学术报告

Deep Neural Networks in EXO-200

byProf.Igor Ostrovskiy(University of Alabama)

→Asia/Shanghai
122 Multidisciplinary Building

122 Multidisciplinary Building

Description

Abstract:

EXO-200was an experiment searching theneutrinolessdouble beta decay using a liquid-phasetime projectionchamber filled with Xe-136. It accumulated a large low-background dataset of physics data, as well as uniquecalibrationdata that remain to be a perfect testbed for novel analysis algorithms, such as deep neural networks (DNNs).Afterintroducing the EXO-200 experiment, this talk describes successfulapplicationsof DNNs to simulate, reconstruct, and analyze the EXO-200 data. We demonstrate that DNNs are able to extract necessary high-level features directly from raw waveforms, with minimal pre- processing. The developed DNNs andconventionalanalyses approaches are compared back-to-back. Importantly, the DNNs described in this talk are validated with real detector data, and some are even trained on the data. This shows thepotentialof DNNs in reducing or eveneliminatingthe reliance on theconventionalMonte Carlo and itsimperfections. The talk concludes by discussing advantages and challenges of DNNs for the upcomingnext-generationlow-background experiments.

About the speaker:

Dr.IgorOstrovskiyworked at JINR and INFN before receiving his Ph.D. at the University of Alabama, where he performed the first analysis of the DoubleChoozfar detector data that provided a hint of non-zero value of the neutrino mixing angleθ13.He then did a postdoc at Stanford before returning to the University of Alabama where he currently holds an associate professorposition. His group's two primary research foci are searches formagneticmonopoles at the LHC and R&D for the nextgenerationof liquid-xenon detectors that will search for dark mater andneutrinolessdouble beta decay. He is a member of theMoEDAL, DARWIN, EXO, and MONUMENTcollaborations.

----------


会议号 Meeting ID: 9472 1492 781
会议链接 Meeting URL:https://zoom.us/j/94721492781?pwd=bVM0QWFnV3VTMnhqaytuRHE1cy9lUT09
入会密码Password:240314
主持密码Host key:068014

Baidu
map