Konuşmacılar
Açıklama
In high-energy particle physics, detecting the position of particles provides critical insights into particles about them. There are some traditional methods for this determination. However, the amount of data is large, so processing time and cost are high. Instead, artificial intelligence support can solve this challenge. We aimed to use an artificial intelligence sparse readout system to reduce the cost and to look at how to use a lot of data more effectively.
We employed a plastic scintillation detector, which emits light (scintillation) upon interaction with incident charged particles. We set the scintillation detector as a square block and selected four silicon photomultipliers (SiPMs) that read electrical signals, one in each corner of the square block. The entire detector geometry was simulated using the Geant4 toolkit, and analyzed the signals from the detectors with the ROOT data analysis framework.
We chose the muon particle, one of the cosmic rays, as the particle. These muons strike towards the scintillation detector from various (x, y) coordinates. Then, from the signals created, we trained an artificial neural network (DNN) model to determine where the muons strike and interact more. As a result, the model showed a success rate in position detection.
