Habilitando previsões de desastres em cenários de IoT com transferência de aprendizado

• Ano de defesa: 2022
• Autor(a) principal: Colantoni, Marcus Artiaga
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Ciência da Computação
• Palavras-chave em Português: Aprendizado de máquina; IoT; Transferência de aprendizado; Gerenciamento de desastres; infraestrutura virtualizada; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::SISTEMAS DE COMPUTACAO::ARQUITETURA DE SISTEMAS DE COMPUTACAO; Computação; Internet das coisas; Inteligência artificial
• Link de acesso: https://repositorio.ufu.br/handle/123456789/36517; http://doi.org/10.14393/ufu.di.2022.617

Information and communication technology have changed the way humanity sees the world around it. Faced with the maturation of the Internet of Things (IoT) and the possibilities offered by artificial intelligence, we have a greater ability to observe and estimate things and situations, based on information collected in the environment. This work investigates the use of transfer of learning as an enabling tool for monitoring remote locations, aiming at disaster prevention. The main objective is to provide a viable solution for scenarios with limited bandwidth for data communication, such as rural areas. As a way to validate the proposed architecture, we present results obtained in a prototype built in our laboratories that allows us to evaluate the accuracy of machine learning models under different conditions for the transmission of monitoring data. The results demonstrate the effectiveness of the transfer of learning technique as one of the pillars for the scalability of a disaster monitoring system.