Machine Learning Sensors: A design paradigm for the future of intelligent sensors.

In the last decade, there has been a significant increase in the use of machine learning (ML) for commercial purposes. At the same time, advancements in wireless communications have led to the widespread adoption of cloud-connected devices, such as Internet of Things "smart devices." These devices, while appearing intelligent, mostly rely on centralized cloud infrastructure, raising concerns about data storage, usage, and access. This has led to the need for enhanced transparency and the implementation of rules or systems to safeguard user privacy and apprise users about the data their devices are gathering. As a solution, the authors present the concept of the ML sensor, which offers a structured framework for creating embedded systems equipped with machine learning features with a strong emphasis on privacy. By limiting the data interface, the ML sensor approach guarantees that user data cannot be accessed beyond the sensor's intended purpose.