Computational Biology and Chemistry
| Hi-index | 0.00 |
Feature selection is necessary to reduce the dimensional- ity of spectral image data. Particle swarm optimization was originally developed to search only continuous spaces and, although many applications on discrete spaces had been proposed, it could not tackle the problem of feature selec- tion directly. We developed a formulation utilizing two par- ticles swarms in order to optimize a desired performance criterion and the number of selected features, simultane- ously. Candidate feature sets were evaluated on a regres- sion problem modeled using neural networks, which were trained to construct models of chemical concentration of glucose in soybeans. We present experimental results uti- lizing real-world spectral image data to attest the viability of the method. The particle swarms approach presented su- perior performance for linear modeling of chemical con- tents when compared to a conventional feature extraction method.