SensorAI: A Machine Learning Framework for Sensor Data.

Area of Science:

• Cyber-Physical Systems Engineering
• Machine Learning Applications
• Signal Processing

Background:

• The integration of machine learning (ML) and artificial intelligence (AI) into cyber-physical systems (CPS) necessitates specialized engineering knowledge.
• There is a growing need for engineers to understand appropriate ML models for time-series sensor data and effective signal processing techniques.
• The Center for Cyber-Physical Systems (CCPS) at the University of Georgia (UGA) identifies these skill gaps in student researchers.

Purpose of the Study:

• To develop a practical machine learning framework tailored for time-series sensor data analysis within CPS.
• To provide a tool for rapidly building, training, and evaluating multiple ML models.
• To serve as an educational resource for student researchers to grasp essential ML and signal processing concepts for CPS.

Main Methods:

• Development of a generalized machine learning framework.
• Application of the framework to time-series sensor data from CCPS testbeds.
• Facilitation of rapid model construction, training, and testing.

Main Results:

• Demonstration of a functional ML framework for time-series sensor data.
• Successful application of the framework on CCPS testbed data.
• Creation of a valuable tutorial tool for student researchers.

Conclusions:

• The developed ML framework effectively supports the analysis of time-series sensor data in CPS.
• The framework serves as an accessible educational tool, enhancing student researchers' understanding and skills.
• This initiative addresses the critical need for ML expertise in the field of cyber-physical systems.