Interplay of Traditional Methods and Machine Learning Algorithms for Tagging Boosted Objects

TL;DR

This paper reviews the integration of traditional jet substructure techniques with modern machine learning methods for tagging boosted objects like Higgs bosons and top quarks at the LHC, aiming to improve interpretability and performance.

Contribution

It introduces a hybrid approach combining traditional and machine learning methods for boosted object tagging, enhancing interpretability and potential for new physics searches.

Findings

Hybrid taggers improve interpretability of ML methods.

Combining traditional and ML techniques enhances tagging accuracy.

Proposed methods are relevant for both SM and BSM physics.

Abstract

Interest in deep learning in collider physics has been growing in recent years, specifically in applying these methods in jet classification, anomaly detection, particle identification etc. Among those, jet classification using neural networks is one of the well-established areas. In this review, we discuss different tagging frameworks available to tag boosted objects, especially boosted Higgs boson and top quark, at the Large Hadron Collider (LHC). Our aim is to study the interplay of traditional jet substructure based methods with the state-of-the-art machine learning ones. In this methodology, we would gain some interpretability of those machine learning methods, and which in turn helps to propose hybrid taggers relevant for tagging of those boosted objects belonging to both Standard Model (SM) and physics beyond the SM.