
dc.rights | Attribution-NonCommercial-NoDerivatives 4.0 International | |
dc.subject | Aprendizado de máquina | |
dc.subject | Rede neural | |
dc.subject | Reversibilidade | |
dc.subject | Processos gerativos | |
dc.subject | Treinamento local | |
dc.subject | Programação diferenciável | |
dc.subject | Equações diferenciais | |
dc.subject | Modelos de spin | |
dc.subject | Machine learning | |
dc.subject | Neural network | |
dc.subject | Reversibility | |
dc.subject | Generative processes | |
dc.subject | Local training | |
dc.subject | Differentiable programming | |
dc.subject | Differential equations | |
dc.subject | Spin models | |
dc.title | Redes neurais reversíveis e caracterização de problemas físicos através de programação diferenciável | |
dc.type | Tese | |