اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدBackground rejection in NEXT using deep neural networks
96
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
We investigate the potential of using deep learning techniques to reject background events in searches for neutrinoless double beta decay with high pressure xenon time projection chambers capable of detailed track reconstruction. The differences in the topological signatures of background and signal events can be learned by deep neural networks via training over many thousands of events. These networks can then be used to classify further events as signal or background, providing an additional background rejection factor at an acceptable loss of efficiency. The networks trained in this study performed better than previous methods developed based on the use of the same topological signatures by a factor of 1.2 to 1.6, and there is potential for further improvement.
قيم البحث
اقرأ أيضاً
Convolutional neural networks (CNNs) are widely used state-of-the-art computer vision tools that are becoming increasingly popular in high energy physics. In this paper, we attempt to understand the potential of CNNs for event classification in the N
EXT experiment, which will search for neutrinoless double-beta decay in $^{136}$Xe. To do so, we demonstrate the usage of CNNs for the identification of electron-positron pair production events, which exhibit a topology similar to that of a neutrinoless double-beta decay event. These events were produced in the NEXT-White high-pressure xenon TPC using 2.6-MeV gamma rays from a $^{228}$Th calibration source. We train a network on Monte Carlo-simulated events and show that, by applying on-the-fly data augmentation, the network can be made robust against differences between simulation and data. The use of CNNs offer significant improvement in signal efficiency/background rejection when compared to previous non-CNN-based analyses.
We demonstrate that the application of an external magnetic field could lead to an improved background rejection in neutrinoless double-beta (0nbb) decay experiments using a high pressure xenon (HPXe) TPC. HPXe chambers are capable of imaging electro
n tracks, a feature that enhances the separation between signal events (the two electrons emitted in the 0nbb decay of 136Xe) and background events, arising chiefly from single electrons of kinetic energy compatible with the end-point of the 0nbb decay (Qbb ). Applying an external magnetic field of sufficiently high intensity (in the range of 0.5-1 Tesla for operating pressures in the range of 5-15 atmospheres) causes the electrons to produce helical tracks. Assuming the tracks can be properly reconstructed, the sign (direction) of curvature can be determined at several points along these tracks, and such information can be used to separate signal (0nbb) events containing two electrons producing a track with two different directions of curvature from background (single-electron) events producing a track that should spiral in a single direction. Due to electron multiple scattering, this strategy is not perfectly efficient on an event-by-event basis, but a statistical estimator can be constructed which can be used to reject background events by one order of magnitude at a moderate cost (approx. 30%) in signal efficiency. Combining this estimator with the excellent energy resolution and topological signature identification characteristic of the HPXe TPC, it is possible to reach a background rate of less than one count per ton-year of exposure. Such a low background rate is an essential feature of the next generation of 0nbb experiments, aiming to fully explore the inverse hierarchy of neutrino masses.
Future upgrades to the LHC will pose considerable challenges for traditional particle track reconstruction methods. We investigate how artificial Neural Networks and Deep Learning could be used to complement existing algorithms to increase performanc
e. Generating seeds of detector hits is an important phase during the beginning of track reconstruction and improving the current heuristics of seed generation seems like a feasible task. We find that given sufficient training data, a comparatively compact, standard feed-forward neural network can be trained to classify seeds with great accuracy and at high speeds. Thanks to immense parallelization benefits, it might even be worthwhile to completely replace the seed generation process with the Neural Network instead of just improving the seed quality of existing generators.
For a large liquid argon time projection chamber (LArTPC) operating on or near the Earths surface to detect neutrino interactions, the rejection of cosmogenic background is a critical and challenging task because of the large cosmic ray flux and the
long drift time of the TPC. We introduce a superior cosmic background rejection procedure based on the Wire-Cell three-dimensional (3D) event reconstruction for LArTPCs. From an initial 1:20,000 neutrino to cosmic-ray background ratio, we demonstrate these tools on data from the MicroBooNE experiment and create a high performance generic neutrino event selection with a cosmic contamination of 14.9% (9.7%) for a visible energy region greater than O(200)~MeV. The neutrino interaction selection efficiency is 80.4% and 87.6% for inclusive $ u_mu$ charged-current and $ u_e$ charged-current interactions, respectively. This significantly improved performance compared to existing reconstruction algorithms, marks a major milestone toward reaching the scientific goals of LArTPC neutrino oscillation experiments operating near the Earths surface.
The Dark Matter Time Projection Chamber (DMTPC) collaboration is developing a low pressure gas TPC for detecting Weakly Interacting Massive Particle (WIMP)-nucleon interactions. Optical readout with CCD cameras allows for the detection of the daily m
odulation of the direction of the dark matter wind. In order to reach sensitivities required for WIMP detection, the detector needs to minimize backgrounds from electron recoils. This paper demonstrates that a simplified CCD analysis achieves $7.3times10^{-5}$ rejection of electron recoils while a charge analysis yields an electron rejection factor of $3.3times10^{-4}$ for events with $^{241}$Am-equivalent ionization energy loss between 40 keV and 200 keV. A combined charge and CCD analysis yields a background-limited upper limit of $1.1times10^{-5}$ (90% confidence level) for the rejection of $gamma$ and electron events. Backgrounds from alpha decays from the field cage are eliminated by introducing a veto electrode that surrounds the sensitive region in the TPC. CCD-specific backgrounds are reduced more than two orders of magnitude when requiring a coincidence with the charge readout.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
