Neutrino Fingerprints: Image-Based Encodings of IceCube Events for CNN Direction Reconstruction

Abstract

Reconstructing the direction of incoming neutrinos in the IceCube Neutrino Observatory is an important problem in astrophysics. The public IceCube--Neutrinos in Deep Ice Kaggle competition provided 140 million simulated events to benchmark reconstruction techniques. To address this challenge from a novel perspective we introduce neutrino fingerprints compact 72 × 72 × 3 images in which each pixel represents a single detector, with pulse timing and charge statistics encoded as color channels. This representation transforms sparse, irregular pulse data into dense images suitable for convolutional processing. Our ResNet18 model achieves a mean angular error of $1.10$ rad, indicating that convolutional networks trained on fingerprints rival more complex architectures while offering an effective, interpretable baseline for IceCube event reconstruction.

IceCube neutrino point-source searches in the direction of the KM3NeT ultra-high-energy event
R. Abbasi, M. Ackermann, J. Adams +416
arXiv10 Sep 2026PreprintCC BY 4.0Full textHigh-energy astrophysicsarXiv
Hidden Neutrino Active Galaxy Candidates: Prospects for Catalog and Stacking Searches with IceCube, IceCube-Gen2, and KM3NeT
Mainak Mukhopadhyay, Jose Alonso Carpio, Heer Patel +2
arXiv1 Nov 2026PreprintCC BY 4.0Full textGalaxies High-energy astrophysicsarXiv
Resolving the Galactic Center Region's Gamma-Ray Emission and Searching for its Neutrino Counterpart with HAWC and IceCube
S. Yun-Carcamo, R. Babu, for the HAWC +1
arXiv18 Sep 2026PreprintCC BY 4.0Full textHigh-energy astrophysicsarXiv
Limit on high energy neutrino emission from Abell 119 using IceCube 10-year muon track data
Sri Devaki Meduri, Shantanu Desai
arXiv5 Oct 2026PreprintCC BY 4.0Full textCosmology High-energy astrophysicsarXiv
Multiple-Mediator-Affected Ultra-High-Energy Neutrino Attenuation: Hints for 5D U(1)_{L_μ− L_τ} in IceCube
Atri Bhattacharya, Ayushi Kaushik, Kenji Nishiwaki
arXiv29 Sep 2026PreprintarXiv licenceFull textHigh-energy astrophysics High-energy & particle physicsarXiv

Cite

BibTeX RIS

BibTeX

@misc{tori2026neutrino,
  title = {{Neutrino Fingerprints: Image-Based Encodings of IceCube Events for CNN Direction Reconstruction}},
  author = {Floriano Tori and Brecht F. Verbeken and Vincent Ginis},
  howpublished = {arXiv},
  year = {2026},
  month = jun,
  doi = {10.48550/arxiv.2606.02788},
  eprint = {2606.02788},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.02788},
  note = {CC BY 4.0},
}

RIS

TY  - UNPB
TI  - Neutrino Fingerprints: Image-Based Encodings of IceCube Events for CNN Direction Reconstruction
AU  - Floriano Tori
AU  - Brecht F. Verbeken
AU  - Vincent Ginis
PY  - 2026
DA  - 2026/06/01
PB  - arXiv
DO  - 10.48550/arxiv.2606.02788
UR  - https://arxiv.org/abs/2606.02788
AB  - Reconstructing the direction of incoming neutrinos in the IceCube Neutrino Observatory is an important problem in astrophysics. The public IceCube--Neutrinos in Deep Ice Kaggle competition provided 140 million simulated events to benchmark reconstruction techniques. To address this challenge from a novel perspective we introduce neutrino fingerprints compact 72 × 72 × 3 images in which each pixel represents a single detector, with pulse timing and charge statistics encoded as color channels. This representation transforms sparse, irregular pulse data into dense images suitable for convolutional processing. Our ResNet18 model achieves a mean angular error of $1.10$ rad, indicating that convolutional networks trained on fingerprints rival more complex architectures while offering an effective, interpretable baseline for IceCube event reconstruction.
ER  -

Abstract

Related

Cite