Anomaly Detection with Density Estimation

Abstract

We leverage recent breakthroughs in neural density estimation to propose a new unsupervised anomaly detection technique (ANODE). By estimating the probability density of the data in a signal region and in sidebands, and interpolating the latter into the signal region, a likelihood ratio of data vs. background can be constructed. This likelihood ratio is broadly sensitive to overdensities in the data that could be due to localized anomalies. In addition, a unique potential benefit of the ANODE method is that the background can be directly estimated using the learned densities. Finally, ANODE is robust against systematic differences between signal region and sidebands, giving it broader applicability than other methods. We demonstrate the power of this new approach using the LHC Olympics 2020 R\&D Dataset. We show how ANODE can enhance the significance of a dijet bump hunt by up to a factor of 7 with a 10\% accuracy on the background prediction. While the LHC is used as the recurring example, the methods developed here have a much broader applicability to anomaly detection in physics and beyond.

28 pages, 11 figures, v2: appendix on optimality, minor modifications, journal version

References in corpus (16)

• PYTHIA 6.4 Physics and Manual
• MADE: Masked Autoencoder for Distribution Estimation
• Classification without labels: Learning from mixed samples in high energy physics
• Extending the Bump Hunt with Machine Learning
• Search for new resonances in mass distributions of jet pairs using 139 fb${}^{-1}$ of $pp$ collisions at $\sqrt{s}=13$ TeV with the ATLAS detector
• A generic anti-QCD jet tagger
• Simulation Assisted Likelihood-free Anomaly Detection
• Search for high mass dijet resonances with a new background prediction method in proton-proton collisions at $\sqrt{s}=$ 13 TeV
• DisCo Fever: Robust Networks Through Distance Correlation
• Fast simulation of muons produced at the SHiP experiment using Generative Adversarial Networks
• LHC analysis-specific datasets with Generative Adversarial Networks
• Cherenkov Detectors Fast Simulation Using Neural Networks
• Search for new phenomena with large jet multiplicities and missing transverse momentum using large-radius jets and flavour-tagging at ATLAS in 13 TeV $pp$ collisions
• Machine Learning Templates for QCD Factorization in the Search for Physics Beyond the Standard Model
• Modeling Smooth Backgrounds and Generic Localized Signals with Gaussian Processes
• Learning Multivariate New Physics

Cited by in corpus (117)

• A Unifying Review of Deep and Shallow Anomaly Detection
• Getting High: High Fidelity Simulation of High Granularity Calorimeters with High Speed
• Simulation Assisted Likelihood-free Anomaly Detection
• Dijet resonance search with weak supervision using $\sqrt{s}=13$ TeV $pp$ collisions in the ATLAS detector
• The Dark Machines Anomaly Score Challenge: Benchmark Data and Model Independent Event Classification for the Large Hadron Collider
• Machine Learning and LHC Event Generation
• Classifying Anomalies THrough Outer Density Estimation (CATHODE)
• Tag N' Train: A Technique to Train Improved Classifiers on Unlabeled Data
• Invertible Networks or Partons to Detector and Back Again
• Anomaly detection with Convolutional Graph Neural Networks
• CaloFlow: Fast and Accurate Generation of Calorimeter Showers with Normalizing Flows
• Finding New Physics without learning about it: Anomaly Detection as a tool for Searches at Colliders
• Understanding Event-Generation Networks via Uncertainties
• Better Latent Spaces for Better Autoencoders
• Variational Autoencoders for Anomalous Jet Tagging
• Simulation-Assisted Decorrelation for Resonant Anomaly Detection
• GANplifying Event Samples
• Bump Hunting in Latent Space
• Machine Learning for Anomaly Detection in Particle Physics
• ABCDisCo: Automating the ABCD Method with Machine Learning
• Rare and Different: Anomaly Scores from a combination of likelihood and out-of-distribution models to detect new physics at the LHC
• Casting a graph net to catch dark showers
• Challenges for Unsupervised Anomaly Detection in Particle Physics
• Quasi Anomalous Knowledge: Searching for new physics with embedded knowledge
• Resonant anomaly detection without background sculpting
• Unsupervised clustering for collider physics
• CURTAINs for your Sliding Window: Constructing Unobserved Regions by Transforming Adjacent Intervals
• Online-compatible Unsupervised Non-resonant Anomaly Detection
• Comparing Weak- and Unsupervised Methods for Resonant Anomaly Detection
• Deep Set Auto Encoders for Anomaly Detection in Particle Physics
• Mass Unspecific Supervised Tagging (MUST) for boosted jets
• Symmetries, Safety, and Self-Supervision
• FETA: Flow-Enhanced Transportation for Anomaly Detection
• Via Machinae: Searching for Stellar Streams using Unsupervised Machine Learning
• Improving Variational Autoencoders for New Physics Detection at the LHC with Normalizing Flows
• How to Understand Limitations of Generative Networks
• Unravelling physics beyond the standard model with classical and quantum anomaly detection
• The LHC Olympics 2020: A Community Challenge for Anomaly Detection in High Energy Physics
• E Pluribus Unum Ex Machina: Learning from Many Collider Events at Once
• Unsupervised Hadronic SUEP at the LHC
• Measuring QCD Splittings with Invertible Networks
• Jet Substructure from Dark Sector Showers
• Self-supervised Anomaly Detection for New Physics
• Learning new physics efficiently with nonparametric methods
• Neural Network-based Top Tagger with Two-Point Energy Correlations and Geometry of Soft Emissions
• ν-Flows: Conditional Neutrino Regression
• Learning the language of QCD jets with transformers
• The MadNIS Reloaded
• The Hough Stream Spotter: A New Method for Detecting Linear Structure in Resolved Stars and Application to the Stellar Halo of M31
• The Interplay of Machine Learning--based Resonant Anomaly Detection Methods
• An Iterative Method for Unsupervised Robust Anomaly Detection Under Data Contamination
• Deep Generative Models for Detector Signature Simulation: A Taxonomic Review
• What's Anomalous in LHC Jets?
• Anomaly Detection under Coordinate Transformations
• IRC-safe Graph Autoencoder for unsupervised anomaly detection
• Anomaly detection from mass unspecific jet tagging
• High-dimensional and Permutation Invariant Anomaly Detection
• Learning Likelihood Ratios with Neural Network Classifiers
• Unsupervised in-distribution anomaly detection of new physics through conditional density estimation
• Ephemeral Learning -- Augmenting Triggers with Online-Trained Normalizing Flows
• Nanosecond anomaly detection with decision trees and real-time application to exotic Higgs decays
• Boosting mono-jet searches with model-agnostic machine learning
• Precision-Machine Learning for the Matrix Element Method
• A Method to Simultaneously Facilitate All Jet Physics Tasks
• A semi-supervised approach to dark matter searches in direct detection data with machine learning
• Anomalies, Representations, and Self-Supervision
• Data-Directed Search for New Physics based on Symmetries of the SM
• GPU coprocessors as a service for deep learning inference in high energy physics
• Neural Embedding: Learning the Embedding of the Manifold of Physics Data
• Anomaly Awareness
• Normalizing Flows for High-Dimensional Detector Simulations
• Calorimeter shower superresolution
• Combining Resonant and Tail-based Anomaly Detection
• Convolutional L2LFlows: Generating Accurate Showers in Highly Granular Calorimeters Using Convolutional Normalizing Flows
• Differentiable MadNIS-Lite
• Non-Parametric Data-Driven Background Modelling using Conditional Probabilities
• Tagging boosted hadronic objects with dynamical grooming
• Neural Bridge Sampling for Evaluating Safety-Critical Autonomous Systems
• Resonant Anomaly Detection with Multiple Reference Datasets
• Enhancing the hunt for new phenomena in dijet final-states using anomaly detection filters at the High-Luminosity Large Hadron Collider
• Mixture-of-theories Training: Can We Find New Physics and Anomalies Better by Mixing Physical Theories?
• Jet tagging made easy
• Flows for Flows: Morphing one Dataset into another with Maximum Likelihood Estimation
• Comparison of Affine and Rational Quadratic Spline Coupling and Autoregressive Flows through Robust Statistical Tests
• A Quantum Algorithm for Model-Independent Searches for New Physics
• Anomaly detection with flow-based fast calorimeter simulators
• OneFlow: One-class flow for anomaly detection based on a minimal volume region
• High-dimensional Anomaly Detection with Radiative Return in $e^{+}{e}^{-}$ Collisions
• Weakly-Supervised Anomaly Detection in the Milky Way
• Non-resonant Anomaly Detection with Background Extrapolation
• Exploring the ex-situ components within Gaia DR3
• Quantum density estimation with density matrices: Application to quantum anomaly detection
• Invariant Representation Driven Neural Classifier for Anti-QCD Jet Tagging
• Unsupervised and lightly supervised learning in particle physics
• Simulation-based Anomaly Detection for Multileptons at the LHC
• Generative Invertible Quantum Neural Networks
• Detecting anomaly in vector boson scattering
• Finding Wombling Boundaries in LHC Data with Voronoi and Delaunay Tessellations
• How to GAN Event Unweighting
• TopicFlow: Disentangling quark and gluon jets with normalizing flows
• Signal-background interference for digluon resonances at the Large Hadron Collider
• Unifying Simulation and Inference with Normalizing Flows
• Weakly supervised anomaly detection for resonant new physics in the dijet final state using proton-proton collisions at $\sqrt{s}=13$ TeV with the ATLAS detector
• Exotic diboson $Z^{\prime }$ decays in the U$\mu \nu$SSM
• Fair Normalizing Flows
• New Methods and Datasets for Group Anomaly Detection From Fundamental Physics
• Latent Space Refinement for Deep Generative Models
• Observable Optimization for Precision Theory: Machine Learning Energy Correlators
• Preserving New Physics while Simultaneously Unfolding All Observables
• OmniLearned: A Foundation Model Framework for All Tasks Involving Jet Physics
• Nonparametric Density Estimation from Markov Chains
• Detecting New Physics as Novelty -- Complementarity Matters
• Learning New Physics from an Imperfect Machine
• Event Generation and Density Estimation with Surjective Normalizing Flows
• Graph theory inspired anomaly detection at the LHC
• Creating Simple, Interpretable Anomaly Detectors for New Physics in Jet Substructure
• Reweighting and Analysing Event Generator Systematics by Neural Networks on High-Level Features