MACHINE LEARNING PERSPECTIVE FOR ANALYSIS OF GEOSPATIAL DATA

Abstract

The characteristics of spatially explicit data are often inadequately handled in machine learning
for spatial domains of application. At the same time, resources that can identify these properties
and explore their impacts and how machine learning applications handle them are lagging
behind. In this paper, we seek to identify and discuss the spatial properties of data that influence
the performance of machine learning. We address existing research efforts and challenges in
three main areas of machine learning: data analysis, deep learning and statistical inference. We
will also discuss the existing end-to-end systems and highlight unresolved issues and
challenges for future research in this area