astro-ph.IM · 2606.02788 Copy arXiv ID · Jun 1, 2026 Save Neutrino Fingerprints: Image-Based Encodings of IceCube Events for CNN Direction Reconstruction Authors: Floriano Tori , Brecht Verbeken , Vincent Ginis
Organizations: Data Analytics Lab, Vrije Universiteit Brussel, Pleinlaan 5, 1050 Brussels, Belgium · imec-SMIT, Vrije Universiteit Brussel, Pleinlaan 9, 1050 Brussels, Belgium · School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138, USA
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 72 \times 72 \times 3 72 × 72 × 3 1.10 1.10 1.10
Explore similar work Apr 21, 2026 · R. Abbasi, M. Ackermann, J. Adams +411 Modern Collider Experiments Sphere
May 11, 2026 · A. Matseiko, G. Plotnikov, I. Kharuk Modern Collider Experiments
Nov 20, 2025 · Asya Y. Akkus, Bradley T. Wolfe, Pinghan Chu +7 Image Reconstruction Inertial Confinement Fusion
Apr 21, 2026 · hep-ex J/K move · Enter open · S save
R. Abbasi, M. Ackermann, J. Adams, J. A. Aguilar +410
IceCube is a cubic-kilometer-scale neutrino detector located at the geographic South Pole. A precise directional reconstruction of IceCube neutrinos is vital for associations with astronomical objects. In this context, we discuss neural posterior estimation of the neutrino direction via a transformer encoder that maps to a normalizing flow on the 2-sphere. It achieves a new state-of-the-art angular resolution for the two main event morphologies in IceCube - tracks and showers - while being significantly faster than traditional B-spline-based likelihood reconstructions. All-sky scans can be performed within seconds rather than hours, and take constant computation time, regardless of whether the posterior extent is arc-minutes or spans the whole sky. We utilize a combination of
C 2 C^2 C 2 -smooth rational-quadratic splines, scale transformations and rotations to define a novel spherical normalizing-flow distribution whose parameters are predicted as a whole as the output of the transformer encoder. We test several structural choices diverting from the vanilla transformer architecture. In particular, we find dual residual streams, nonlinear QKV projection and a separate class token with its own cross-attention processing to boost test-time performance. The angular resolution for both showers and tracks improves substantially over the whole trained energy range from 100 GeV to 100 PeV. At 100 TeV deposited energy, for example, the median angular resolution improves by a factor of
1.3 1.3 1.3 for throughgoing tracks, by a factor of
1.7 1.7 1.7 for showers and by a factor of
2.5 2.5 2.5 for starting tracks compared to state-of-the art likelihood reconstructions based on B-splines. While previous machine-learning (ML) efforts have managed to obtain competitive shower resolutions, this is the first time an ML-based method outperforms likelihood-based muon reconstructions above 100 GeV.