BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:PDE-FOAM - a probability-density estimation method based on self-a
 dapting phase-space binning
DTSTART;VALUE=DATE-TIME:20081103T135000Z
DTEND;VALUE=DATE-TIME:20081103T141500Z
DTSTAMP;VALUE=DATE-TIME:20130519T171800Z
UID:indico-contribution-14@cern.ch
DESCRIPTION:Speakers: Dr. DANNHEIM\, Dominik (CERN)\nProbability-Density E
 stimation (PDE) is a multivariate discrimination technique based on sampli
 ng signal and background densities in a multi-dimensional phase space. The
  signal and background densities are defined by event samples (from data o
 r monte carlo) and are evaluated using a binary search tree (range searchi
 ng). This method is a powerful classification tool for problems with highl
 y non-linearly correlated observables. In this paper\, we present an innov
 ative improvement of the PDE method that uses a self-adapting binning meth
 od to divide the multi-dimensional phase space in a finite number of hyper
 -rectangles (boxes). For a given number of boxes\, the binning algorithm a
 djusts the size and position of the boxes inside the multidimensional phas
 e space\, minimizing the variance of the signal and background densities i
 nside the boxes. The binned density information is stored in binary trees\
 , allowing for a very fast and memory-efficient classification of events. 
 The implementation of the binning algorithm (PDE-FOAM) is based on the mon
 te-carlo integration package TFOAM included in the analysis package ROOT a
 nd has been developed within the framework of the Toolkit for Multivariate
  Data Analysis with ROOT (TMVA). We present performance results for repres
 entative examples (toy models) and discuss the dependence of the obtained 
 results on the choice of parameters. The new PDE-FOAM is compared to the o
 riginal PDE method based on range-searching.\n\nhttp://indico.cern.ch/cont
 ributionDisplay.py?contribId=14&sessionId=5&confId=34666
LOCATION:Ettore Majorana Foundation and Centre for Scientific Culture
URL:http://indico.cern.ch/contributionDisplay.py?contribId=14&sessionId=5&
 confId=34666
END:VEVENT
END:VCALENDAR