Zheng N., Mazumder P. Learning in Energy-Efficient Neuromorphic Computing: Algorithm and Architecture Co-Design

New York: Wiley-IEEE Press, 2020. — 289 p.Explains current co-design and co-optimization methodologies for building hardware neural networks and algorithms for machine learning applications.
Machine learning, especially deep learning, has emerged as an important discipline through which many conventionally difficult problems, such as pattern recognition, decision making, and natural language processing, can be addressed. Nowadays, millions and even billions of neural networks are running in data centers, personal computers and portable devices to perform various tasks. In the future, it is expected that more complex neural networks with larger sizes will be needed. Such a trend demands specialized hardware to accommodate the ever-increasing requirements on power consumption and response time.
In this book, we focus on the topic of how to build energy-efficient hardware for neural networks with a learning capability. This book strives to provide co-design and co-optimization methodologies for building hardware neural networks that can learn to perform various tasks. The book provides a complete picture from high-level algorithms to low-level implementation details. Hardware-friendly algorithms are developed with the objective to ease implementation in hardware, whereas special hardware architectures are proposed to exploit the unique features of the algorithms.
- Includes cross-layer survey of hardware accelerators for neuromorphic algorithms
- Covers the co-design of architecture and algorithms with emerging devices for much-improved computing efficiency
- Focuses on the co-design of algorithms and hardware, which is especially critical for using emerging devices, such as traditional memristors or diffusive memristors, for neuromorphic computing
Learning in Energy-Efficient Neuromorphic Computing: Algorithm and Architecture Co-Design is an ideal resource for researchers, scientists, software engineers, and hardware engineers dealing with the ever-increasing requirement on power consumption and response time. It is also excellent for teaching and training undergraduate and graduate students about the latest generation neural networks with powerful learning capabilities.OverviewHistory of Neural Networks
Neural Networks in Software
Artificial Neural Network
Spiking Neural Network
Need for Neuromorphic Hardware
Objectives and Outlines of the Book
Fundamentals and Learning of Artificial Neural Networks
Operational Principles of Artificial Neural Networks
Inference
Learning
Neural Network Based Machine Learning
Supervised Learning
Reinforcement Learning
Unsupervised Learning
Case Study: Action-Dependent Heuristic Dynamic Programming
Deep Learning
Artificial Neural Networks in HardwareOverviewGeneral-Purpose Processors
Digital Accelerators
A Digital ASIC Approach
Analog/Mixed-Signal Accelerators
Case Study: An Energy-Efficient Accelerator for Adaptive Dynamic Programming
Operational Principles and Learning in Spiking Neural Networks
Spiking Neural Networks
Learning in Shallow SNNs
Learning in Deep SNNs
Hardware Implementations of Spiking Neural Networks
The Need for Specialized Hardware
Digital SNNs
Analog/Mixed-Signal SNNs