Combined neural network - QCD classifier for quark and gluon jet separation

I. Csabai, F. Czakó, Z. Fodor

Research output: Contribution to journalArticlepeer-review

9 Scopus citations


In this paper we show a possible quark/gluon jet discrimination method using a neural network, combined with QCD, in electron-positron annihilation. The network has been trained by quark and gluon jets of the same energy, therefore only the internal structure of the jets has been considered. The input data correspond to a total basis, in the sense that all the longitudinal and transverse momenta of the outgoing hadrons with respect to the jet axis (excluding soft particles) have been used in the analysis. The neural network alone has reached ≈ 71% identification accuracy for jets of the same energy. On the other hand the QCD matrix element for definite jet energies also gives the probability of a jet being a gluon or quark one. By the combination of the neural network, and the QCD second-order matrix element, we have reached 92% accuracy in identifying jets.

Original languageEnglish (US)
Pages (from-to)288-308
Number of pages21
JournalNuclear Physics, Section B
Issue number2
StatePublished - Apr 27 1992

All Science Journal Classification (ASJC) codes

  • Nuclear and High Energy Physics


Dive into the research topics of 'Combined neural network - QCD classifier for quark and gluon jet separation'. Together they form a unique fingerprint.

Cite this