Please visit Jefferson Lab Event Policies and Guidance before planning your next event: https://www.jlab.org/conference_planning.

Apr 12 – 14, 2023
Minneapolis, Minnesota
US/Central timezone

Deep learning models for deeply virtual exclusive processes

Apr 12, 2023, 2:00 PM
20m
Orchestra A

Orchestra A

Speaker

Brandon Kriesten (University of Virginia)

Description

Deeply virtual exclusive reactions are theorized to be sensitive to the dynamics of bound partons in hadrons through 3D quantum mechanical phase space distributions - the generalized parton distributions; however, there are many steps in the analysis from experimental data to information on hadron structure. The FemtoNet framework was developed to analyze deeply virtual exclusive experimental data using physics-informed deep learning models in order to quantify information loss and reconstruction through the many inverse problems encountered. Simultaneously, the FemtoNet framework leverages a suite of uncertainty quantification techniques to separate epistemic (reducible) and aleatoric (irreducible) errors from the analysis and properly propagate experimental uncertainty. I will demonstrate what physics-informed deep neural networks are capable of in the context of reconstructing lost information from inverse problems in exclusive scattering experiments and give prospects for the future of such a program and consequences for an EIC.

Primary author

Brandon Kriesten (University of Virginia)

Co-authors

simonetta liuti (university of virginia) Huey-Wen Lin (Michigan State University) Yaohang Li (Old Dominion University) Manal Almaeen (Old Dominion University)

Presentation materials