Building Generalizable Deep Learning Solutions for AIoT Applications
Huatao Xu · 2025
In an era where embedded and mobile devices are becoming ubiquitous, the intersection of Artificial Intelligence and the Internet of Things (AIoT) is rapidly transforming various fields. This paper delves into the challenges and innovations in this domain, particularly focusing on the development of generalized sensing models for wearable sensor data. We introduce novel approaches to leverage the abundant unlabeled sensor data, physical sensing knowledge, and common knowledge embedded in Large Language Models (LLMs) to enhance the generalizability of AIoT models significantly.