Deep neural networks help to identify the neutrinoless double beta decay signal


The simulated tracks arised from electrons discharged in a neutrinoless double beta decay occasion. Credit: ©ScienceChinaPress

A group of scientists from Shanghai Jiao Tong University and Peking University significantly enhanced the discrimination power of tracks from various particles travelling through the gaseous detector with the help of deep convolutional neuralnetworks The work will help to enhance the level of sensitivity of detection for the PandaX-III neutrinoless double beta decay experiment, and deepen our understanding of the nature of neutrinos.

This work is released by ScienceChina Physics, Mechanics & &Astronomy(SCPMA) withthe title “Signal-background discrimination with convolutional neural networks in the PandaX-III experiment using MC simulation.” Hao Qiao, a master trainee from Peking University, is the very first author.

.

Doublebeta decay is a phenomenon where 2 electrons and 2 neutrinos are discharged. The so called neutrinoless double beta decay, without the emission of neutrinos, is straight associated to the nature of the neutrino itself, and has actually not been observed in any experiments. The procedure is just possible when neutrino is Majorana fermion, or, neutrino and anti-neutrino are the exact same. Scientists presume that such residential or commercial property is likewise essential to understanding the asymmetry in between matter and antimatter in our universe.

.

Specially created experiments may be able to discover the uncommon neutrinoless double beta decay procedure. Among them, the PandaX-III experiment strategies to look for the neutrinoless double beta decay of the Xe-136 isotope with a high pressure gaseous xenon time forecast chamber. It will not just record the energy of the electrons produced in the procedure, however likewise take “snapshot” of their tracks in the detector, acquiring the forecasts on 2 equally perpendicular aircrafts parallel to the wandering instructions. The functions of the tracks can be utilized for the discrimination in between preferred signals and backgrounds. But the stochastic residential or commercial property makes it tough to specify and identify the functions of neutrinoless double beta decay signals. One example of the simulated tracks arised from a neutrinoless double beta decay occasion is displayed inFig 1.

.

“The traditional topological method for track discrimination employs only two widely used parameters. More features are desirable, but are difficult to find out,” stated Siguang Wang, associated teacher of Peking University and author of this work.

.

Onthe other hand, excellent enhancements have actually been attained in image category with innovations based upon the convolutional neuralnetworks This influenced the scientists of this work to use these innovations in the discrimination of various particle tracks. The matching author, Xun Chen, assistant research study fellow in Shanghai Jiao Tong University, believed that it may lead to much better lead to contrast with the conventional technique.

.

The curve of background rejection performance versus signal performance from the experienced design. Credit: ©ScienceChinaPress

The scientists built a comprehensive simulation of the PandaX-III detector in a computer system and created information of both signal and background occasions. The information were transformed to images, and the photos taken by the detector were encoded to the color channels of the images.

.

Hao described, “You need to consider the spatial and time distribution of the detected signal simultaneously and make a cut on them when converting the data to images, because the neutral network limits the size of the input image.”

.

About560,000 pictures of the neutrinoless double beta decay occasions and the exact same variety of images for the high energy gamma backgrounds were created. The scientists utilized 80 percent of the information set to train the neural network designs.

.

“We started from a simple model with only three layers, and then tried more complicated models. We finally chose the ResNet50 model for our studies,” stated Chunyu Lu, an undergraduate trainee from Shanghai Jiao Tong University, co-author of the post.

.

The last experienced design revealed outstanding signal- background discrimination power. Relative high signal performance is kept with a high background rejection performance, as displayed inFig 2. The performance has actually been enhanced by 62 percent in contrast with the standard technique specified in the style of PandaX-III experiment. This likewise suggests that the detection level of sensitivity of the neutrinoless double beta decay procedure in the PandaX-III experiment might be enhanced significantly.

.

KeHan, a recognized research study fellow of Shanghai Jiao Tong University and a co-author, commented,”CNN mades the tracking capability of PandaX-III more distinctive and attractive in the crowded field of neutrinoless double beta decay experiments.”


Explore even more:
Investigatingthe Neutrino Mass Scale with the ultra-low background KamLAND-Zen detector.

.
More details:
HaoQiao et al, Signal- background discrimination with convolutional neural networks in the PandaX-III experiment utilizing MC simulation, ScienceChina Physics, Mechanics & & Astronomy(2018). DOI: 10.1007/ s11433-018-9233 -5.

Provided by:
ScienceChinaPress

Recommended For You

About the Author: livescience

Leave a Reply

Your email address will not be published. Required fields are marked *