Advanced Northern Tracks Selection using a Graph Convolutional Neural Network for the IceCube Neutrino Observatory

TL;DR

This paper presents a novel graph convolutional neural network approach for the IceCube detector to improve the identification of neutrino-induced muons by leveraging sensor data and detector geometry.

Contribution

The study introduces an advanced GCN-based method for muon track selection, enhancing background rejection over previous techniques.

Findings

Improved background rejection efficiency in muon track selection.

Effective integration of sensor data and detector geometry using GCN.

Demonstrated superiority over existing methods.

Abstract

The IceCube Neutrino Observatory is a cubic-kilometer detector located in the Antarctic ice at the geographic South Pole. It reads out over 5,000 photomultiplier tubes (PMTs) to detect Cherenkov light produced by secondary particles, enabling IceCube to identify both atmospheric and astrophysical neutrinos. One of the main challenges in this effort is effectively distinguishing between muons induced by neutrinos and those generated by cosmic-ray air showers. To address this challenge, the Advanced Northern Tracks Selection (ANTS) employs a graph convolutional neural network. This network is designed to utilize both the sensor data and the geometric arrangement of the detector's photomultiplier tubes (PMTs). By representing each module as a node in a graph and extracting features from each module, the network can capture and integrate both local and global features. This work details the implementation of the network architecture and highlights the improvements in background rejection efficiency compared to existing methods for selecting muon tracks.