Energy Flow Networks: Deep Sets for Particle Jets

Abstract

A key question for machine learning approaches in particle physics is how to best represent and learn from collider events. As an event is intrinsically a variable-length unordered set of particles, we build upon recent machine learning efforts to learn directly from sets of features or "point clouds". Adapting and specializing the "Deep Sets" framework to particle physics, we introduce Energy Flow Networks, which respect infrared and collinear safety by construction. We also develop Particle Flow Networks, which allow for general energy dependence and the inclusion of additional particle-level information such as charge and flavor. These networks feature a per-particle internal (latent) representation, and summing over all particles yields an overall event-level latent representation. We show how this latent space decomposition unifies existing event representations based on detector images and radiation moments. To demonstrate the power and simplicity of this set-based approach, we apply these networks to the collider task of discriminating quark jets from gluon jets, finding similar or improved performance compared to existing methods. We also show how the learned event representation can be directly visualized, providing insight into the inner workings of the model. These architectures lend themselves to efficiently processing and analyzing events for a wide variety of tasks at the Large Hadron Collider. Implementations and examples of our architectures are available online in our EnergyFlow package.

arXiv comment31+16 pages, 21 figures, 5 tables; v2: updated to match JHEP version; code available at https://energyflow.network

References in corpus (58)

• PYTHIA 6.4 Physics and Manual
• PYTHIA 6.2 Physics and Manual
• The anti-k_t jet clustering algorithm
• An Introduction to PYTHIA 8.2
• FastJet user manual
• DELPHES 3, A modular framework for fast simulation of a generic collider experiment
• Searching for Exotic Particles in High-Energy Physics with Deep Learning
• Identifying Boosted Objects with N-subjettiness
• Soft Drop
• Fast inference of deep neural networks in FPGAs for particle physics
• Energy Correlation Functions for Jet Substructure
• Deep Learning and its Application to LHC Physics
• Maximizing Boosted Top Identification by Minimizing N-subjettiness
• Pileup Per Particle Identification
• Boosted objects: a probe of beyond the Standard Model physics
• Jet Substructure at the Tevatron and LHC: New results, new tools, new benchmarks
• Classification without labels: Learning from mixed samples in high energy physics
• Learning Particle Physics by Example: Location-Aware Generative Adversarial Networks for Physics Synthesis
• Deep learning in color: towards automated quark/gluon jet discrimination
• Jet-Images -- Deep Learning Edition
• Accelerating Science with Generative Adversarial Networks: An Application to 3D Particle Showers in Multi-Layer Calorimeters
• Deep-learning Top Taggers or The End of QCD?
• Anomaly Detection for Resonant New Physics with Machine Learning
• Gaining (Mutual) Information about Quark/Gluon Discrimination
• Quark and Gluon Tagging at the LHC
• Jet Shapes with the Broadening Axis
• Deep-learned Top Tagging with a Lorentz Layer
• New Angles on Energy Correlation Functions
• Jet-Images: Computer Vision Inspired Techniques for Jet Tagging
• Systematics of quark/gluon tagging
• A generic anti-QCD jet tagger
• Constraining Effective Field Theories with Machine Learning
• Quark and Gluon Jet Substructure
• Jet Charge at the LHC
• Energy flow polynomials: A complete linear basis for jet substructure
• Pulling Out All the Tops with Computer Vision and Deep Learning
• The Lund Jet Plane
• Jets and Quantum Field Theory
• Pileup Mitigation with Machine Learning (PUMML)
• Calculating Track-Based Observables for the LHC
• Enhanced Higgs to ${\tau }^{+}{\tau }^{-}$ Searches with Deep Learning
• LHCb Topological Trigger Reoptimization
• On the Topic of Jets: Disentangling Quarks and Gluons at Colliders
• Unsafe but Calculable: Ratios of Angularities in Perturbative QCD
• Boosting $H\to b\bar{b}$ with Machine Learning
• Casimir Meets Poisson: Improved Quark/Gluon Discrimination with Counting Observables
• How Much Information is in a Jet?
• Weakly Supervised Classification in High Energy Physics
• Recursive Neural Networks in Quark/Gluon Tagging
• Novel Jet Observables from Machine Learning
• Measuring Multijet Structure of Hadronic Energy Flow Or What IS A Jet?
• Jet Charge and Machine Learning
• Toward Multi-Differential Cross Sections: Measuring Two Angularities on a Single Jet
• (Machine) Learning to Do More with Less
• An operational definition of quark and gluon jets
• Joint resummation of two angularities at next-to-next-to-leading logarithmic order
• Learning to pinpoint effective operators at the LHC: a study of the $t\bar{t}b\bar{b}$ signature
• Machine learning techniques in searches for $t\bar{t}h$ in the $h\to b\bar{b}$ decay channel

Cited by in corpus (169)

• Machine learning and the physical sciences
• ParticleNet: Jet Tagging via Particle Clouds
• Jet Substructure at the Large Hadron Collider: A Review of Recent Advances in Theory and Machine Learning
• The Machine Learning Landscape of Top Taggers
• Graph Neural Networks in Particle Physics
• OmniFold: A Method to Simultaneously Unfold All Observables
• QCD or What?
• Simulation Assisted Likelihood-free Anomaly Detection
• Learning representations of irregular particle-detector geometry with distance-weighted graph networks
• JEDI-net: a jet identification algorithm based on interaction networks
• The Metric Space of Collider Events
• The Dark Machines Anomaly Score Challenge: Benchmark Data and Model Independent Event Classification for the Large Hadron Collider
• An Efficient Lorentz Equivariant Graph Neural Network for Jet Tagging
• Sampling using $SU(N)$ gauge equivariant flows
• Analyzing N-point Energy Correlators Inside Jets with CMS Open Data
• Tag N' Train: A Technique to Train Improved Classifiers on Unlabeled Data
• ABCNet: An attention-based method for particle tagging
• Neural Networks for Full Phase-space Reweighting and Parameter Tuning
• DisCo Fever: Robust Networks Through Distance Correlation
• Energy-energy correlations at next-to-next-to-leading order
• MLPF: Efficient machine-learned particle-flow reconstruction using graph neural networks
• Interaction networks for the identification of boosted $H\to b\overline{b}$ decays
• GANplifying Event Samples
• A Neural Resampler for Monte Carlo Reweighting with Preserved Uncertainties
• Exploring the Space of Jets with CMS Open Data
• ABCDisCo: Automating the ABCD Method with Machine Learning
• Quark-Gluon Tagging: Machine Learning vs Detector
• A guide for deploying Deep Learning in LHC searches: How to achieve optimality and account for uncertainty
• Deep-Learning Jets with Uncertainties and More
• Unveiling Hidden Physics at the LHC
• Point Cloud Transformers applied to Collider Physics
• Casting a graph net to catch dark showers
• Mapping Machine-Learned Physics into a Human-Readable Space
• The Hidden Geometry of Particle Collisions
• Interpretable Deep Learning for Two-Prong Jet Classification with Jet Spectra
• Supervised Jet Clustering with Graph Neural Networks for Lorentz Boosted Bosons
• EPiC-GAN: Equivariant Point Cloud Generation for Particle Jets
• PC-JeDi: Diffusion for Particle Cloud Generation in High Energy Physics
• A Theory of Quark vs. Gluon Discrimination
• Deep Set Auto Encoders for Anomaly Detection in Particle Physics
• Symmetries, Safety, and Self-Supervision
• Equivariant Energy Flow Networks for Jet Tagging
• End-to-End Jet Classification of Quarks and Gluons with the CMS Open Data
• Per-Object Systematics using Deep-Learned Calibration
• Permutationless Many-Jet Event Reconstruction with Symmetry Preserving Attention Networks
• E Pluribus Unum Ex Machina: Learning from Many Collider Events at Once
• The Boosted Higgs Jet Reconstruction via Graph Neural Network
• Linearized Optimal Transport for Collider Events
• Measuring QCD Splittings with Invertible Networks
• Automating the Construction of Jet Observables with Machine Learning
• Topological Obstructions to Autoencoding
• Quarks and gluons in the Lund plane
• Energy-weighted Message Passing: an infra-red and collinear safe graph neural network algorithm
• Binary JUNIPR: an interpretable probabilistic model for discrimination
• Towards Machine Learning Analytics for Jet Substructure
• Jet Flavour Tagging for Future Colliders with Fast Simulation
• Neural Network-based Top Tagger with Two-Point Energy Correlations and Geometry of Soft Emissions
• Higgs boson tagging with the Lund jet plane
• Calculating the primary Lund Jet Plane density
• Three-point energy correlator in $\mathcal{N}=4$ super Yang-Mills Theory
• Explainable Equivariant Neural Networks for Particle Physics: PELICAN
• OmniJet-$\alpha$: The first cross-task foundation model for particle physics
• Disentangling Quarks and Gluons with CMS Open Data
• Deep Generative Models for Detector Signature Simulation: A Taxonomic Review
• Shared Data and Algorithms for Deep Learning in Fundamental Physics
• Lorentz group equivariant autoencoders
• Probing triple Higgs coupling with machine learning at the LHC
• Machine Learning in High Energy Physics: A review of heavy-flavor jet tagging at the LHC
• A method to challenge symmetries in data with self-supervised learning
• Machine learning-based jet and event classification at the Electron-Ion Collider with applications to hadron structure and spin physics
• Combine and Conquer: Event Reconstruction with Bayesian Ensemble Neural Networks
• Non-Gaussianities in Collider Energy Flux
• Which Metric on the Space of Collider Events?
• HHH Whitepaper
• CapsNets Continuing the Convolutional Quest
• Jet grooming through reinforcement learning
• Cutting Multiparticle Correlators Down to Size
• Secondary Vertex Finding in Jets with Neural Networks
• Study of energy deposition patterns in hadron calorimeter for prompt and displaced jets using convolutional neural network
• A Method to Simultaneously Facilitate All Jet Physics Tasks
• Collinear Drop
• Reconstructing boosted Higgs jets from event image segmentation
• Jet tagging algorithm of graph network with HaarPooling message passing
• A Detailed Study of Interpretability of Deep Neural Network based Top Taggers
• A Holistic Approach to Predicting Top Quark Kinematic Properties with the Covariant Particle Transformer
• Resolving Extreme Jet Substructure
• A Lorentz-Equivariant Transformer for All of the LHC
• Exploring Jet Substructure in Semi-visible jets
• Learning to Identify Semi-Visible Jets
• Parameter Estimation using Neural Networks in the Presence of Detector Effects
• Jet Tagging with More-Interaction Particle Transformer
• Learning Uncertainties the Frequentist Way: Calibration and Correlation in High Energy Physics
• Does Lorentz-symmetric design boost network performance in jet physics?
• Maximum performance of strange-jet tagging at hadron colliders
• Jet Discrimination with Quantum Complete Graph Neural Network
• Parameter Inference from Event Ensembles and the Top-Quark Mass
• Semi-supervised Graph Neural Networks for Pileup Noise Removal
• Jets with electrons from boosted top quarks
• BIP: Boost Invariant Polynomials for Efficient Jet Tagging
• Equivariant, Safe and Sensitive -- Graph Networks for New Physics
• Ultrafast jet classification on FPGAs for the HL-LHC
• Comparing Point Cloud Strategies for Collider Event Classification
• Heterogeneous Graph Neural Network for Identifying Hadronically Decayed Tau Leptons at the High Luminosity LHC
• Search for new physics in final states with semi-visible jets or anomalous signatures using the ATLAS detector
• Reconstruction of decays to merged photons using end-to-end deep learning with domain continuation in the CMS detector
• Systematic Quark/Gluon Identification with Ratios of Likelihoods
• Is infrared-collinear safe information all you need for jet classification?
• Accurate Surrogate Amplitudes with Calibrated Uncertainties
• Probing Higgs exotic decay at the LHC with machine learning
• Learning Physics at Future $e^{-}{e}^{+}$ Colliders with Machine
• Learning to Isolate Muons
• Hypergraphs in LHC Phenomenology -- The Next Frontier of IRC-Safe Feature Extraction
• A Comparison Between Invariant and Equivariant Classical and Quantum Graph Neural Networks
• Accuracy versus precision in boosted top tagging with the ATLAS detector
• Interpretable deep learning models for the inference and classification of LHC data
• High-dimensional Anomaly Detection with Radiative Return in $e^{+}{e}^{-}$ Collisions
• Meta-learning and data augmentation for mass-generalised jet taggers
• Deep Learning Jet Substructure from Two-Particle Correlation
• Searches for pair-produced multijet resonances using data scouting in proton-proton collisions at $\sqrt{s}$ = 13 TeV
• Going off topics to demix quark and gluon jets in ${\alpha }_S$ extractions
• Automatic detection of boosted Higgs boson and top quark jets in an event image
• Safe but Incalculable: Energy-weighting is not all you need
• Jet Classification Using High-Level Features from Anatomy of Top Jets
• Lorentz-Equivariant Quantum Graph Neural Network for High-Energy Physics
• Reweighting simulated events using machine-learning techniques in the CMS experiment
• A rotation-equivariant graph neural network for learning hadronic SMEFT effects
• Search for long-lived heavy neutral leptons in proton-proton collision events with a lepton-jet pair associated with a secondary vertex at $\sqrt{s}$ = 13 TeV
• Attention to the strengths of physical interactions: Transformer and graph-based event classification for particle physics experiments
• TopicFlow: Disentangling quark and gluon jets with normalizing flows
• Interplay of Traditional Methods and Machine Learning Algorithms for Tagging Boosted Objects
• Longitudinal Spin Transfer to $\Lambda$ Hyperons in CLAS12
• Domain-informed neural networks for interaction localization within astroparticle experiments
• Unveiling the Secrets of New Physics Through Top Quark Tagging
• Refinable modeling for unbinned SMEFT analyses
• End-to-End Jet Classification of Boosted Top Quarks with the CMS Open Data
• Rapidly Encoding Generalizable Dynamics in a Euclidean Symmetric Neural Network
• Classify the Higgs decays with the PFN and ParticleNet at electron-positron colliders
• Quark-versus-gluon tagging in CMS Open Data with CWoLa and TopicFlow
• Spurious correlations between galaxies and multi-epoch image stacks in the DESI Legacy Surveys
• Semi-supervised permutation invariant particle-level anomaly detection
• Jet energy calibration with deep learning as a Kubeflow pipeline
• Foundations of automatic feature extraction at LHC--point clouds and graphs
• Transforming jet flavour tagging at ATLAS
• Top-philic Machine Learning
• Machine Learning-Assisted Unfolding for Neutrino Cross-section Measurements with the OmniFold Technique
• BitHEP -- The Limits of Low-Precision ML in HEP
• Learning Broken Symmetries with Approximate Invariance
• Amplitude Uncertainties Everywhere All at Once
• Particle-level transformers for 95 GeV Higgs boson searches at future $e^{+}{e}^{-}$ Higgs factories
• Preserving New Physics while Simultaneously Unfolding All Observables
• Jet Rotational Metrics
• Event Generation and Density Estimation with Surjective Normalizing Flows
• Identifying the Quantum Properties of Hadronic Resonances using Machine Learning
• Performance studies of jet flavor tagging and measurement of $R_b(R_c)$ using ParticleNet at CEPC
• Jet Reconstruction with Mamba Networks in Collider Events
• Guided Graph Compression for Quantum Graph Neural Networks
• OmniLearned: A Foundation Model Framework for All Tasks Involving Jet Physics
• The Physics Behind ML-based Quark-Gluon Taggers
• Vision Transformers for End-to-End Quark-Gluon Jet Classification from Calorimeter Images
• Stable and Interpretable Jet Physics with IRC-Safe Equivariant Feature Extraction
• Graph theory inspired anomaly detection at the LHC
• Power Counting Energy Flow Polynomials
• Search for exotic Higgs boson decays H $\to$ $\mathcal{AA}$ with $\mathcal{AA}$ $\to$ $\gamma \gamma$ in events with a semi-merged topology in proton-proton collisions at $\sqrt{s}$ = 13 TeV
• jBOT: Semantic Jet Representation Clustering Emerges from Self-Distillation
• Measurement of event shapes in minimum-bias events from proton-proton collisions at $\sqrt{s}$ = 13 TeV
• Learnable cut flow for high energy physics
• Reweighting and Analysing Event Generator Systematics by Neural Networks on High-Level Features
• Enhancing generalization in high energy physics using white-box adversarial attacks
• Deep learning for jet modification in the presence of the QGP background