Clusterização intervalar incremental bottom-up a partir de fluxos de dados intervalares

• Ano de defesa: 2020
• Autor(a) principal: Pereira, Tamyres
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Programa de Pós-graduação em Engenharia de Sistemas e Automação; UFLA; brasil; Departamento de Engenharia
• Palavras-chave em Português: Clusterização incremental; Aprendizado de máquina; Fluxos de dados; Matemática intervalar; Computação granular; Incremental clustering; Machine learning; Data streams; Interval mathematics; Granular computing; Teoria da Computação
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/41472

This work proposes a method of bottom-up incremental interval clustering from interval data streams. The method is supported by concepts, definitions and mathematical tools of the gra- nular computation theory, in particular interval algebra. Differently from other evolutionary methods of processing and modeling numerical data flows, the proposed method deals with data streams that exhibits unstructured uncertainty represented by interval values, and also nu- merical data streams as a particular case. The proposed method is able to model complex processes presented as a data stream and subject to changes in the environment. The learning algorithm develop the structure of the model in a bottom-up manner, without prior knowledge about of the process, and adapts the parameters of the model as needed, thus avoiding that the model be reconstructed and retrained when there is a change in the environment or system - this being a clear advantage over pre-designed models based on specialized knowledge or historical data. For the development of granules (local models), the learning algorithm is equipped with recursive formulas to calculate the similarity between interval objects and with the Xie-Beni incremental validation index.