TÍTULO DA PROPOSTA DE DISSERTAÇÃO: FedWS: Advancing with HeterogeneousData on Federated Learning.
Resumo: Investments in Internet of Things and artificial intelligence have a large growth forecast for 2030, considering estimates made in 2022.People and objects are increasingly connected, raising the amount of data available for new services using machine learning, and privacy concerns. Federated learning is a technique that allows distributed machine learning without the need to transfer client data to a central server, reducing privacy issues. Nonetheless, the existence of heterogeneous data samples among clients leads to low accuracy and high communications costs. In this context, this research proposal aims to understand how current studies deal with this problem, evaluating how it developsand proposing an approach aiming to advance in the state of the art. The research methodwill be an experiment conducted on heterogeneous datasets, whose results will be used to understand its behavior and confirm the effectiveness of the proposed approach. The expected contributions are a greater understanding of the characteristics and impacts of heterogeneous distributions and a new approach aimed at reducing its negative effects.
Palavras-chave: Federated Learning, Intelligent Transportation Systems, Heterogeneous Data, Data Transformation, Image Classification
BANCA EXAMINADORA:
CARLOS ALBERTO CAMPOS VIEIRA-UNIRIO
CARLOS EDUARDO RIBEIRO DE MELLO-UNIRIO
ALINE MARINS PAES CARVALHO- IC/UFF
DATA/HORA DE DEFESA: 01/09/2023 13:00
