Machine learning slightly improves invisible dark boson detection at collider
Machine Learning for Invisible Dark Boson Searches at the Electron-Ion Collider
Machine Learning
Summary
Finding invisible particles called dark bosons is challenging because they leave no direct signals in detectors. The authors studied whether a machine learning method called boosted decision trees (BDTs) can do better than simple threshold cuts at spotting these particles in electron-gold collisions. Using only electron data, the machine learning approach showed no clear improvement. However, when including a specific measured quantity called momentum transfer, BDTs performed slightly better for certain particle masses. This suggests machine learning may offer small benefits in these searches when more detailed data is used.
What this means in practice
- •For particle physicists: Enhance signal detection methods in electron-ion collider experiments by incorporating momentum transfer data with machine learning models.
- •For data scientists in physics: Optimize classification algorithms for rare event searches by integrating additional kinematic variables beyond traditional inputs.
Authors
Rojae Mighty, Ankush Reddy Kanuganti
Abstract
We investigate whether adding $t$, the positive magnitude of the squared nuclear four-momentum transfer, enables boosted decision trees (BDTs) to improve invisible-dark-boson selection relative to optimized rectangular cuts at the Electron-Ion Collider. We model coherent exclusive scalar and vector production at generator level in electron-gold collisions at 18 GeV by 100 GeV per nucleon. Both methods use identical weighted samples, inputs, preselection, and optimization objectives. Using only electron information, the BDT provided no consistent advantage over optimized cuts for signal selection across 11 masses for each boson type. When both methods also use $t$, the BDT distinguishes signal from background slightly better than optimized cuts at 10 GeV for both boson types. These results motivate further investigation of machine learning in EIC dark-boson searches through exclusive processes where $t$ can be reconstructed.