Development of a method for fast, easy and optimized tuning of selection cuts
Modification of the Fisher Linear Discriminant Analysis for use in the low signal-to-noise environments
INFN sezione di Padova, Padova, Italy
* e-mail: firstname.lastname@example.org
Published online: 3 August 2006
In situations where the signal of the analysed particle is tangled up with orders of magnitude more background, its analysis may benefit from the use of a pattern classification method to discriminate the signal out of the background candidates. We present and explain the basic Linear Discriminant Analysis and the modifications brought – the use of cascaded cuts and of a locally optimized criterion – to adapt it to the conditions encountered in the field of heavy ion Physics. We show that this optimized multicut Linear Discriminant Analysis has a higher performance than classical selection cuts and provides a very fast and easy selection cut optimization.
© Springer-Verlag, 2006