… for training AI models on edges. In addition, we provide an overview of edge AI system
architectures for communication-efficient edge training and edge inference. In the next section, …

… required with distributed AI limits its ability to be as proficient at the IoT edge. In …
communication-efficient processing techniques for training AI in resource-constrained devices at the …

… to develop technologies for deploying AI algorithms at the network edge, called edge AI [1,
2]. Edge learning refers to the first phase of edge AI, training of an AI model, while the other …

… communication-efficient data shuffling strategy in the wireless distributed computing system
for on-device distributed AI … presented the communication-efficient designs for edge inference…

… This article proposes communication-efficient FedAvg (CE-… 1) We propose the CE-FedAvg
algorithm, which is composed of … 4) We perform further experimentation on an edgecomputing-…

… structures of edge AI models and algorithms. The system performance metrics for edge AI
are … Specifically, to design a communication-efficient edge AI training system, we will provide …

… edge systems, for optimizing mobile edge computing, caching and communication. And thus,
we design the “In-Edge AI” … collaboration among devices and edge nodes to exchange the …

… The mainstream edge learning approach, federated learning, has been developed based …
at edge devices and then transmitted to an edge server for updating a global AI model. Since …

… the distributed devices in edge networks, conventional centralized AI algorithms can lead to
poor … Distributed AI algorithms have shown great potential to be applied in edge networks to …

… (AI), which can be used in IIoT for data analysis and mining. The conventional cloud-based
architectures [2] transmit the user data to a cloud server, and executing AI algorithms on the …