Speaker
Brian Humensky
(University of Maryland)
Description
Boosted decision trees (BDTs) consistently provide enhanced background-rejection power for imaging air Cherenkov telescope (IACT) data analysis, compared to box cuts on conventional image-characterization parameters, including those based on second-order moment analysis, so-called “mean scaled width” and “mean scaled length.” In this presentation, we discuss in-progress work towards a new BDT-based classifier for VERITAS that incorporates both conventional parameters and goodness-of-fit parameters from a template-based shower reconstruction algorithm, including both challenges and successes.
| Type of Contribution | talk |
|---|
Primary author
Brian Humensky
(University of Maryland)
