Rule-based knowledge discovery of satellite imagery using evolutionary classification tree

Abstract

The classification tree (CT) may be used to establish explicit classification rules for Satellite Imagery (SI). However, the accuracy of explicit classification rules attained by this method is poor. Back-propagation networks (BPN) and the support vector machine (SVM) may both be used to establish highly accurate models for predicting the classification of SI. However, neither is able to generate explicit rules. This study proposes the evolutionary classification tree (ECT) as a novel mining rule method. Composed of the particle bee algorithm (PBA) and classification tree (CT), the ECT produces self-organized rules automatically to predict the classification of SI. In ECT, CT serves as the architecture to represent explicit rules and PBA acts as the optimization mechanism to optimize CT in order to fit the experimental data. A total of 600 experimental datasets were used to compare the accuracy and complexity of four model-building techniques: CT, BPN, SVM, and ECT. The results demonstrate the ability of ECT to produce rules that are more accurate than CT and SVM but less accurate than BPN. However, because BPN is black box model, the ability of ECT to generate explicit rules makes ECT the best model for users wanting to mine the explicit rules and knowledge in practical applications.