Machine learning experienced a great advance in the 1980s and 1990s due to the active research in artificial neural networks, adaptive schemes and fuzzy systems. These methodologies demonstrated good results in many real applications, especially for classification and regression tasks, since neither a priori knowledge about the model of the distribution of the available data nor the relationships among the independent variables should be necessarily assumed. These desirable properties are at the basis of the success of these methods in the field of the analysis of remote sensing images, where a wide literature refers to the definition of classifiers and estimation algorithms based on neural networks and fuzzy systems.