Название: Neuromorphic Computing Principles and Organization, 2nd EditionАвтор: Abderazek Ben Abdallah, Khanh N. Dang
Издательство: Springer
Год: 2025
Страниц: 328
Язык: английский
Формат: pdf (true), epub
Размер: 46.0 MB
The second edition of Neuromorphic Computing Principles and Organization delves deeply into neuromorphic computing, focusing on designing fault-tolerant, scalable hardware for spiking neural networks. Each chapter includes exercises to enhance understanding. All existing chapters have been meticulously revised, and a new chapter on advanced neuromorphic prosthesis design serves as a comprehensive case study.
The book starts with an overview of neuromorphic systems and fundamental artificial neural network concepts. It explores artificial neurons, neuron models, storage technologies, inter-neuron communication, learning mechanisms, and design approaches. Detailed discussions cover challenges in constructing spiking neural networks and emerging memory technologies. A dedicated chapter addresses circuits and architectures, including Network-on-Chip (NoC) fabric, Address Event Representation (AER), memory access methods, and photonic interconnects.
Reliability issues, recovery methods for multicore systems, and reconfigurable designs supporting multiple applications are examined. The book also describes the hardware-software design of a three-dimensional neuromorphic processor, focusing on high integration density, minimal spike delay, and scalable design. The book concludes with a comprehensive review of neuromorphic systems, providing a detailed analysis of the field and an overarching understanding of the key concepts discussed throughout the text.
One of the standout features of this edition is its exploration of emerging memory technologies and their integration into neuromorphic systems. Memory has long been a critical bottleneck in computing systems, and the book addresses this challenge head-on. From photonic interconnects to Address Event Representation (AER) protocols, the authors provide innovative solutions and forward-looking insights that are both timely and impactful.
The chapter dedicated to circuits and architectures is another highlight. The discussions on Network-on-Chip (NoC) fabric, memory access methods, and fault tolerance are particularly relevant in building scalable and reliable neuromorphic systems. These concepts are theoretically sound and rooted in practical implementation, reflecting the authors’ deep understanding of the interplay between hardware and software in neuromorphic design.
Reliability and scalability are recurring themes throughout the book. The emphasis on recovery methods for multicore systems and reconfigurable designs capable of supporting multiple applications is especially significant. These features are essential for building systems adapting to the dynamic demands of modern computing environments. The book’s exploration of a three-dimensional neuromorphic processor is a testament to this adaptability. By focusing on high integration density, minimal spike delay, and scalability, the authors provide a roadmap for future innovations in neuromorphic hardware.
Contents:
Скачать Neuromorphic Computing Principles and Organization, 2nd Edition