Machine Learning and Deep Learning in Remote Sensing Data Analysis

Abstract

Machine and deep learning have revolutionized remote sensing by extracting valuable information from large datasets and contributing to advancements in remote sensing applications. Viana et al. (2023) recently raised four fundamental questions to help land change model users decide which model to use: “(1) Can the user understand the model? (2) Can the audience understand the model? (3) Can the user control the model? (4) Does the model address the goals of the specific application?” We argued that one should ask the same questions before choosing a machine learning or deep learning model in remote sensing data analysis and applications. This chapter aims to help users answer the questions (1) and (2) by describing the history, principles, mathematical formulations, and training methods for commonly used machine learning and deep learning methods in remote sensing. Each method description is followed by a list of literature demonstrating the method applications. The deep learning method will be introduced in a way to highlight the difference between the deep learning originated problems and the remote sensing problems, as well as the caveats in using deep learning in remote sensing data analysis. It is difficult to comprehensively explain diverse machine learning and deep learning architectures and training techniques in one chapter. However, this chapter covers their introduction by citing seminal literature on method and selected literature on application that readers can refer to.