Mateus S. Gilbert

Mateus S. Gilbert

D.Sc. student

UFRJ

Biography

Mateus S. Gilbert is a D.Sc. student in Electrical Engineering enrolled in the Electrical Engineering Program of the Alberto Luiz Coimbra Institute of Graduate Studies and Research in Engineering (PEE/COPPE/UFRJ) under the co-orientation of Miguel Elias M. Campista, D.Sc., Marcello L. R. de Campos, Ph.D., and Marcelo Dias de Amorim, Dr., since march 2024.

In 2024, he received his M.Sc. in Electric Engineering at the PEE/COPPE/UFRJ. Previously in 2022, he received his B.Sc. in Electronic and Computer Engineering at the UFRJ Polytechnic School and B.Math at the UFRJ Intitute of Mathematics. He is a research student in the Grupo de Teleinformática e Automação (GTA), since 2019. His research interests include neural network and deep learning, IoT, wireless communication, computer vision, cloud and fog computing, and data science.

Interests
  • Neural Networks
  • Deep Learning
  • IoT
  • Computer Vision
  • Wireless Communication
  • Cloud and Fog Computing
  • Data Science
Education

  • DSc in Electric Engineering

    COPPE/UFRJ

  • MSc in Electric Engineering, 2024

    COPPE/UFRJ

  • BMath, 2022

    UFRJ Institute of Mathematics

  • BSc in Electronic and Computer Engineering, 2022

    UFRJ Polytechnic School

Recent Publications

Search for an older work by filtering publications.
Asymmetric Autoencoders: an NN alternative for resource-constrained devices in IoT networks
Many recent publications resort to DNNs to deal with IoT data. Models with multiple layers and other parameters can be hard to implement with resource-constrained devices given its computational and energy demands. This work proposes an autoencoder models that aims to mitigate these issues.
Mateus S. Gilbert, Marcello L. R. de Campos, Miguel Elias M. Campista
Apr 1, 2024
PDF Cite DOI
Unlocking Early-Exiting Semantic Segmentation with Branched Networks
Given the success of Early-Exit DNNs (EE-DNNs) with image classification, this work aims to construct an EE-DNN capable of performing semantic segmentation, which can be more challenging as it requires pixel-level classification.
Mateus S. Gilbert, Roberto G. Pacheco, Rodrigo S. Couto, Marcello L. R. de Campos, Miguel Elias M. Campista
Dec 25, 2023
PDF Cite DOI
Towards edge computing using early-exit convolutional neural networks
Early-exit CNNs for edge-cloud co-inference. Analysis of a branched neural network designed to minimize cloud offloading. Employs an architecture where layers are distributed between edge and cloud instances, and that enables classifications attempts in earlier stages.
Roberto G. Pacheco, Kaylani Bochie, Mateus S. Gilbert, Rodrigo S. Couto, Miguel Elias M. Campista
Oct 19, 2021
PDF Cite DOI

Contact