Deep Learning Framework for Enhanced Neutrino Reconstruction of Single-line Events in the ANTARES Telescope
Abstract
We present the N-fit algorithm designed to improve the reconstruction of neutrino events detected by a single line of the ANTARES underwater telescope, usually associated with low energy neutrino events (∼ 100 GeV). N-Fit is a neural network model that relies on deep learning and combines several advanced techniques in machine learning --deep convolutional layers, mixture density output layers, and transfer learning. This framework divides the reconstruction process into two dedicated branches for each neutrino event topology --tracks and showers-- composed of sub-models for spatial estimation --direction and position-- and energy inference, which later on are combined for event classification. Regarding the direction of single-line events, the N-Fit algorithm significantly refines the estimation of the zenithal angle, and delivers reliable azimuthal angle predictions that were previously unattainable with traditional χ2-fit methods. Improving on energy estimation of single-line events is a tall order; N-Fit benefits from transfer learning to efficiently integrate key characteristics, such as the estimation of the closest distance from the event to the detector. N-Fit also takes advantage from transfer learning in event topology classification by freezing convolutional layers of the pretrained branches. Tests on Monte Carlo simulations and data demonstrate a significant reduction in mean and median absolute errors across all reconstructed parameters. The improvements achieved by N-Fit highlight its potential for advancing multimessenger astrophysics and enhancing our ability to probe fundamental physics beyond the Standard Model using single-line events from ANTARES data.
Related
- The road to H0, Ωm, and w with a global neutrino-telescope network
- Evaluation of the upgraded 3-inch Hamamatsu photomultiplier for the KM3NeT Neutrino Telescope
- Search for Neutrino-Produced Long-Lived Particles Using the SBND Cosmic Ray Tagger Beam Telescope
- First demonstration of 50 ps four-dimensional vertex reconstruction with the Timepix4 beam telescope
- Quantum-Enhanced Learning Framework for Intelligent and AI-Native 6G Wireless Networks
BibTeX
@article{albert2026deep,
title = {{Deep Learning Framework for Enhanced Neutrino Reconstruction of Single-line Events in the ANTARES Telescope}},
author = {A Albert and S Alves and M André and M Ardid and S Ardid and J.-J Aubert and J Aublin and B Baret and S Basa and Y Becherini and B Belhorma and F Benfenati and V Bertin and S Biagi and J Boumaaza and M Bouta and M.C Bouwhuis and H Brânzaş and R Bruijn and J Brunner and J Busto and B Caiffi and D Calvo and S Campion and A Capone and F Carenini and J Carr and V Carretero and T Cartraud and S Celli and L Cerisy and M Chabab and R. Cherkaoui El Moursli and T Chiarusi and M Circella and J.A.B Coelho and A Coleiro and R Coniglione and P Coyle and A Creusot and A.F Díaz and B de Martino and C Distefano and I Di Palma and C Donzaud and D Dornic and D Drouhin and T Eberl and A Eddymaoui and T van Eeden and D van Eijk and S. El Hedri and N. El Khayati and A Enzenhöfer and P Fermani and G Ferrara and F Filippini and L Fusco and S Gagliardini and J García-Méndez},
journal = {Machine Learning : Science and Technology},
year = {2026},
month = oct,
doi = {10.1088/2632-2153/ae5d84},
eprint = {2511.16614},
archivePrefix = {arXiv},
url = {https://hal.science/hal-05397585v2},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Deep Learning Framework for Enhanced Neutrino Reconstruction of Single-line Events in the ANTARES Telescope
AU - A Albert
AU - S Alves
AU - M André
AU - M Ardid
AU - S Ardid
AU - J.-J Aubert
AU - J Aublin
AU - B Baret
AU - S Basa
AU - Y Becherini
AU - B Belhorma
AU - F Benfenati
AU - V Bertin
AU - S Biagi
AU - J Boumaaza
AU - M Bouta
AU - M.C Bouwhuis
AU - H Brânzaş
AU - R Bruijn
AU - J Brunner
AU - J Busto
AU - B Caiffi
AU - D Calvo
AU - S Campion
AU - A Capone
AU - F Carenini
AU - J Carr
AU - V Carretero
AU - T Cartraud
AU - S Celli
AU - L Cerisy
AU - M Chabab
AU - R. Cherkaoui El Moursli
AU - T Chiarusi
AU - M Circella
AU - J.A.B Coelho
AU - A Coleiro
AU - R Coniglione
AU - P Coyle
AU - A Creusot
AU - A.F Díaz
AU - B de Martino
AU - C Distefano
AU - I Di Palma
AU - C Donzaud
AU - D Dornic
AU - D Drouhin
AU - T Eberl
AU - A Eddymaoui
AU - T van Eeden
AU - D van Eijk
AU - S. El Hedri
AU - N. El Khayati
AU - A Enzenhöfer
AU - P Fermani
AU - G Ferrara
AU - F Filippini
AU - L Fusco
AU - S Gagliardini
AU - J García-Méndez
PY - 2026
DA - 2026/10/12
JO - Machine Learning : Science and Technology
DO - 10.1088/2632-2153/ae5d84
UR - https://hal.science/hal-05397585v2
AB - We present the N-fit algorithm designed to improve the reconstruction of neutrino events detected by a single line of the ANTARES underwater telescope, usually associated with low energy neutrino events (∼ 100 GeV). N-Fit is a neural network model that relies on deep learning and combines several advanced techniques in machine learning --deep convolutional layers, mixture density output layers, and transfer learning. This framework divides the reconstruction process into two dedicated branches for each neutrino event topology --tracks and showers-- composed of sub-models for spatial estimation --direction and position-- and energy inference, which later on are combined for event classification. Regarding the direction of single-line events, the N-Fit algorithm significantly refines the estimation of the zenithal angle, and delivers reliable azimuthal angle predictions that were previously unattainable with traditional χ^2-fit methods. Improving on energy estimation of single-line events is a tall order; N-Fit benefits from transfer learning to efficiently integrate key characteristics, such as the estimation of the closest distance from the event to the detector. N-Fit also takes advantage from transfer learning in event topology classification by freezing convolutional layers of the pretrained branches. Tests on Monte Carlo simulations and data demonstrate a significant reduction in mean and median absolute errors across all reconstructed parameters. The improvements achieved by N-Fit highlight its potential for advancing multimessenger astrophysics and enhancing our ability to probe fundamental physics beyond the Standard Model using single-line events from ANTARES data.
SN - 2632-2153
SN - 2632-2153
ER -