Embedded Deep Learning: Algorithms, Architectures and Circuits for Always-on Neural Network Processing

This book discusses algorithmic and hardware implementation techniques to enable embedded deep learning, with four silicon prototypes demonstrating the impact of these techniques. It covers optimization of neural networks for embedded deployment, efficient Convolutional Neural Network processors, and cross-layer design concepts.

\n Format: Hardback
\n Length: 206 pages
\n Publication date: 03 November 2018
\n Publisher: Springer International Publishing AG
\n

This comprehensive book delves into algorithmic and hardware implementation techniques to empower embedded deep learning systems. The authors present synergistic design approaches spanning application, algorithmic, computer architecture, and circuit levels, aimed at significantly reducing the computational costs of deep learning algorithms. The impact of these techniques is showcased through four silicon prototypes, demonstrating their practical applications in the field of embedded deep learning.

Gives a comprehensive overview of a range of efficient solutions for energy-efficient neural networks on battery-constrained wearable devices.

Discusses the optimization of neural networks for embedded deployment across various design hierarchies, including applications, algorithms, hardware architectures, and circuits, with support from real silicon prototypes.

Elaborates on the design of efficient Convolutional Neural Network processors, leveraging parallelism, data reuse, sparse operations, and low-precision computations.

Supports the introduced theory and design concepts through four real silicon prototypes, providing detailed discussions on their physical realizations, implementation, and achieved performances.

Illustrates and highlights the introduced cross-layer design concepts through comprehensive discussions, providing a deeper understanding of the underlying principles and their practical applications.

\n Weight: 462g\n
Dimension: 165 x 242 x 17 (mm)\n
ISBN-13: 9783319992228\n
Edition number: 1st ed. 2019\n