The self-organizing network can map data sets from high-dimensional to low-dimensional topologies. The forest data of Cambodia were applied to train Kohonen's Self-Organizing Feature Maps (SOFM). After training we have the neighborhood structure among the training data implicit on the map. Using visualization tools, in particular the U-matrix method, the neighbourhood structure of learned SOFM can be visually recognized. This method also facilitates to visualize the clusters presented in the input data. The algorithm takes the training data with the classification detected through SOFM as input and generates symbolic rules.
Rule extraction extracts forest classes produced by self-organizing methods for the queries of knowledge. This is a good method to learn cluster techniques to automate the process of classification of certain scientific data, thereby reducing the human effort required. The forest classifications in Cambodia can then be drawn on the geographical map in different colors for different classes. However, the knowledge acquisition of the rule extraction can be further interpreted by the geological expert or can be used thru knowledge-based system. Rule extraction algorithms significantly enhance the capabilities of artificial neural networks to explore data for the benefit of the users.