Semi-supervised relevance index for feature selection
| dc.creator | Frederico Gualberto Ferreira Coelho | |
| dc.creator | Cristiano Leite de Castro | |
| dc.creator | Antônio Braga | |
| dc.creator | Michel Verleysen | |
| dc.date.accessioned | 2025-04-07T13:35:39Z | |
| dc.date.accessioned | 2025-09-09T00:13:25Z | |
| dc.date.available | 2025-04-07T13:35:39Z | |
| dc.date.issued | 2017 | |
| dc.identifier.doi | 10.1007/s00521-017-3062-0 | |
| dc.identifier.issn | 1433-3058 | |
| dc.identifier.uri | https://hdl.handle.net/1843/81329 | |
| dc.language | eng | |
| dc.publisher | Universidade Federal de Minas Gerais | |
| dc.relation.ispartof | Neural computing and applications | |
| dc.rights | Acesso Restrito | |
| dc.subject | Indices | |
| dc.subject | Fontes de informação | |
| dc.subject.other | This work presented a new method for feature selection that is capable of considering sources of information from labeled as well as from unlabeled data. In this semi-supervised feature selection framework, the method is based on the idea of eliminating redundancy by feature clustering. | |
| dc.subject.other | The method adopts a novel semi-supervised approach, since labeled and unlabeled data are taken into account in the new similarity index, which is also proposed in this work. SSFC can be directly applied to multiple variables by incorporating them to the MI estimation. | |
| dc.subject.other | Stopping criterion for feature clustering can also incorporate further overall performance strategies, since it is based only on the significance level of S. The method, however, achieved competitive results with less features then previous works in the literature with the same data sets. It is interesting to highlight that the proposed method performed well even with a small number of labeled data. | |
| dc.title | Semi-supervised relevance index for feature selection | |
| dc.type | Artigo de periódico | |
| local.citation.epage | 9 | |
| local.citation.spage | 1 | |
| local.citation.volume | 28 | |
| local.description.resumo | This paper presents a new relevance index based on mutual information that is based on labeled and unlabeled data. The proposed index, which is based in Mutual Information, takes into account the similarity between features and their joint influence on the output variable. Based on this principle, a method to select features is developed to eliminate redundant and irrelevant features when the relevance index value is less then a threshold value. A strategy to set the threshold is also proposed in this work. Experiments show that the new method is capable of capturing important joint relations between input and output variables, which are incorporated into a new feature selection clustering approach. | |
| local.publisher.country | Brasil | |
| local.publisher.department | ENG - DEPARTAMENTO DE ENGENHARIA ELÉTRICA | |
| local.publisher.department | ENG - DEPARTAMENTO DE ENGENHARIA ELETRÔNICA | |
| local.publisher.initials | UFMG | |
| local.url.externa | https://link.springer.com/article/10.1007/s00521-017-3062-0 |
Arquivos
Licença do pacote
1 - 1 de 1