# Uncovering latent jet substructure

## TL;DR

This paper introduces an unsupervised Bayesian generative modeling approach using Latent Dirichlet Allocation to analyze jet substructure, enabling data-driven top-quark tagging and potential discovery of new physics signatures in multi-jet events.

## Contribution

It applies LDA, a mixed membership model, to jet substructure data for the first time, providing a novel unsupervised method for top-quark tagging and new physics searches.

## Key findings

- Effective top-quark tagging demonstrated
- Comparable or superior to traditional methods
- Potential for model-independent new physics discovery

## Abstract

We apply techniques from Bayesian generative statistical modeling to uncover hidden features in jet substructure observables that discriminate between different a priori unknown underlying short distance physical processes in multi-jet events. In particular, we use a mixed membership model known as Latent Dirichlet Allocation to build a data-driven unsupervised top-quark tagger and $t\bar t$ event classifier. We compare our proposal to existing traditional and machine learning approaches to top jet tagging. Finally, employing a toy vector-scalar boson model as a benchmark, we demonstrate the potential for discovering New Physics signatures in multi-jet events in a model independent and unsupervised way.