Artificial Intelligence of Things: Implementation, Performance Analysis, Computational Feasibility

The main objective of this Thesis dissertation is to explore AI models in the context of resource-constrained IoT devices for a variety of applications. In particular, this dissertation aims to develop, optimize, and deploy ML/DL models intended for embedded systems, considering a number of trade-offs between accuracy, computational efficiency, and constraints during deployment that are critical for the real-world evaluations. Thus, investigating the balance between computational complexity and prediction accuracy in the collaboration between AI and tiny IoT devices will provide a deeper understanding of their potential and assist in selecting the most feasible and efficient approaches depending on the application.