Reducing model bias in a deep learning classifier using domain adversarial neural networks in the MINERvA experiment
We present a simulation-based study using deep convolutional neural networks (DCNNs) to identify neutrino interaction vertices in the MINERvA passive targets region, and illustrate the application of domain adversarial neural networks (DANNs) in this context. DANNs are designed to be trained in one domain (simulated data) but tested in a second domain (physics data) and utilize unlabeled data from the second domain so that during training only features which are unable to discriminate between the domains are promoted. MINERvA is a neutrino-nucleus scattering experiment using the NuMI beamline at Fermilab. A-dependent cross sections are an important part of the physics program, and these measurements require vertex finding in complicated events. To illustrate the impact of the DANN we used a modified set of simulation in place of physics data during the training of the DANN and then used the label of the modified simulation during the evaluation of the DANN. We find that deep learning based methods offer significant advantages over our prior track-based reconstruction for the task of vertex finding, and that DANNs are able to improve the performance of deep networks by leveraging available unlabeled data and by mitigating network performance degradation rooted in biases in the physics models used for training.
- Research Organization:
- Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States). Oak Ridge Leadership Computing Facility (OLCF); Fermi National Accelerator Lab. (FNAL), Batavia, IL (United States)
- Sponsoring Organization:
- National Science Foundation (NSF); CNPq (Brazil); CoNaCyT (Mexico); IDI/IGI-UNI (Peru); Latin American Center for Physics (CLAF); Russian Ministry of Education and Science (Russia); National Science Centre of Poland; USDOE Office of Science (SC), High Energy Physics (HEP)
- Contributing Organization:
- MINERvA collaboration; MINERvA Collaboration
- Grant/Contract Number:
- AC05-00OR22725; AC02-07CH11359
- OSTI ID:
- 1484978
- Alternate ID(s):
- OSTI ID: 1487058
- Report Number(s):
- arXiv:1808.08332; FERMILAB-PUB-18-432-CD-ND
- Journal Information:
- Journal of Instrumentation, Vol. 13, Issue 11; ISSN 1748-0221
- Publisher:
- Institute of Physics (IOP)Copyright Statement
- Country of Publication:
- United States
- Language:
- English
Web of Science
The NuMI neutrino beam
|
journal | January 2016 |
Gradient-based learning applied to document recognition
|
journal | January 1998 |
ROOT — An object oriented data analysis framework
|
journal | April 1997 |
CNN Features Off-the-Shelf: An Astounding Baseline for Recognition
|
conference | June 2014 |
Deep Learning and Its Application to LHC Physics
|
journal | October 2018 |
A convolutional neural network neutrino event classifier
|
journal | September 2016 |
Learning representations by back-propagating errors
|
journal | October 1986 |
Vertex reconstruction of neutrino interactions using deep learning
|
conference | May 2017 |
Caffe: Convolutional Architecture for Fast Feature Embedding
|
conference | January 2014 |
Neutrino-Nucleus Interactions
|
journal | November 2011 |
Design, calibration, and performance of the MINERvA detector
|
journal | April 2014 |
The GENIE neutrino Monte Carlo generator
|
journal | February 2010 |
Measurement of Ratios of Charged-Current Cross Sections on C, Fe, and Pb to CH at Neutrino Energies 2–20 GeV
|
journal | June 2014 |
Measurement of partonic nuclear effects in deep-inelastic neutrino scattering using MINERvA
|
journal | April 2016 |
Geant4—a simulation toolkit
|
journal | July 2003 |
MINERvA neutrino detector response measured with test beam data
|
journal | July 2015 |
A New Approach to Linear Filtering and Prediction Problems
|
journal | March 1960 |
Neutrino flux predictions for the NuMI beam
|
journal | November 2016 |
Evolving Deep Networks Using HPC
|
conference | January 2017 |
Similar Records
Domain adaptation techniques for improved cross-domain study of galaxy mergers
Domain Adaptation for Measurements of Strong Gravitational Lenses