Analysis Meeting 31Oct 05T. Burnett1 Classification, etc. status at UW Implement “nested trees” Generate “boosted” trees for all of the Atwood categories.

Slides:

Advertisements

Similar presentations

Advanced XSLT. Branching in XSLT XSLT is functional programming –The program evaluates a function –The function transforms one structure into another.

Advertisements

Heuristic Search techniques

SAS 9.2 Getting Started. 2006/03/01. SAS Main Window.

Simple but slow: O(n 2 ) algorithms Serial Algorithm Algorithm compares each particle with every other particle and checks for the interaction radius Most.

Introduction Training Complexity, Pruning CART vs. ID3 vs. C4.5

1 Progress report on Calorimeter design comparison simulations MICE detector phone conference Rikard Sandström.

Simulation Studies of a (DEPFET) Vertex Detector for SuperBelle Ariane Frey, Max-Planck-Institut für Physik München Contents: Software framework Simulation.

Chapter 7 – Classification and Regression Trees

Chapter 7 – Classification and Regression Trees

Searching for Single Top Using Decision Trees G. Watts (UW) For the DØ Collaboration 5/13/2005 – APSNW Particles I.

7 June 06 1 UW Point Source Detection and Localization: Compare with DC2 truth Toby Burnett University of Washington.

Assessment and Proposal W. B. Atwood UCSC, April 6, 2005.

Science Tools 27 Sep 06 1 Generating post-DC2 IRFs: a proposal Toby Burnett.

My Beats Karen Tipping Dylan Barrett Mike Smith. Overview Goal: To dynamically create tracks for guitar hero using any song as input. Terminology for.

Analysis Meeting 24Oct 05T. Burnett1 UW classification: new background rejection trees.

Analysis Meeting 26 Sept 05T. Burnett1 UW classification: background rejection (A summary – details at a future meeting)

Bill Atwood, July, 2003 GLAST 1 A GLAST Analysis Agenda Overarching Approach & Strategy Flattening Analysis Variables Classification Tree Primer Sorting.

Bill Atwood, SCIPP/UCSC, Oct, 2005 GLAST 1 DC2 Discussion – What Next? 1)Alternatives - Bill's IM analysis - ???? 2)Backing the IM Analysis into Gleam.

T. Burnett: IRF status 30-Jan-061 DC2 IRF Status Meeting agenda, references at:

Introduction to Single-Top Single-Top Cross Section Measurements at ATLAS Patrick Ryan (Michigan State University) The measurement.

Bill Atwood, SCIPP/UCSC, Dec., 2003 GLAST 1 DC1 Instrument Response Agenda Where are we? V3R3P7 Classification Trees Covariance Scaled PSF Pair Energies.

Analysis Meeting 31Oct 05T. Burnett1 Classification, etc. status Implement pre-filters Energy estimation.

Analysis Meeting 31Oct 05T. Burnett1 Classification, etc. status at UW Background rejection for the v7r2 “500M” run at SLAC”: first look at v7r3.

Bill Atwood, SCIPP/UCSC, Nov, 2003 GLAST 1 Update on Rome Results Agenda A eff Fall-off past 10 GeV V3R3P7 Classification Trees Covariance Scaled PSFs.

C&A 8 May 06 1 Point Source Localization: Optimizing the CTBCORE cut Toby Burnett University of Washington.

Analysis Meeting 17 Oct 05T. Burnett1 UW classification: background rejection Update with new all gamma and background samples.

THB Source Catalog Meeting 12/21/05 1 Healpixels and Wavelets Toby Burnett and Bruce Lesnick University of Washington.

Bill Atwood, Nov. 2002GLAST 1 Classification PSF Analysis A New Analysis Tool: Insightful Miner Classification Trees From Cuts Classification Trees: Recasting.

Analysis Meeting 4/09/05 - T. Burnett 1 Classification tree Update Toby Burnett Frank Golf.

1 Using Insightful Miner Trees for Glast Analysis Toby Burnett Analysis Meeting 2 June 2003.

Bill Atwood, SCIPP/UCSC, DC2 Closeout, May 31, 2006 GLAST 1 Bill Atwood & Friends.

T. Burnett: IRF status 5-feb-061 DC2 IRF Status Meeting agenda, references at:

Analysis Meeting 14Nov 05T. Burnett1 Classification, etc. status at UW New application to apply trees to existing tuple Background rejection for the v7r2.

Classification and Regression Trees for Glast Analysis: to IM or not to IM? Toby Burnett Data Challenge Meeting 15 June 2003.

Analysis Meeting 8 Aug 05T. Burnett1 Status of UW classification and PSF fits Data selection Classification PSF.

Bill Atwood, SCIPP/UCSC, Oct, 2005 GLAST 1 Back Ground Rejection Status for DC2 a la Ins. Miner Analysis.

Bill Atwood, SCIPP/UCSC, Jan., 2006 GLAST 1 The 3 rd Pass Back Rejection Analysis using V7R3P4 (repo) 1)Training sample: Bkg: Runs (SLAC) v7r3p4.

T. Burnett: PSF status 3Feb061 DC2 PSF Status The function: 8 bins in energy (2/decade), 6 in cos(  ) (0.1) Two classes A: GoodEvent3; C: GoodEvent1 &&

Data Mining and Decision Tree CS157B Spring 2006 Masumi Shimoda.

Validation and TestEm series Michel Maire for the Standard EM group LAPP (Annecy) July 2006.

Chapter 9 – Classification and Regression Trees

LOGO Ensemble Learning Lecturer: Dr. Bo Yuan

DC2 at GSFC - 28 Jun 05 - T. Burnett 1 DC2 C++ decision trees Toby Burnett Frank Golf Quick review of classification (or decision) trees Training and testing.

PPR meeting Marcello Lunardon 1 Semi-electronic beauty detection: status and perspectives THE COLLABORATION Rosario Turrisi and Marcello Lunardon.

Ensembles. Ensemble Methods l Construct a set of classifiers from training data l Predict class label of previously unseen records by aggregating predictions.

David Adams ATLAS DIAL status David Adams BNL November 21, 2002 ATLAS software meeting GRID session.

Measurement of the Atmospheric Muon Neutrino Energy Spectrum with IceCube in the 79- and 86-String Configuration Tim Ruhe, Mathis Börner, Florian Scheriau,

Preliminary analysis of p-Pb data update n. 6 Lorenzo Bonechi LHCf Catania meeting – 19 December 2013.

T. Burnett1 GLAST LAT ProjectDOE/NASA Baseline-Preliminary Design Review, January 9, 2002 SAS Software: Sources Detector geometry model Simulation Event.

Training of Boosted DecisionTrees Helge Voss (MPI–K, Heidelberg) MVA Workshop, CERN, July 10, 2009.

MACHINE LEARNING 10 Decision Trees. Motivation  Parametric Estimation  Assume model for class probability or regression  Estimate parameters from all.

Feb. 7, 2007First GLAST symposium1 Measuring the PSF and the energy resolution with the GLAST-LAT Calibration Unit Ph. Bruel on behalf of the beam test.

David Adams ATLAS DIAL: Distributed Interactive Analysis of Large datasets David Adams BNL August 5, 2002 BNL OMEGA talk.

Beam Extrapolation Fit Peter Litchfield  An update on the method I described at the September meeting  Objective;  To fit all data, nc and cc combined,

Mark Dorman UCL/RAL MINOS Collaboration Meeting Fermilab, Oct. 05 Data/MC Comparisons and Estimating the ND Flux with QE Events ● Update on QE event selection.

GLAST/LAT analysis group 31Jan05 - T. Burnett 1 Creating Decision Trees for GLAST analysis: A new C++-based procedure Toby Burnett Frank Golf.

Copyright © Curt Hill Other Trees Applications of the Tree Structure.

Σ Eγ comparison using Woochun Park Univ of South Carolina Feb 26,

Feb. 3, 2007IFC meeting1 Beam test report Ph. Bruel on behalf of the beam test working group Gamma-ray Large Area Space Telescope.

T. Burnett: Anagroup 5/24/04 1 Latest PSF and Point-source sensitivity prediction status Anagroup meeting Toby Burnett 24 May 04.

This was written with the assumption that workbooks would be added. Even if these are not introduced until later, the same basic ideas apply Hopefully.

Marcello Barisonzi First results of AOD analysis on t-channel Single Top DAD 30/05/2005 Page 1 AOD on Single Top Marcello Barisonzi.

Background Rejection Activities in Italy

Tree based validation tool for track reconstruction

Event Classification & Background Rejection

LAT Software Quality Assurance

A First DC2 Analysis with New Recons

Using decision trees and their ensembles for analysis of NIR spectroscopic data WSC-11, Saint Petersburg, 2018 In the light of morning session on superresolution.

LAT performance for DC2 and the “4-panel plot”

Statistical Learning Dong Liu Dept. EEIS, USTC.

Presentation transcript:

Analysis Meeting 31Oct 05T. Burnett1 Classification, etc. status at UW Implement “nested trees” Generate “boosted” trees for all of the Atwood categories Background rejection status

Analysis Meeting 31Oct 05T. Burnett2 “Nested” Trees Objectives: –define a CT model that is consistent with current sources Insightful Miner (IM) (via Bill and Tracy) Using the classify package (only TB so far) Riccardo Rando’s “forests” –Generalize (and minimize) the “glue” code: it must Provide connection between the named variables and values found in the ROOT tree Decide which trees to evaluate Create and fill output tuple variables The role of the filter (see last week) –Preselection cuts for a train/evaluation to simplify the actual tree –Implemented as a simple tree with one path. –New idea: use it to select the sample to train/evaluate as well Riccardo splits up the data into 8 energy bins Bill (and TB) split up the parametric energy into 3 bins New feature: evaluate multiple trees, assumed to have exclusive filters, return the first giving a non-zero value.

Analysis Meeting 31Oct 05T. Burnett3 Apply to the energy CT The param estimator is unique in that it covers all energies. (See my plot last week, and following) With this feature, since the number of trees is now data-driven, it is easy to replace a single tree with multiple trees. Implement simply by including a file in the param folder, named nested.txt –lines are relative paths to new CT folders # the nested trees for energy-param low med high

Analysis Meeting 31Oct 05T. Burnett4 UW Energy resolution: (param only, three trees)

Analysis Meeting 31Oct 05T. Burnett5 How about XML? The current CT descriptions use the file system to define the structure. Tracy is a big advocate of XML (and so am I!) It would be easy to convert this to a single XML file that reproduces the same structure. Only needs a simple tool to expand/contract

Analysis Meeting 31Oct 05T. Burnett6 UW background rejection status Reminder: –We currently mimic the UCSC structure, and only this is supported in Gleam. (Supporting Padova would require that UCSC adopt filters that also select the subset.) 8 different branches: –Vertex or Track better for PSF? highcal: High energy (>3.5 GeV) medcal: Medium energy (>350 MeV) thick: Low energy, thick converter section thin: Low energy, thin section Each event is examined, and the appropriate CT evaluated. –CTgamma is set to the value from that node. –CTgammaType is set to 0-7.

Analysis Meeting 31Oct 05T. Burnett7 UW background rejection Acceptance for the given CTgamma and CTgammaType cut (Note that the on-axis effective area is very close to dividing by 0.8  : The “SRD” minimum is 2 m 2 sr) Spectra for 8640 sec (0.1 day): expected extra-galactic diffuse (EGRET “standard”) and measured background)