An assessment of support vector machines for land cover classification

TLDR

Support vector machines are theoretically superior machine learning algorithms. This study introduces the theoretical development of SVMs and experimentally evaluates their accuracy, stability, and training speed for land‑cover classification from satellite imagery. The authors compared SVMs to maximum‑likelihood, neural‑network, and decision‑tree classifiers, assessing the effects of kernel choice and training data selection on performance. SVMs proved competitive with the best available algorithms for classifying high‑dimensional datasets.

Abstract

The support vector machine (SVM) is a group of theoretically superior machine learning algorithms. It was found competitive with the best available machine learning algorithms in classifying high-dimensional data sets. This paper gives an introduction to the theoretical development of the SVM and an experimental evaluation of its accuracy, stability and training speed in deriving land cover classifications from satellite images. The SVM was compared to three other popular classifiers, including the maximum likelihood classifier (MLC), neural network classifiers (NNC) and decision tree classifiers (DTC). The impacts of kernel configuration on the performance of the SVM and of the selection of training data and input variables on the four classifiers were also evaluated in this experiment.