Undergraduate Research at Jefferson Lab

AI-Driven Detector Reconstruction Enhancement for Pixelated Detectors for the Electron-Ion Collider

Student: Nathan Branson

School: Messiah College

Mentored By: Dmitry Romanov

With artificial intelligence (AI) becoming widespread in many professional fields, AI detector optimization is a logical step for nuclear physics. This project focuses on enhancing pixelated detector data using AI. Current algorithms cannot differentiate multiple particles that hit in similar locations. AI will allow us to differentiate between these hits and get a better resolution from our detectors. The main model we look at is the convolutional autoencoder. The model was created from the tensorflow and keras python libraries and tested on Geant3 calorimeter data. Here we show how AI can be a tremendous benefit to detector resolution and to physics as a whole. Implementation of ML will allow the EIC detectors to have a much better resolution for event reconstruction.

[Watch the presentation on YouTube]