Classification of imbalanced land-use/land-cover data using variational semi-supervised learning

Abstract

Classification of Land Use/Land Cover (LULC) data is a typical task in remote-sensing domain. However, because the classes distribution in LULC data is naturally imbalance, it is difficult to do the classification. In this paper, we employ Variational Semi-Supervised Learning (VSSL) to solve imbalance problem in LULC of Jakarta City. This VSSL exploits the use of semi-supervised learning on deep learning model. Therefore, it is suitable for classifying data with abundant unlabeled like LULC. The result shows that VSSL achieves 80.17% of overall accuracy, outperforming other algorithms in comparison.