Deep-Learning Unfolding for Extraction of Drell-Yan Angular Parameters in $pp$ Collisions with 120 GeV Beam Energy

Extracting particle-level information from detector measurements (unfolding) often relies on matrix methods, which become computationally expensive in high-dimensional phase spaces. We propose an unbinned unfolding approach using deep neural networks to tackle this challenge. This method is applied to the precise measurement of the $\cos2\phi$ asymmetry in the Drell-Yan process with SeaQuest data, where $\phi$ is the azimuthal angle of the $\mu^{+}\mu^{-}$ pair in the Collins-Soper frame. SeaQuest, a fixed-target Drell-Yan experiment at Fermilab, scattered unpolarized protons with unpolarized LH$_{2}$ and LD$_{2}$ targets. This measurement offers valuable insights into the proton's structure and the transverse momentum ($q_{T}$) dependence of the $\cos2\phi$ asymmetry. In this presentation, we will discuss the design of our neural network architecture, training strategies, and plans to achieve definitive results.