Iterative subtraction method for Feature Ranking

TL;DR

This paper compares various feature ranking methods for classifying top-quark pairs with Higgs bosons at the LHC, highlighting the superior performance of the iterative subtraction approach.

Contribution

The paper introduces and evaluates an iterative subtraction method for feature ranking, demonstrating its effectiveness over existing techniques in high-energy physics classification tasks.

Findings

Iterative subtraction outperforms traditional ranking methods.

Permutation-based methods are computationally expensive.

Iterative removal and addition procedures are highly effective.

Abstract

Training features used to analyse physical processes are often highly correlated and determining which ones are most important for the classification is a non-trivial tasks. For the use case of a search for a top-quark pair produced in association with a Higgs boson decaying to bottom-quarks at the LHC, we compare feature ranking methods for a classification BDT. Ranking methods, such as the BDT Selection Frequency commonly used in High Energy Physics and the Permutational Performance, are compared with the computationally expense Iterative Addition and Iterative Removal procedures, while the latter was found to be the most performant.