Artificial
Neural Networks
Artificial
Neural Networks
(ANN) have been widely used in High Energy Physics for more than 10
years in the following areas:
- Mainly for off-line
classification :
The main interest in High Energy
Physics is now feed forward back propagation networks, which are able
to "learn" sophisticated
cuts that physicists cannot
deduce by hand, or it is extremely time consuming to attempt to
discover them all.
I have extensively used ANN techniques for the data analysis of
the DONUT experiment :
Working with G. Tzanakos & M.
Zois, from the MINOS Athens Group, we have started to implement
the same idea for neutrino event classification for both the Far and
Near Detector data.
(Talk
presented at the September 2003 Collaboration Meeting)
Continuing the work on event classification using ANN we addressed some
of the issues we have discussed in the September 2003 presentation
related with :
-
Event Classification as a function of visible energy
- A priori probabilities and their significance in event
classification
- Enlarged set of input variables including tracking & shower
infomation .
The details of this study are
described in the next presentation that we gave at the Fermilab March
2004 Collaboration meeting.
(Talk
presented at the March 2004 Collaboration Meeting)
Created by Niki Saoulidou