Comparison of Machine Learning Approach to other Unfolding Methods

TL;DR

This paper compares traditional unfolding methods in high energy physics with a machine learning approach called Omnifold, highlighting differences in efficiency and coverage of variables.

Contribution

It provides a direct comparison between standard unfolding techniques and the novel Omnifold machine learning method.

Findings

Omnifold allows event-by-event unfolding of multiple variables.

Machine learning approach offers potential advantages over traditional methods.

Comparison highlights strengths and limitations of each approach.

Abstract

Unfolding in high energy physics represents the correction of measured spectra in data for the finite detector efficiency, acceptance, and resolution from the detector to particle level. Recent machine learning approaches provide unfolding on an event-by-event basis allowing to simultaneously unfold a large number of variables and thus to cover a wider region of the features that affect detector response. This study focuses on a simple comparison of commonly used methods in RooUnfold package to the machine learning package Omnifold.