Data driven background estimation in HEP using Generative Adversarial Networks

TL;DR

This paper introduces a novel GAN-based method for data-driven background estimation in high energy physics, enabling the generation of physics objects that preserve correlations with event properties, improving analysis accuracy.

Contribution

The paper presents a new approach using GANs to generate physics objects that maintain correlations, reducing bias in background estimation for HEP analyses.

Findings

GAN can generate coherent physics objects with proper correlations.

Method improves background modeling in HEP experiments.

Demonstrated on photon misidentification in LHC data.

Abstract

Data-driven methods are widely used to overcome shortcomings of Monte Carlo simulations (lack of statistics, mismodeling of processes, etc.) in experimental high energy physics. A precise description of background processes is crucial to reach the optimal sensitivity for a measurement. However, the selection of the control region used to describe the background process in a region of interest biases the distribution of some physics observables, rendering the use of such observables impossible in a physics analysis. Rather than discarding these events and/or observables, we propose a novel method to generate physics objects compatible with the region of interest and properly describing the correlations with the rest of the event properties. We use a generative adversarial network (GAN) for this task, as GANs are among the best generator models for various applications. We illustrate the method by generating a new misidentified photon for the $\gamma + \mathrm{jets}$ background of the $\mathrm{H}\to\gamma\gamma$ analysis at the CERN LHC, and demonstrate that this GAN generator is able to produce a coherent object correlated with the different properties of the rest of the event.