Development of a Machine Learning Platform

• Autor(a) principal: David, Henrique Miguel Castilho
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/160388
• topic: Machine Learning; Machine Learning Platform; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Adoption of machine learning is becoming widespread, thus, it is natural to see a more comprehensive adoption of this technology by companies to, not only to enhance their products and services, but also to offer greater market competitiveness. Having said that, and attending to this new paradigm, the present dissertation is focused on the implementation of a platform to optimize and enhance the development of projects in the area of machine learning. This challenge arises from a proposal put forward by company GMV, which aims to make the machine learning process more accessible and intuitive for its workers and, in parallel, to ensure high levels of consistency and productivity in the development of its projects. Based on all these assumptions, a first approach is made in this dissertation, laying both on how a machine learning project is organized as well as on the problems that arise throughout its development. First, a study was made of the functioning of some platforms already present in the market, in order to understand which problems they intend to solve and which solution or solutions have been developed to address them. Then, the characteristics to be integrated in the platform were identified. The study and comparison of some technologies present in the market allowed us to select and implement the most promising ones regarding the characteristics previously identified. Finally, the proposed solution is presented, explaining both the functioning of the platform and the options taken throughout its development.