Creating Simple, Interpretable Anomaly Detectors for New Physics in Jet Substructure

Abstract

Anomaly detection with convolutional autoencoders is a popular method to search for new physics in a model-agnostic manner. These techniques are powerful, but they are still a "black box," since we do not know what high-level physical observables determine how anomalous an event is. To address this, we adapt a recently proposed technique by Faucett et al., which maps out the physical observables learned by a neural network classifier, to the case of anomaly detection. We propose two different strategies that use a small number of high-level observables to mimic the decisions made by the autoencoder on background events, one designed to directly learn the output of the autoencoder, and the other designed to learn the difference between the autoencoder's outputs on a pair of events. Despite the underlying differences in their approach, we find that both strategies have similar ordering performance as the autoencoder and independently use the same six high-level observables. From there, we compare the performance of these networks as anomaly detectors. We find that both strategies perform similarly to the autoencoder across a variety of signals, giving a nontrivial demonstration that learning to order background events transfers to ordering a variety of signal events.

arXiv comment17 pages, 8 figures; Corrections made for p_T, m calculations; analysis rerun with conclusions largely unchanged. Published version

References in corpus (61)

• The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations
• The anti-k_t jet clustering algorithm
• An Introduction to PYTHIA 8.2
• FastJet user manual
• Knowledge Distillation: A Survey
• DELPHES 3, A modular framework for fast simulation of a generic collider experiment
• Dispelling the N^3 myth for the Kt jet-finder
• Searching for Exotic Particles in High-Energy Physics with Deep Learning
• Energy Correlation Functions for Jet Substructure
• Anomaly Detection with Density Estimation
• Principles of general final-state resummation and automated implementation
• Searching for New Physics with Deep Autoencoders
• Anomaly Detection for Resonant New Physics with Machine Learning
• Power Counting to Better Jet Observables
• Gaining (Mutual) Information about Quark/Gluon Discrimination
• Learning New Physics from a Machine
• New Angles on Energy Correlation Functions
• Systematics of quark/gluon tagging
• QCD or What?
• A generic anti-QCD jet tagger
• Simulation Assisted Likelihood-free Anomaly Detection
• Novelty Detection Meets Collider Physics
• What is the Machine Learning?
• Adversarially-trained autoencoders for robust unsupervised new physics searches
• Energy flow polynomials: A complete linear basis for jet substructure
• Dijet resonance search with weak supervision using $\sqrt{s}=13$ TeV $pp$ collisions in the ATLAS detector
• Pulling Out All the Tops with Computer Vision and Deep Learning
• Guiding New Physics Searches with Unsupervised Learning
• Variational Autoencoders for New Physics Mining at the Large Hadron Collider
• The Dark Machines Anomaly Score Challenge: Benchmark Data and Model Independent Event Classification for the Large Hadron Collider
• Uncovering latent jet substructure
• Autoencoders for unsupervised anomaly detection in high energy physics
• Tag N' Train: A Technique to Train Improved Classifiers on Unlabeled Data
• Anomaly detection with Convolutional Graph Neural Networks
• Jet Substructure Without Trees
• Finding New Physics without learning about it: Anomaly Detection as a tool for Searches at Colliders
• Better Latent Spaces for Better Autoencoders
• Simulation-Assisted Decorrelation for Resonant Anomaly Detection
• Autoencoders on FPGAs for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider
• Search for a Higgs boson in the decay channel H to ZZ(*) to q qbar l-l+ in pp collisions at sqrt(s) = 7 TeV
• Bump Hunting in Latent Space
• Rare and Different: Anomaly Scores from a combination of likelihood and out-of-distribution models to detect new physics at the LHC
• Challenges for Unsupervised Anomaly Detection in Particle Physics
• Learning the latent structure of collider events
• Quasi Anomalous Knowledge: Searching for new physics with embedded knowledge
• Mapping Machine-Learned Physics into a Human-Readable Space
• Unsupervised clustering for collider physics
• Online-compatible Unsupervised Non-resonant Anomaly Detection
• Mass Unspecific Supervised Tagging (MUST) for boosted jets
• Deep Set Auto Encoders for Anomaly Detection in Particle Physics
• Transferability of Deep Learning Models in Searches for New Physics at Colliders
• Use of a Generalized Energy Mover's Distance in the Search for Rare Phenomena at Colliders
• Unsupervised Outlier Detection in Heavy-Ion Collisions
• Anomalous Jet Identification via Sequence Modeling
• Topological Obstructions to Autoencoding
• Learning to Identify Electrons
• A Data-Directed Paradigm for BSM searches: the bump-hunting example
• Disentangling a Deep Learned Volume Formula
• Identifying the relevant dependencies of the neural network response on characteristics of the input space
• Learning to Isolate Muons
• High-dimensional Anomaly Detection with Radiative Return in $e^{+}{e}^{-}$ Collisions