08.07.2026
Classification of muon tracks from charmonium decays and pion tracks in the model of the SPD detector using neural networks
The seventh issue of the international electronic scientific journal Natural Science Review has published an article by A. R. Didenko, I. V. Eletskikh, and A. O. Gridin, researchers at the Laboratory of Nuclear Problems, titled "Classification of muon tracks from charmonium decays and pion tracks in the model of the SPD detector using neural networks".
Researchers study problems of muon identification in the model of the SPD detector at the NICA collider using machine learning. The objective of identification of muons and pions in the momentum region (1.5–2.5 GeV/c) is complicated by the similarity of their signals in tracking systems.
The authors applied a deep neural network optimized by an evolutionary algorithm, that uses kinematic and track characteristics. The classifier achieves high accuracy while retaining 99% of muons and rejecting 48% of pions, and under tighter selection criteria, it suppresses up to 95.7% of the pion background, as demonstrated in the reconstruction of J/ψ → μ⁺μ⁻ decays.
Source: JINR
