Please use this identifier to cite or link to this item:
http://acervodigital.unesp.br/handle/11449/71961
- Title:
- Optimizing optimum-path forest classification for huge datasets
- Universidade Estadual Paulista (UNESP)
- Universidade Estadual de Campinas (UNICAMP)
- 1051-4651
- Traditional pattern recognition techniques can not handle the classification of large datasets with both efficiency and effectiveness. In this context, the Optimum-Path Forest (OPF) classifier was recently introduced, trying to achieve high recognition rates and low computational cost. Although OPF was much faster than Support Vector Machines for training, it was slightly slower for classification. In this paper, we present the Efficient OPF (EOPF), which is an enhanced and faster version of the traditional OPF, and validate it for the automatic recognition of white matter and gray matter in magnetic resonance images of the human brain. © 2010 IEEE.
- 18-Nov-2010
- Proceedings - International Conference on Pattern Recognition, p. 4162-4165.
- 4162-4165
- Brain image classification
- Optimum-Path forest
- Supervised classification
- Support Vector machines
- Automatic recognition
- Brain images
- Computational costs
- Data sets
- Forest classification
- Gray matter
- Human brain
- Large datasets
- Magnetic resonance images
- Pattern recognition techniques
- Recognition rates
- Support vector
- White matter
- Image analysis
- Image classification
- Magnetic resonance
- Magnetic resonance imaging
- Support vector machines
- Classification (of information)
- http://dx.doi.org/10.1109/ICPR.2010.1012
- Acesso restrito
- outro
- http://repositorio.unesp.br/handle/11449/71961
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.