Draft: Introduce two new classifiers to reduce ghost tracks in SciFi and Downstream Tracking
This merge request is part of the ongoing effort to enhance the HLT2 tracking algorithm for reconstructing long-lived particles using machine learning techniques.
Two new ML-based classifiers have been added to Pr/PrAlgorithms, along with their associated trained weight files, targeting improvements in: SciFi Seed tracking and Downstream tracking
These classifiers do not replace existing logic or workflows. Instead, they provide an additional layer aimed at reducing ghost tracks.
Preliminary evaluations show a notable reduction in ghost rate with minimal impact on processing time.
This is a reworked merge request, with most development completed. It may still require technical review and validation for the final adjustments.
Edited by Sabin Hashmi Kalavan Kadavath