The 3rd Workshop on Machine Learning on Edge in Sensor Systems (SenSys-ML 2024)

Sensors have become more ubiquitous through the spread of the Internet of Things and increased market penetration of smartphones. The ensuing flood of data has the promise to advance state of the art in ways that could change the life of every human on the planet, including improvements in healthcare, environmental management, and city management. To enable this revolution machine learning needs to fill the gap to turn raw data into an understandable and actionable system. However, resource constraints, complex architectures, and challenging study designs and ground truth collection are some of the many hurdles that must be overcome to bring a promising idea to reality in this domain.

Sensys-ML focuses on providing extensive feedback from a diverse pool of people on Work In Progress papers involving machine learning on sensor systems (TinyML). The focus is on work that combines sensor signals from the physical world with machine learning, particularly in ways that are distributed to the device or use edge and fog computing. The development and deployment of ML at the very edge remains a technological challenge constrained by computing, memory, energy, network bandwidth and data privacy and security limitations. This is especially true for battery operated devices and always-on use cases and applications. Our end goal is to help all attendees design systems that are more scientifically advanced, robust to failure, efficient, and well validated.