Algorithmic Background for Machine Learning

This chapter provides a detailed discussion of the machine learning models typically deployed on (extreme) edge devices. This is necessary to understand the rest of the book, which delves into the hardware-algorithm co-optimization and hardware acceleration of these models. The first part presents the basic concepts of one of the classical machine learning models, i.e., support vector machines (SVMs), followed by a discussion on a sub-field of machine learning (ML), called deep learning (DL), and the models used in it. Rather than giving a detailed overview of the wide range of models used in machine learning, these sections focus on the models extensively used in the subsequent chapters of this book. Finally, the chapter presents some of the standard feature extraction algorithms used in the machine learning pipeline for the applications accelerated in this book.