An investigation of multiple self-organizing feature maps for classification of multisource data

Multiple Self-Organizing Feature Maps (MSOMs) can be considered attractive for classification of remote sensing data with many input features. The MSOMs have several advantages, e.g., they are non-parametric, the computational cost for them only grows linearly as a function of the number of features, and they have been shown to approximate posterior probabilities. In the paper MSOMs are investigated for classification of a multisource remote sensing and geographic data set. In the experiments, the MSOM showed potential for classification of the multisource data set.