Speaker
Description
Summary
Advanced data analysis algorithms have a gradually increasing presence in High
Energy Physics. Neural Networks or Fisher Discriminant techniques are commonly used
in many experiments. Other techniques such as Support Vector Machine, Kernel
Estimation Technique or Evolutionary Algorithms have also been successfully tested
in this field.
Evolutionary Algorithms, inspired by the evolutionary theories from biology, are
based on the idea that solutions to a problem can be represented as entities that
evolve throughout generations as a consequence of interactions with other candidate
solutions, and the application of genetic operators. Genetic Algorithms (GA) and
Genetic Programming (GP) are the most known algorithms from this class. Genetic
Algorithms were applied mainly to problems such as discrimination and parameter
optimisation in both experimental and theoretical particle physics for the last ten
years [1]. Genetic Programming was only recently applied to event selection type
problems in two particle physics studies [2].
Gene Expression Programming (GEP), invented in 2001 [3], is a new technique of
Evolutionary Algorithms for data analysis. GEP uses fixed-length linear character
strings to represent solutions of a problem in a form of expression trees of
different shapes and sizes, and implements a genetic algorithm to find the best
solution. Subsequent studies [4] show that GEP combines the advantages of both GA
and GP, while overcoming some of their individual limitations.
A first application of GEP to High Energy Physics data analysis that I recently
presented [5] indicates this algorithm as a promising technique.
The present paper will present a detailed analysis of GEP applicability to High
Energy Physics event selection. This will include the description of the technique,
the results of its application to specific physics processes, the analysis of the
algorithm performances and their comparison with performances of the traditional
event selection methods. Based on this comparison, advantages and disadvantages of
GEP technique for High Energy Physics event selection will be discussed.
Bibliography
[1] see, for example, D.G. Ireland et. al., “A Genetic Algorithm Analysis of N*
Resonances in P(Gamma,K+)Lambda Reactions”, Nucl. Phys. A740 (2004)147; B.C.
Allanach et.al., Genetic Algorithms and Experimental Discrimination of SUSY Models”,
hep-ph/0406277; S. Abdullin, “Genetic Algorithm for SUSY Trigger Optimisation in CMS
Detector at LHC”, Nucl. Instr. Meth. A502 (2003) 693; Y. Azusa, “Genetic Algorithm
for SU(N) Gauge Theory on a Lattice, hep-lat/9808001.
[2] K. Cranmer, R. Sean Bowman, “PhysicsGP: A Genetic Programming Approach to Event
Selection”, physics/0402030; J.M. Link et. al. (FOCUS Coll.), Application of Genetic
Programming to High Energy Physics Event Selection”, hep-ex/0503007.
[3] C. Ferreira, “Gene Expression Programming: A New Adaptive Algorithm for Solving
Problems”, Complex Systems, Vol. 13, issue 2 (2001) 87.
[4] C. Zhou et. al., Evolving Accurate and Compact Classification Rules with Gene
Expression Programming”, IEEE Transactions on Evolutionary Computation, Vol. 7, No.
6 (2003) 519;
[5] L. Teodorescu, “High Energy Physics Data Analysis with Gene Expression
Programming”, IEEE NSS/MIC, October 2005, Puerto Rico.
