Improving $Λ$ Signal Extraction with Domain Adaptation via Normalizing Flows

TL;DR

This study investigates the ability of flow based neural networks to improve signal extraction of $\Lambda$ Hyperons at CLAS12 and was successful in training a flow network to transform between the latent physics space and a normal distribution.

Abstract

The present study presents a novel application for normalizing flows for domain adaptation. The study investigates the ability of flow based neural networks to improve signal extraction of $Λ$ Hyperons at CLAS12. Normalizing Flows can help model complex probability density functions that describe physics processes, enabling uses such as event generation. $Λ$ signal extraction has been improved through the use of classifier networks, but differences in simulation and data domains limit classifier performance; this study utilizes the flows for domain adaptation between Monte Carlo simulation and data. We were successful in training a flow network to transform between the latent physics space and a normal distribution. We also found that applying the flows lessened the dependence of the figure of merit on the cut on the classifier output, meaning that there was a broader range where the cut results in a similar figure of merit.

Figures (4)

• Figure 1: A few dimensions are shown for the latent data (left) and the normalized data (passed through the first NF model, but not the second) (right).
• Figure 2: FOM and purity for transformed data (left) and non-transformed data (right).
• Figure 3: ROC curve (left) for the latent MC and classifier output across latent data, transformed data, and MC. The ROC curve had an AUC of 0.90
• Figure 4: Plots of the proton $p_T$ distribution for the MC $p_T$, distorted $p_T$, and transformed $p_T$ with transformed $p_T$ plotted against distorted $p_T$. Proton $p_T$ is distorted by adding a value drawn from a normal distribution with $\sigma = 0.1$ (left) and by shifting every $p_T$ 0.1 GeV (right)