Introduction

This chapter introduces the recent trend of deploying machine learning workloads on resource- and energy-constrained (extreme) edge devices, detailing the opportunities and challenges in this paradigm. This (extreme) edge computing paradigm is broken down into three fundamental blocks: application, algorithm, and hardware, and a holistic approach of cross-domain optimization is proposed. Subsequently, four open research questions are discussed, targeting fast exploration and implementation of flexible, high-performance, energy-efficient, and low-power hardware design. Finally, this book’s main contributions to addressing these research questions through a holistic hardware-algorithm co-design are summarized.