Sam Young

I’m a second year PhD student in the Physics Department at Stanford University interested in helping computers learn understand particle physics on their own. I am advised by Kazu Terao and Hiro Tanaka in the neutrino group at SLAC National Accelerator Laboratory and work on reconstruction techniques for the Deep Underground Neutrino Experiment and it’s predecessors. The experiment aims to understand the properties of neutrinos, invisible fundamental particles that rarely interact with matter.

I’m broadly interested in:

Self-supervised representation learning for scientific imaging data
3D computer vision for particle physics
Differentiable surrogate models for simulation calibration

I’m grateful to be supported by the Stanford Institute for Human-Centered Artificial Intelligence (HAI) Graduate Fellowship.

Previously, I worked with Julia Gonski of the SLAC ATLAS group on applying machine learning techniques for anomaly detection in the ATLAS experiment at CERN, and with Giorgio Gratta of the Stanford neutrino group on detector R&D for the upcoming nEXO experiment. Before that, I was a research assistant at the University of Pennsylvania working with Prof. Josh Klein on novel neutrino detector instrumentation for THEIA, a proposed hybrid optical neutrino detector. I graduated summa cum laude and Phi Beta Kappa from Penn in 2023 with a bachelor’s and master’s in physics.

publications

2025

Particle Trajectory Representation Learning with Masked Point Modeling

Sam Young, Yeon-jae Jwa, and Kazuhiro Terao

2025

Abstract Project Page arXiv Cite

Effective self-supervised learning (SSL) techniques have been key to unlocking large datasets for representation learning. While many promising methods have been developed using online corpora and captioned photographs, their application to scientific domains, where data encodes highly specialized knowledge, remains in its early stages. We present a self-supervised masked modeling framework for 3D particle trajectory analysis in Time Projection Chambers (TPCs). These detectors produce globally sparse (<1% occupancy) but locally dense point clouds, capturing meter-scale particle trajectories at millimeter resolution. Starting with PointMAE, this work proposes volumetric tokenization to group sparse ionization points into resolution-agnostic patches, as well as an auxiliary energy infilling task to improve trajectory semantics. This approach – which we call Point-based Liquid Argon Masked Autoencoder (PoLAr-MAE) – achieves 99.4% track and 97.7% shower classification F-scores, matching that of supervised baselines without any labeled data. While the model learns rich particle trajectory representations, it struggles with sub-token phenomena like overlapping or short-lived particle trajectories. To support further research, we release PILArNet-M – the largest open LArTPC dataset (1M+ events, 5.2B labeled points) – to advance SSL in high energy physics (HEP). Project site: this https URL
@misc{young2025particletrajectoryrepresentationlearning, title = {Particle Trajectory Representation Learning with Masked Point Modeling}, author = {Young, Sam and Jwa, Yeon-jae and Terao, Kazuhiro}, year = {2025}, eprint = {2502.02558}, archiveprefix = {arXiv}, primaryclass = {hep-ex}, }