1 / 4

FlavourTagInputImportance processor

FlavourTagInputImportance processor. At request of CALICE Dave Bailey added code to the neural network package to calculate the relative importance of the net inputs for a given data set ; based on the same approach that is used in the TMVA package , manual p66:. implementation :

ricky
Download Presentation

FlavourTagInputImportance processor

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. FlavourTagInputImportance processor • At request of CALICE Dave Bailey added code to the neural network package • to calculate the relative importance of the net inputs for a given data set; • based on the same approach that is used in the TMVA package, manual p66: • implementation: • …/LCFIVertex/vertex_lcfi/nnet/src/InputImportance.cpp: • std::vector<double> InputImportance::operator()(const NeuralNet &theNet,const NeuralNetDataSet &theData) const • call of this code added to FlavourTag processor & values for all inputs printed to screen • for input sample

  2. Jet Charge – initial results Erik Devetak • Renormalization

  3. LCFIVertex input to ILD DST files • ILD concept is starting mass production of MC for LoI benchmark studies • based on “DST-files” with only the information needed for physics analyses • LCFIVertex will be run as part of the standard software chain • Following LCFIVertex results will be available in the DST-files: • flavour tag NN-outputs (b-, c-, bc-) • 4 MC truth values for flavour and charge • number of vertices in the jet • flavour tag inputs for the >=2 vertex case: • JointProbRPhi, JointProbZ, NumTracksInVertices, DecayLength, • DecayLengthSignficance, RawMomentum, PTCorrectedMass, SecondaryVertexProbability • This information will be added to an LCFloatVec object in the ReconstructedParticles that • Correspond to the jets in the event • NOTE: all output collections will be written to the “Full Files” for more detailed studies

  4. LCFIVertex code parameter tuning • The effort to optimise code parameters is making good progress (Roberval Walsh). • Detailed results shown in recent physics meetings; • Following track selection obtained for ZVRES: • chi2/ndf < 4; • d0 < 2mm; • d0 error < 0.007mm; • z0 (not ready!) • z0 error < 0.025mm (no change); • pT > 0.2 GeV; Idea: look for cut values that don’t degrade the reconstructed quantities, such as the secondary vertices, used as an input for the flavour tagging

More Related