Difference between revisions of "Bruna Stahlhofer at NOvA"

From AstroNuWiki
Jump to: navigation, search
(Created page with "{| cellspacing="10" || 200px || '''Working time''' : * 8 weeks July-August 2018 within [http://students.jinr.ru/en/jinr2018-participants SSP...")
 
 
Line 14: Line 14:
 
'''Tasks description'''
 
'''Tasks description'''
  
'''Title''':  
+
'''Title''': Investigating Alternative Deep Learning Methods in the NOvA Experiment
  
'''Abstract''':  
+
'''Abstract''': In this paper, we will review the basic concepts of convolutional neural networks (CNN) and how CNNs are used to classify events in the NOvA experiment. We will review neutrinos in the standard model, and how the detection of such particles is necessary for the better understanding of the interaction between fundamental particles. We have investigated methods of improving NOvA’s CNN classification. Our main focus has been the addition of event reconstruction variables to the network input. This has shown to improve categorization accuracy for small data samples, and a reduction in network over training in large data samples (with negligible improvement in accuracy).
  
Details of her work will be appeared soon here.
+
Details of her work in [[Media:SSP2018report_BrunaStahlhofer.pdf|report.pdf]].

Latest revision as of 23:37, 19 November 2018

BrunaStahlhofer.jpg

Working time :

  • 8 weeks July-August 2018 within SSP-2018

Advisors :

Tasks description

Title: Investigating Alternative Deep Learning Methods in the NOvA Experiment

Abstract: In this paper, we will review the basic concepts of convolutional neural networks (CNN) and how CNNs are used to classify events in the NOvA experiment. We will review neutrinos in the standard model, and how the detection of such particles is necessary for the better understanding of the interaction between fundamental particles. We have investigated methods of improving NOvA’s CNN classification. Our main focus has been the addition of event reconstruction variables to the network input. This has shown to improve categorization accuracy for small data samples, and a reduction in network over training in large data samples (with negligible improvement in accuracy).

Details of her work in report.pdf.