Uncertainty Quantification and Propagation for ACORN, a geometric deep learning tracking pipeline for HEP experiments

TL;DR

This paper introduces an uncertainty quantification method for the ACORN particle tracking pipeline in high-energy physics, analyzing how uncertainties propagate and are affected by data properties, ultimately demonstrating high confidence in track reconstruction.

Contribution

It applies Monte Carlo Dropout to quantify uncertainties in the ACORN pipeline and studies their propagation and dependence on data size and properties, advancing uncertainty analysis in HEP tracking.

Findings

Uncertainty becomes dominated by aleatoric uncertainty with more data.

ACORN pipeline maintains high confidence and proper calibration.

Uncertainty propagation is influenced by data geometry and physical properties.

Abstract

We have developed an Uncertainty Quantification process for multistep pipelines and applied it to the ACORN particle tracking pipeline. All our experiments are made using the TrackML open dataset. Using the Monte Carlo Dropout method, we measure the data and model uncertainties of the pipeline steps, study how they propagate down the pipeline, and how they are impacted by the training dataset's size, the input data's geometry and physical properties. We will show that for our case study, as the training dataset grows, the overall uncertainty becomes dominated by aleatoric uncertainty, indicating that we had sufficient data to train the ACORN model we chose to its full potential. We show that the ACORN pipeline yields high confidence in the track reconstruction and does not suffer from the miscalibration of the GNN model.