Abstract
In this article, a novel Local Experts Organization (LEO) model for processing tree structures with its application of natural scene images classification is presented. Instead of relatively poor representation of image features in a flat vector form, we proposed to extract the features and encode them into a binary tree representation. The proposed LEO model is used to generalize this tree representation in order to perform the classification task. The capabilities of the proposed LEO model are evaluated in simulations running under different image scenarios. Experimental results demonstrate that the LEO model is consistent in terms of robustness amongst the other tested classifiers.
Original language | English |
---|---|
Pages (from-to) | 83-99 |
Number of pages | 17 |
Journal | Neural Processing Letters |
Volume | 26 |
Issue number | 2 |
DOIs | |
Publication status | Published - Oct 2007 |
Externally published | Yes |
Keywords
- Image classification
- Multivariate polynomial model
- Support Vector Machines
ASJC Scopus subject areas
- Software
- Neuroscience (all)
- Computer Networks and Communications
- Artificial Intelligence