ABSTRACT
This chapter aims to define and interpret phases of the AI Lifecycle for Edge AI applications. We highlight common pitfalls that can arise when developing and maintaining AI models at the edge and outline best practices that are recognized in academia and industry, with the goal of developing a well-established taxonomy and pipeline for the lifecycle of Edge AI. We lay out that edge-based AI is seen as a natural extension of the cloud-based AI paradigm, solving problems related to real-time responsiveness, privacy, and independent operation closer to the source of the data. The challenges of edge-use cases are summarized, including limited network access, limited computational resources, and the need for customised deployment and maintenance procedures.
