Abstract:With the advent of the Internet of Everything era, there has been a dramatic increase in the number of edge devices, leading to the generation of massive amounts of data at the network edge. The development of artificial intelligence (AI) technology provides powerful support for analyzing and processing these data. However, the traditional centralized processing model of cloud computing fails to meet users' demands for low latency of tasks and low power consumption of devices. In addition, it poses potential threats to data privacy and security. At the same time, the development of embedded high-performance chips has greatly enhanced the computing capabilities of edge devices, enabling them to process computation-intensive tasks in real-time at the edge. In light of this, edge computing (EC) and AI are organically integrated, giving rise to a new computing paradigm known as edge intelligence (EI). This paper focuses on the frontiers and advances in EI and collaborative computing. Firstly, we introduce the relevant background, basic principles, and development trends of EC, AI, and EI. Secondly, we review EI methods for individual devices, covering edge training, edge inference, and edge caching. Thirdly, we present the collaborative EI works on multiple devices from the perspectives of architecture, technology, and functionality. Finally, we summarize the wide-ranging applications of EI in various fields, such as the industrial Internet of Things, smart cities, and virtual reality.