Progress on Multivariate Data Analysis with <i>T</i>MVA

Friday 21 Jan 2011, 08:00 → 19:00 Europe/Zurich

60-6-015 (CERN)

Andreas Hoecker (CERN), Eckhard Von Toerne (Rheinische-Friedrich-Wilhelms- Univ.-Univ. Bonn, Physikalisches), Helge Voss (Physik-Institut), Jan Therhaag (Univ. Bonn, Physikalisches Institut), Joerg Stelzer (DESY), Peter Speckmayer (CERN)

Given the increasing complexity of modern HEP experiments, multivariate analysis techniques have proven indispensable to extract the maximum information content out of the data. At this workshop we will review recent developments on multivariate methods in TMVA (the Toolkit for Multivariate Data Analysis in ROOT), discuss user experience and feedback, and explain the use of TMVA by means of a hands-on tutorial. The workshop directly succeeds Phystat 2011 at CERN. EVO phone connection will be provided for remote participants. The schedule leaves room for user feedback on TMVA, and ideas for new multivariate approaches or improvements on existing methods. Please submit talk proposals to tmva-workshop@cern.ch If you wish to attend the workshop and/or TMVA tutorial please register as soon as possible, so that we can plan for an appropriate location at CERN.

PNG PDF PPTX

09:00 → 09:30 Introduction

Speaker: Andreas Hoecker (CERN)

Slides Introduction_hoecker_TMVAWorkshop2011.pdf Introduction_hoecker_TMVAWorkshop2011.pptx

09:30 → 10:00 Likelihood Classification and PDF Estimators

Speaker: Joerg Stelzer (DESY)

Slides Likelihoods.pdf Likelihoods.pptx

10:00 → 10:30 Neural Networks - Defeating the Black Box

Speaker: Jan Therhaag (Univ. Bonn, Physikalisches Institut)

Slides NeuralNetworks_TMVA_new.pptx NeuralNetworks_TMVA_noeffects.pdf

10:30 → 11:00 Coffee Break

11:00 → 11:30 Decision Trees and Boosting

Speaker: Helge Voss (Physik-Institut)

Slides DecisionTreesAndBoosting.pdf DecisionTreesAndBoosting.ppt

11:30 → 12:00 Metamethods and Data Preprocessing in TMVA

Speaker: Peter Speckmayer (CERN)

Slides 20110121_TMVA_Workshop.pdf

12:00 → 12:30 Practical Tips and Tricks for TMVA Users

Speaker: Eckhard Von Toerne (Rheinische-Friedrich-Wilhelms- Univ.-Univ. Bonn, Physikalisches)

Slides TMVATipsAndTricks.pdf

12:30 → 14:00 Lunch Break

14:00 → 14:30 Bayesian neural networks and their application in ATLAS analysis

Speaker: Mr Jiahang Zhong (Academia Sinica)

Slides TMVA_BNN.pdf

14:30 → 15:00 PanTau -- using TMVA for tau lepton ID

Speaker: Sebastian Fleischmann (U. Wuppertal)

Slides 110121-tmva-workshop-pantau4.pdf

15:00 → 15:30 Coffee Break

15:30 → 18:00 TMVA Tutorial

Participants are kindly requested to bring a notebook with ROOT, Version >=5.24 preinstalled. We recommend to use ROOT 5.28.

We will use TMVA 4.1.0 for this tutorial. This version can either be installed at the beginning of the tutorial or following the description here: https://twiki.cern.ch/twiki/bin/view/TMVA/WebHome

15:30 TMVA Tutorial

Participant are kindly requested to bring a notebook with ROOT, Version >=5.24. We will use TMVA 4.1.0 for this tutorial. This version can either be installed at the beginning of the tutorial or follwoing the description here: https://twiki.cern.ch/twiki/bin/view/TMVA/WebHome

Speaker: Eckhard Von Toerne (Rheinische-Friedrich-Wilhelms- Univ.-Univ. Bonn, Physikalisches)