The H1 Neural Network Trigger Project

ADVANCED COMPUTING AND ANALYSIS TECHNIQUES IN PHYSICS RESEARCH(2001)

Cited 8|Views11
No score
Abstract
We present a short overview of neuromorphic hardware and some of the physics projects making use of such devices. As a concrete example we describe an innovative project within the H1-Experiment at the electron-proton collider HERA, instrumenting hardwired neural networks as pattern recognition machines to discriminate between wanted physics and uninteresting background at the trigger level. The decision time of the system is less than 20 microseconds, typical for a modem second level trigger. The neural trigger has been successfully running for the past four years and has turned out new physics results from H1 unobtainable so far with other triggering schemes. We describe the concepts and the technical realization of the neural network trigger system, present the most important physics results, and motivate an upgrade of the system for the future high luminosity running at HERA. The upgrade concentrates on "intelligent preprocessing" of the neural inputs which help to strongly improve the networks' discrimination power.
More
Translated text
Key words
data analysis,operating system,pattern recognition,neural network,fuzzy logic,new physics,data management,artificial intelligent
AI Read Science
Must-Reading Tree
Example
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined