Vinicius Arruda
Bio
I am a M.Sc. student at Federal University of Espírito Santo in the High Performance Computing Laboratory, advised by Prof. Thiago Oliveira-Santos. My research interest lies in studying the ability of neural networks to learn general representations to perceive the visual world as we humans do. I am currently working with unsupervised domain adaptation for object detection, taking advantage of Generative Adversarial Networks to artificially generate data. I have previously interned at a startup that uses computer vision and deep learning to weigh cattle without using a scale. In my spare time, you will probably meet me at sea.
Publications
Cross-Domain Car Detection Using Unsupervised Image-to-Image Translation: From Day to Night
International Joint Conference on Neural Networks (IJCNN), 2019
Vinicius F. Arruda, Thiago M. Paixão, Rodrigo F. Berriel, Alberto F. De Souza, Claudine Badue, Nicu Sebe, Thiago Oliveira-Santos
[Paper] [Project Page] [Code] [Demo Video]
Particle Swarm Optimization for training artificial neural networks of type ELM: A case study for time series prediction
XLVIII Brazilian Symposium of Operational Research, 2016
Vinicius F. Arruda, Renato A. Krohling
[Paper]
Posts
A Brief Introduction to Artificial Neural Networks
An introduction with numerical examples solving logical gates and a simple training with genetic algorithm to play Flappy Bird.
"What I cannot create, I do not understand."
— Richard Feynman
adapted from genforce.github.io
inspired by vsitzmann.github.io