Page 2 - Demo

Utilizing machine learning to enable a generalizable in situ calibration method
• We present an in situ calibration of the C-2W electromagnetic NPA
• In situ calibrations are generally faster, easier, and much lower-cost than lengthy “benchtop” calibrations
• Machine learning is used to enhance the calibration method:
• Machine learning techniques maximize data leverage
• They also enable rapid evaluation of high-dimensionality parameter spaces for the best calibration
results with minimal computation time
• Test results on synthetic data are highly promising:
• Accurate while requiring minimal data
• Results on experimental data are well within theoretical expectations
• This methodology is highly generalizable:
• While the specific example presented is the EM-NPA, nothing about the method is unique to that diagnostic
Presenter: Gabriel Player
2