Detalhes bibliográficos
| Ano de defesa: |
2017 |
| Autor(a) principal: |
Caldas, Weslley Lioba |
| Orientador(a): |
Não Informado pela instituição |
| Banca de defesa: |
Não Informado pela instituição |
| Tipo de documento: |
Dissertação
|
| Tipo de acesso: |
Acesso aberto |
| Idioma: |
por |
| Instituição de defesa: |
Não Informado pela instituição
|
| Programa de Pós-Graduação: |
Não Informado pela instituição
|
| Departamento: |
Não Informado pela instituição
|
| País: |
Não Informado pela instituição
|
| Palavras-chave em Português: |
|
| Link de acesso: |
http://www.repositorio.ufc.br/handle/riufc/24913
|
Resumo: |
Semi-supervised learning is an important fild of machine learning, combining the use of labeled data with unlabelled data, and has gained attention of academic community in the last years. This is mainly due to the large amount of data available and the work required to label these data, making semi-supervised learning an attractive methodology because it requires a reduced amount of labeled data. Regarding the various approaches of semi-supervised learning, Co-Training has become popular because of its simple formulation and promising results in different areas. In this work, we propose Co-MLM, a semi-supervised method that uses the Minimal Learning Machine (MLM), a recent proposed supervised method, in conjunction with the Co-Training methodology. In addition, we also propose a quick version of this same method, named Fast Co-MLM, using as base classifier the NN-MLM method, an MLM variant. Both methods were compared using data sets from the UCI, UCF and DataGov repositories, demonstrating ability to learn about unlabeled data, and promising results when compared with other Co-training based algorithms. |
