Hashing and metric learning for charged particle tracking

by Sabrina Amrouche (Geneva University)

Wednesday 21 Oct 2020, 11:30 → 12:30 Europe/London

Zoom

At the High Luminosity Large Hadron Collider (HL-LHC), up to 200 proton-proton collisions happen during a single bunch crossing. This leads on average to tens of thousands of particles emerging from the interaction region. The CPU time of traditional approaches of constructing hit combinations will grow exponentially as the number of simultaneous collisions increases at the HL-LHC leading to major computational challenges. We introduce a similarity hashing and learning framework for track reconstruction [1] where multiple small regions of the detector, referred to as buckets, are reconstructed in parallel within the ATLAS simulation framework. New developments based on metric learning for hashing optimisation running on CPUs and GPUs will be highlighted with a discussion of the latest performance results on both TrackML [2] and ATLAS simulation datasets.

[1] Amrouche, Sabrina, et al. "Hashing and metric learning for charged particle tracking." https://ml4physicalsciences.github.io/2019/files/NeurIPS_ML4PS_2019_31.pdf
[2] Rousseau. D, et al. "The TrackML challenge." 2018. https://hal.inria.fr/hal-01745714/file/nips-2018-competition-6.pdf

Live

Seminar-Hashing and Metric Learning.pdf

Zoom