Jet grooming through reinforcement learning

Abstract

We introduce a novel implementation of a reinforcement learning (RL) algorithm which is designed to find an optimal jet grooming strategy, a critical tool for collider experiments. The RL agent is trained with a reward function constructed to optimize the resulting jet properties, using both signal and background samples in a simultaneous multi-level training. We show that the grooming algorithm derived from the deep RL agent can match state-of-the-art techniques used at the Large Hadron Collider, resulting in improved mass resolution for boosted objects. Given a suitable reward function, the agent learns how to train a policy which optimally removes soft wide-angle radiation, allowing for a modular grooming technique that can be applied in a wide range of contexts. These results are accessible through the corresponding GroomRL framework.

11 pages, 10 figures, code available at https://github.com/JetsGame/GroomRL, updated to match published version

References in corpus (30)

• The anti-k_t jet clustering algorithm
• An Introduction to PYTHIA 8.2
• FastJet user manual
• Successive Combination Jet Algorithm For Hadron Collisions
• Better Jet Clustering Algorithms
• Jet substructure as a new Higgs search channel at the LHC
• Soft Drop
• Jet Trimming
• OpenAI Gym
• Towards an understanding of jet substructure
• Towards Jetography
• Recombination Algorithms and Jet Substructure: Pruning as a Tool for Heavy Particle Searches
• Learning Particle Physics by Example: Location-Aware Generative Adversarial Networks for Physics Synthesis
• Jet-Images -- Deep Learning Edition
• Deep-learning Top Taggers or The End of QCD?
• Jet Substructure Classification in High-Energy Physics with Deep Neural Networks
• Energy Flow Networks: Deep Sets for Particle Jets
• Deep-learned Top Tagging with a Lorentz Layer
• Jet Flavor Classification in High-Energy Physics with Deep Neural Networks
• Jet-Images: Computer Vision Inspired Techniques for Jet Tagging
• The Lund Jet Plane
• Decorrelated Jet Substructure Tagging using Adversarial Neural Networks
• QCD-Aware Recursive Neural Networks for Jet Physics
• Pileup Mitigation with Machine Learning (PUMML)
• JUNIPR: a Framework for Unsupervised Machine Learning in Particle Physics
• On the Topic of Jets: Disentangling Quarks and Gluons at Colliders
• How Much Information is in a Jet?
• A measurement of the soft-drop jet mass in pp collisions at $\sqrt{s}=13$ TeV with the ATLAS detector
• Measurements of the differential jet cross section as a function of the jet mass in dijet events from proton-proton collisions at $\sqrt{s}=$ 13 TeV
• Recursive Soft Drop

Cited by in corpus (12)

• A guide for deploying Deep Learning in LHC searches: How to achieve optimality and account for uncertainty
• Pileup and Underlying Event Mitigation with Iterative Constituent Subtraction
• Explainable machine learning of the underlying physics of high-energy particle collisions
• Identification of b-jets using QCD-inspired observables
• Reframing Jet Physics with New Computational Methods
• Tagging boosted hadronic objects with dynamical grooming
• Identifying quenched jets in heavy ion collisions with machine learning
• Interpretable deep learning models for the inference and classification of LHC data
• Jet tagging in the Lund plane with graph networks
• Hierarchical clustering in particle physics through reinforcement learning
• Gradient dynamics in reinforcement learning
• Variational inference for pile-up removal at hadron colliders with diffusion models