On this web page we present the combination of opposite side taggers (muon, electron, jet charge and kaon tagger) into one single tagger with a neural network.

An increase of the tagger performance from &epsilon D² = 1.5 % to

&epsilon D² = 1.8 %

The following figures are available for download:

• Figure 1 (gif): Network input variables: Correlations between the network input variables and to the training target
• Figure 2 (gif): Network training: Raw purity as a function of network output
• Figure 3 (gif): Calibration: Measured dilution as function of predicted dilution with fit of scale factor S in l+SVT data
• Figure 4 (gif): Comparison: Predicted dilution times the tagging decision for the combined tagger versus the hierarchical tagger
• Figure 5 (gif): Improvement: Change in D² from hierarchical to combined tagger versus the predicted dilution of the jet charge tagger in events with lepton and jet charge tag