Many Hands Make Light Work-On Ensemble Learning Techniques for Data Fusion in Remote Sensing

In this paper we discuss the use of ensemble methods in remote sensing. After a review of the relevant state of the art in ensemble learning - inside and outside the remote sensing community - we provide the necessary theoretical background of this research field. This includes a discussion of the bias/ variance tradeoff that is a key notion in machine learning and especially ensemble learning. We provide a review of three of the most relevant and prominent techniques in ensemble learning, namely the Random Forest, Extra Trees and the Gradient Boosted Regression Trees algorithms. All algorithms are assessed in terms of their theoretical properties as well as applicability for remote sensing use cases. Finally, in the experimental section we compare their performance in challenging remote sensing datasets with different properties, while discussing again the reasons that the mechanics of each algorithm might give it an advantage under certain conditions.