mu2e Track Quality Selection in Python/Sklearn [Poster]
Scott Israel, A. Edmonds · n/a · 2022
Track Qua I ity • Mu2e signal and background can be hard to distinguish • Want to separate high/low quality measurements • Select high quality measurements to boost signal sensitivity • Used an ANN model in TMVA to do quality selection • Issues • Neural networks are hard to diagnose with hidden layers • TMVA is ROOT/C++ based, lacks accessibility in documentation/tutorials vs Python • Students have more experience with Python, Python has a larger ML community Example of the Decay in orbit (DIO} background burying the signal before quality selection, while the signal can be extracted without quality selection.f 10" w • See DocDB 41505