Transformer networks for Heavy flavor jet tagging

TL;DR

This paper reviews the application of Transformer networks in heavy flavor jet tagging at high-energy colliders, highlighting performance improvements, interpretability, and physics-inspired modifications of deep learning models.

Contribution

It introduces the use of attention-based Transformer networks for heavy flavor jet tagging and discusses enhancements based on physics insights and interpretability methods.

Findings

Transformer networks improve jet tagging accuracy

Physics-inspired modifications enhance model performance

Interpretable methods aid understanding of network decisions

Abstract

In this article, we review recent machine learning methods used in challenging particle identification of heavy-boosted particles at high-energy colliders. Our primary focus is on attention-based Transformer networks. We report the performance of state-of-the-art deep learning networks and further improvement coming from the modification of networks based on physics insights. Additionally, we discuss interpretable methods to understand network decision-making, which are crucial when employing highly complex and deep networks.