Abstract Machine learning-based anomaly detection (AD) methods are promising tools for extending the coverage of searches physics beyond Standard Model (BSM). One class AD that has received significant attention is resonant detection, where BSM assumed to be localized in at least one known variable. While there have been many proposed identify such a signal make use simulated or detected data different ways, not yet study methods’ complementarity. To this end, we address two questions....

Resonant anomaly detection is a promising framework for model-independent searches new particles. Weakly supervised resonant methods compare data with potential signal against template of the Standard Model (SM) background inferred from sideband regions. We propose means to generate this that uses flow-based model create mapping between high-fidelity SM simulations and data. The flow trained in regions region blinded, conditioned on feature (mass) such it can be interpolated into region. To...

We explore the metric space of jets using public collider data from CMS experiment. Starting 2.3/fb 7 TeV proton-proton collisions collected at Large Hadron Collider in 2011, we isolate a sample 1,690,984 central with transverse momentum above 375 GeV. To validate performance detector reconstructing energy flow jets, compare Open Data to corresponding simulated samples for variety jet kinematic and substructure observables. Even without unfolding, find very good agreement track-based...

There is a growing need for machine learning-based anomaly detection strategies to broaden the search Beyond-the-Standard-Model (BSM) physics at Large Hadron Collider (LHC) and elsewhere. The first step of any approach specify observables then use them decide on set anomalous events. One common choice select events that have low probability density. It well-known fact densities are not invariant under coordinate transformations, so sensitivity can depend initial coordinates. broader learning...

We present the first study of anti-isolated Upsilon decays to two muons ($\Upsilon \to \mu^+ \mu^-$) in proton-proton collisions at Large Hadron Collider. Using a machine learning (ML)-based anomaly detection strategy, we "rediscover" $\Upsilon$ 13 TeV CMS Open Data from 2016, despite overwhelming backgrounds. elevate signal significance $6.4 \sigma$ using these methods, starting $1.6 dimuon mass spectrum alone. Moreover, demonstrate improved sensitivity an ML-based estimate multi-feature...

We investigate a method of model-agnostic anomaly detection through studying jets, collimated sprays particles produced in high-energy collisions. train transformer neural network to encode simulated QCD "event space" dijets into low-dimensional "latent representation. optimize the using self-supervised contrastive loss, which encourages preservation known physical symmetries dijets. then binary classifier discriminate BSM resonant dijet signal from background both event space and latent...

A bstract Complete anomaly detection strategies that are both signal sensitive and compatible with background estimation have largely focused on resonant signals. Non-resonant new physics scenarios relatively under-explored may arise from off-shell effects or final states significant missing energy. In this paper, we extend a class of weakly supervised developed for to the non-resonant case. Machine learning models trained reweight, generate, morph background, extrapolated control region....

Many components of data analysis in high-energy physics and beyond require morphing one dataset into another. This is commonly solved via reweighting, but there are advantages preserving weights shifting the points instead. Normalizing flows machine learning models that have shown impressive precision on a variety particle tasks. Naively, normalizing cannot be used for because they knowledge probability density starting dataset; most cases physics, we can generate more examples, do not know...

RR Lyrae variable stars have long been reliable standard candles used to discern structure in the Local Group. With this mind, we present a routine identify groupings containing statistically significant number of variables Milky Way environment. groupings, or substructures, with potential Galactic archaeology applications are found using forest agglomerative, hierarchical clustering trees, whose leaves variables. Each grouping is validated by ensuring that internal proper motions...

We investigate a method of model-agnostic anomaly detection through studying jets, collimated sprays particles produced in high-energy collisions. train transformer neural network to encode simulated QCD "event space" dijets into low-dimensional "latent representation. optimize the using self-supervised contrastive loss, which encourages preservation known physical symmetries dijets. then binary classifier discriminate BSM resonant dijet signal from background both event space and latent...

There are many cases in collider physics and elsewhere where a calibration dataset is used to predict the known / or noise of target region phase space. This usually cannot be out-of-the-box but must tweaked, often with conditional importance weights, maximally realistic. Using resonant anomaly detection as an example, we compare number alternative approaches based on transporting events normalizing flows instead reweighting them. We find that accuracy morphed depends degree which transport...

Resonant anomaly detection is a promising framework for model-independent searches new particles. Weakly supervised resonant methods compare data with potential signal against template of the Standard Model (SM) background inferred from sideband regions. We propose means to generate this that uses flow-based model create mapping between high-fidelity SM simulations and data. The flow trained in regions region blinded, conditioned on feature (mass) such it can be interpolated into region. To...

Determining the form of Higgs potential is one most exciting challenges modern particle physics. pair production directly probes self-coupling and should be observed in near future at High-Luminosity LHC. We explore how to improve sensitivity physics beyond Standard Model through per-event kinematics for di-Higgs events. In particular, we employ machine learning simulation-based inference estimate likelihood ratios gauge gains from including this kinematic information. terms Effective Field...

Determining the form of Higgs potential is one most exciting challenges modern particle physics. pair production directly probes self-coupling and should be observed in near future at High-Luminosity LHC. We explore how to improve sensitivity physics beyond Standard Model through per-event kinematics for di-Higgs events. In particular, we employ machine learning simulation-based inference estimate likelihood ratios gauge gains from including this kinematic information. terms Effective Field...

Many components of data analysis in high energy physics and beyond require morphing one dataset into another. This is commonly solved via reweighting, but there are many advantages preserving weights shifting the points instead. Normalizing flows machine learning models with impressive precision on a variety particle tasks. Naively, normalizing cannot be used for because they knowledge probability density starting dataset. In most cases physics, we can generate more examples, do not know...

Non-Standard-Model parity violation may be occurring in LHC collisions. Any such would go unseen, however, as searches are for it not currently performed. One barrier to is the lack of model-independent methods sensitive all its forms. We remove this by demonstrating an effective and way search parity-violating physics at LHC. The method data-driven makes no reference any particular model. Instead, inspects data construct parity-odd event variables (using machine learning tools), uses these...