A basic introduction to deep learning

The early 2010s witnessed an important phase in machine learning, marked by the rise of deep learning. Fueled by advancements in algorithms, hardware, and vast datasets, deep learning transformed many fields, including Earth observation. This book attempts to capture this evolution, with a particular emphasis on multi-sensor Earth observation. Understanding the details of multi-sensor architectures and their practical applications requires an understanding of the basic principles of deep learning and basic architectures used in tasks like image classification and semantic segmentation. This chapter serves this purpose by providing a brief overview of deep learning, tailored specifically to the requirements of Earth observation applications. Through this overview, the chapter establishes the foundation for further detailed discussions in the subsequent chapters. As such, the chapter may also be useful to researchers working in other areas such as medical image analysis or core computer vision.