Titlebook: Embedded Artificial Intelligence; Principles, Platform Bin Li Book 2024 Tsinghua University Press, Beijing China. 2024 Embedded Artificial

Organization

Embedded Artificial Intelligencering the two implementation modes of embedded artificial intelligence: cloud computing mode and local mode, we clarified the necessity and technical challenges of implementing the local mode and outlined the five essential components needed to overcome these challenges and achieve true embedded AI.

Organonitrile

1分开

Embedded AI Development Processfic development steps for embedded AI development, such as model optimization, conversion, compilation, deployment, etc. Finally, NVIDIA Jetson is taken as an example to introduce its special development process so that developers can gain an intuitive understanding.

nonchalance

Optimizing Embedded Neural Network Modelszation, compression, and compilation collaboration technologies introduced in the previous chapters are used. In order to deepen readers’ understanding, TensorRT, a model optimization tool designed specifically for NVIDIA chips, is introduced in detail.

carotenoids

Nicholas P. Jewell,Stephen C. Shiboskie the challenges of implementing embedded artificial intelligence? With these questions, we defined the topics to be studied in this book. After comparing the two implementation modes of embedded artificial intelligence: cloud computing mode and local mode, we clarified the necessity and technical c

男学院

Joan E. Sieber,James L. Sorensenf GPUs, TPUs, or ASICs and FPGAs designed for specific purposes. When needed, they will be integrated into embedded SoC chips. These chips adopt a parallel computing architecture and introduce concepts such as systolic arrays and multi-level caches to optimize data flow and minimize energy consumpti

tattle

Götz Lechner,Julia Göpel,Anna Passmanneural networks have small sizes and can operate within the constraints of low-power and memory-constrained environments while maintaining accuracy. Firstly, several strategies are introduced to reduce the computational complexity of neural networks without sacrificing accuracy. These strategies incl

organism

Problems of Political Theory and Actionhod to reduce the size of deep neural networks without changing the network structure. Assuming that the neural network model has been generated, techniques such as pruning, weight sharing, quantization, binary/ternary, Winograd convolution, etc. can be used to “compress” the neural network. Model d

Pamphlet

https://doi.org/10.1007/978-3-030-52500-2etworks in embedded devices can also be significantly improved through clever application-level optimizations. This chapter introduces the composition of this hierarchical cascade system, analyzes some key factors that can bring about efficiency improvements, and uses a case to demonstrate the cost

放纵

(Re)Configuring Actors in Practicetraditional deep learning, we clarify the goals and characteristics of lifelong deep learning and explore some methods to implement lifelong deep neural networks, such as dual learning systems, real-time updates, memory merging, and adaptation to real scenarios. Finally, the advantages brought by th