Learning the latent structure of collider events

TL;DR

This paper introduces an unsupervised machine learning method based on Latent Dirichlet Allocation to discover the underlying structure of collider events directly from data, aiding analysis without predefined models.

Contribution

It applies a Bayesian generative model with Variational Inference to learn latent structures in collider event data, enabling clustering and classification without prior theoretical assumptions.

Findings

Successfully learned latent structures in di-jet events

Distinguished between QCD background and signal events

Demonstrated effectiveness in different physics scenarios

Abstract

We describe a technique to learn the underlying structure of collider events directly from the data, without having a particular theoretical model in mind. It allows to infer aspects of the theoretical model that may have given rise to this structure, and can be used to cluster or classify the events for analysis purposes. The unsupervised machine-learning technique is based on the probabilistic (Bayesian) generative model of Latent Dirichlet Allocation. We pair the model with an approximate inference algorithm called Variational Inference, which we then use to extract the latent probability distributions describing the learned underlying structure of collider events. We provide a detailed systematic study of the technique using two example scenarios to learn the latent structure of di-jet event samples made up of QCD background events and either $t\bar{t}$ or hypothetical $W' \to (\phi \to WW) W$ signal events.