Machine Learning on Commodity Tiny Devices: Theory and Practice

This book explores the synergy between tiny machine learning (TinyML) software and hardware for edge intelligence applications, presenting on-device learning techniques for model-level neural network design, algorithm-level training optimization, and hardware-level instruction acceleration. It addresses the limitations of conventional in-cloud computing and highlights the promising direction of on-device learning for edge intelligence. The book discusses the latest research progress in TinyML, including neural network design, training algorithm optimization, and domain-specific hardware acceleration, and provides system-level insights on designing TinyML frameworks. It also identifies the main challenges when deploying TinyML tasks in the real world and guides researchers to deploy reliable learning systems.

Format: Hardback
Length: 250 pages
Publication date: 07 November 2022
Publisher: Taylor & Francis Ltd

This book aims to explore the synergy between tiny machine learning (TinyML) software and hardware for edge intelligence applications. It delves into on-device learning techniques, encompassing neural network design, algorithm-level training optimization, and hardware-level instruction acceleration. By examining the limitations of conventional in-cloud computing, it becomes evident that on-device learning holds great promise in meeting the demands of edge intelligence applications.

In the realm of cutting-edge research in TinyML, one of the fundamental challenges lies in implementing a high-efficiency learning framework and enabling system-level acceleration. This book provides a comprehensive exploration of the latest research advancements and offers insightful system-level insights into designing TinyML frameworks. It covers various aspects, such as neural network design, training algorithm optimization, and domain-specific hardware acceleration. Furthermore, it identifies the primary challenges encountered when deploying TinyML tasks in the real world and provides guidance for researchers in deploying reliable learning systems.

This book is of immense interest to students and scholars in the field of edge intelligence, particularly those with a solid foundation in professional Edge AI skills. It serves as an excellent guide for researchers seeking to implement high-performance TinyML systems. By leveraging the combined strengths of TinyML software and hardware, this book aims to empower developers to create intelligent and efficient solutions for edge computing applications.

Dimension: 254 x 178 (mm)
ISBN-13: 9781032374239