Search NASA⌕ Search

SEARCH · Search NASA

Results for “anomaly identification”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Anomaly Identification of Synchronized Voltage Waveform for Situational Awareness of Low Inertia Systems

Inverter-based resources (IBRs) such as photovoltaics (PVs), wind turbines, and battery energy storage systems (BESSs) are widely deployed in low-carbon power systems. However, these resources typically do not provide the inertia needed for grid stability, resulting in a low-inertia power system. IBRs and lack of inertia have been known to cause anomalies such as waveform distortions and wideband oscillations in power systems due to the limited inertia level, leading to increased generation trips and load shedding. Here, to achieve effective anomaly identification, this paper proposes a synchro-waveform-based algorithm utilizing real-time synchronized voltage waveform measurements from waveform measurement units (WMUs). In the proposed method, different physical characteristics, as well as statistical features, are extracted from synchronized voltage waveform measurements to filter anomalies. Then, the anomaly identification approach based on the random forest is developed and deployed into the FNET/GridEye system considering trade-offs among accuracy, computational burden, and deployment cost. Moreover, four WMUs are specially designed and deployed on Kauai Island to receive instantaneous synchronized voltage waveform measurements. To verify the performance of the proposed algorithm, different experiments are carried out with collected field test data. The result demonstrates that the performance of the proposed synchro-waveform-based anomaly categorization algorithm can accurately identify anomalies 95.35% of the time, which has comparable performance among benchmarking algorithms.

Situational awareness↗
DOE OSTI2025

Unsupervised Process Anomaly Detection and Identification Using the Leave-One-Variable-Out Approach

Automated anomaly detection and identification can signal equipment issues and pinpoint causes in large-scale industrial systems. For systems with limited failure history, unsupervised machine learning methods can be utilized as they do not require past failures. This study introduces the leave-one-variable-out (LOVO) model, which masks one variable at a time to predict the others, learning underlying process correlations. Detection performance was assessed with synthetic and experimental data, while identification performance used only synthetic data due to its ability to generate labeled anomaly types. For detection using synthetic data, the LOVO model generally outperformed comparative models; while using experimental data, the comparative methods outperformed the LOVO model. However, the comparative methods required selecting a latent size, and these conclusions pertain to using the optimal size. In practice, it would not be feasible to always select the optimal value, and incorrect selections impacted performance. In contrast, the LOVO model does not require a latent space. For identification using synthetic data, the LOVO model was slightly outperformed in interpretability and repeatability but still demonstrated impressive results. These outcomes suggest that the LOVO model is an effective model and may be more easily implemented without the challenging tuning process of selecting a latent size.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

Harnessing the Power of AI: Status and Expansion of Current Domestic Transport Security Through Flexible Embedded Hardware

As applications of Artificial Intelligence (AI) continue to expand, there are increasing opportunities to leverage applied AI methodologies with mobile transportation focused embedded systems. Current applications of AI in transportation focus on a variety of areas, including fuel efficiency, safety, security, and other broad fields of optimization or detection. To leverage these AI workflows and methodologies in the field, teams must utilize complex embedded systems capable of implementing these AI-enabled algorithms in real-time. In this paper, we will investigate how these algorithms can be integrated into existing technologies leveraging vehicle data - such as the Controller Area Network Transport Security Tracking and Reporting Unit (C-STAR). The C-STAR technology is an embedded platform with onboard computation capable of running next generation algorithms in vehicle systems AI, such as preventative maintenance, driver authentication, and transport security. As deployed in the field, the C-STAR has a limited AI functionality –this paper will directly discuss how a device like C-STAR can be utilized and the advantages of integrating these new technologies. We will open with relevant background information and transportation projects that leverage AI, focusing specifically on those around transport security such as vehicle identification, anomaly detection, and deterrence. We will then extend this into potential opportunities and scaling for AI methodologies using platforms like the C-STAR. Finally, we will speak directly to the challenges of deploying AI-powered workflows, such as computing power needs, bandwidth, hallucinations, and other regulatory considerations.

Cook, Adian [ORNL] (ORCID:0000000160825395)↗
DOE OSTI2025

Real-Time, Adaptive Radiological Anomaly Detection and Isotope Identification Using Non-Negative Matrix Factorization

Spectroscopic anomaly detection and isotope identification algorithms are integral components in nuclear nonproliferation applications such as search operations. The task is especially challenging in the case of mobile detector systems because the observed gamma-ray background changes more than for a static detector system, and a pretrained background model can easily find itself out of domain. The result is that algorithms may exceed their intended false alarm rate or sacrifice detection sensitivity to maintain the desired false alarm rate. Non-negative matrix factorization (NMF) is a powerful tool for spectral anomaly detection and identification, but, like many similar algorithms that rely on data-driven background models, in its conventional implementation, it is unable to update in real time to account for environmental changes that affect the background spectroscopic signature. Here, we have developed a novel NMF-based algorithm that periodically updates its background model to accommodate changing environmental conditions. The adaptive NMF algorithm involves fewer assumptions about its environment, making it more generalizable than existing NMF-based methods while maintaining or exceeding detection performance on simulated and real-world datasets.

Anomaly detection↗
NASA NTRS2008

An Investigation of State-Space Model Fidelity for SSME Data

In previous studies, a variety of unsupervised anomaly detection techniques for anomaly detection were applied to SSME (Space Shuttle Main Engine) data. The observed results indicated that the identification of certain anomalies were specific to the algorithmic method under consideration. This is the reason why one of the follow-on goals of these previous investigations was to build an architecture to support the best capabilities of all algorithms. We appeal to that goal here by investigating a cascade, serial architecture for the best performing and most suitable candidates from previous studies. As a precursor to a formal ROC (Receiver Operating Characteristic) curve analysis for validation of resulting anomaly detection algorithms, our primary focus here is to investigate the model fidelity as measured by variants of the AIC (Akaike Information Criterion) for state-space based models. We show that placing constraints on a state-space model during or after the training of the model introduces a modest level of suboptimality. Furthermore, we compare the fidelity of all candidate models including those embodying the cascade, serial architecture. We make recommendations on the most suitable candidates for application to subsequent anomaly detection studies as measured by AIC-based criteria.

Space Shuttles↗
DOE OSTI2024

DOC-DICAM: Domain Aware One Class Defect Identification in Composite Aerostructure Material

Fiber-reinforced composites are a common material used in the design of aircraft structures due to their good tensile strength and resistance to compression. During the manufacturing process, these structures are thoroughly inspected for flaws and defects to ensure structural integrity during commercial use. Non-destructive testing (NDT) is a collection of inspection methods that allow inspectors to evaluate material without altering it. Due to the high safety standards in aerospace manufacturing, the NDT process is done manually and can be a significant bottleneck in the development workflow. In this paper, we develop an AI-based assistance tool to drastically reduce inspection time. Typical AI workflows require large amounts of annotated data, but defects rarely occur resulting in strong class imbalance. To overcome this, we formulate the problem of defect identification as an anomaly detection task in which our primary focus is learning non-defect characteristics. To do this, we develop a multi-task self-supervised learning framework that embeds problem specific domain knowledge into the deep learning model. We verify our method using fuselage data generated in a production environment. As a result, we show that our method can effectively identify defects and requires minimal training and inference time.

anomaly detection↗
DOE OSTI2026

RADAI: A Large-Scale Realistic Dataset for Radiation Detection Algorithm Development

Open, realistic datasets are essential for developing and benchmarking radiation detection algorithms, yet they remain scarce. The Radiological Anomaly Detection and Identification (RADAI) project was develop to create datasets that meet the training and testing needs for sophisticated radiation detection algorithms. The RADAI dataset is a large-scale synthetic resource that integrates high-fidelity Monte Carlo simulations with realistic urban scenarios to capture both background variability and source signatures. RADAI models construction-material NORM, people and vehicles, urban clutter, and dynamic environmental effects such as cosmic-ray and rain-induced transients, and they provide list-mode detector data with motion and response modeling suitable for algorithm training and evaluation. The RADAI project resulted in three publicly-released complementary datasets together with an online scoring portal for standardized performance assessment and an open software toolkit that supports data access, augmentation, model development, and evaluation. These resources enable reproducible comparisons across methods and promote rigorous studies at the scale required by contemporary machine learning. By grounding algorithm development in realistic, well-documented conditions, RADAI supports progress toward more robust detection, identification, and localization in complex urban environments.

Ghawaly, James M. [Division of Computer Science an↗
DOE OSTI2025

In situ Visible Light and Thermal Imaging Data from a Laser Powder Bed Fusion Additive Manufacturing Process Co-Registered to X-ray Computed Tomography and Fatigue Data

This dataset is comprised of in situ sensing data collected during a laser-based powder bed fusion additive manufacturing process, as well as rasterized scan path information, post-build X-ray computed tomography (XCT), and fatigue test results. A total of 64 cylinders, approximately 15 mm in diameter and 102 mm tall, were printed out of stainless steel 316H on a Colibrium Additive Concept Laser M2 Series 5 machine. Parameters known to produce dense material were used to construct 56 of these cylinders, while the remaining 8 cylinders were printed with relatively high energy density parameters prone to producing keyhole pores. In addition, two spatter generation blocks were constructed upstream of the 64 cylinders such that ejecta produced during the melting of the spatter generators were stochastically seeded onto the 64 cylinders. Based on previous experiments, these spatter particles were theorized to produce stochastic lack-of-fusion pores. During the construction of the build, high-resolution images of reflected light in the visible spectrum were captured both before and after recoating for each print layer. Additionally, temporally integrated thermal imaging in the near infrared spectrum produced integrated sum and max images on a layerwise basis. The multimodal in situ data has been co-registered to the build plate coordinate system, allowing for identification of process anomalies (e.g., spatter particles) apparent in the two sensors. Following construction of the build, the cylinders were subjected to XCT to identify internal flaws, and the resulting data have also been registered to the build plate coordinate system. Finally, 60 of the 64 cylinders were machined into fatigue coupons conforming to ASTM E466 and subsequently subjected to either high- or -low-cycle fatigue testing. The results of the fatigue tests have also been included in the dataset, and the XCT data corresponded to the approximate location of the gauge sections of the machine fatigue specimen geometry.

42 ENGINEERING↗
DOE OSTI2025

Reinforcement Learning for Anomaly Detection in Nuclear Power Plant Operation and Maintenance

In nuclear power plants (NPPs), timely identification of sensor and human errors is critical to ensure safe and efficient plant operations. Anomaly detection models can be employed for this task. However, traditional anomaly detection approaches may have high dependency on labeled datasets and struggle with adaptability in complex, dynamic environments. Reinforcement learning (RL) has demonstrated significant potential in fault diagnosis and anomaly detection; however, its application to anomaly detection in NPPs remains a relatively underexplored research direction. Hence, to address this gap, in this study, we present a novel physics-informed reinforcement learning model, PIRL-AD: Physics-Informed Reinforcement Learning for Anomaly Detection, that integrates domain knowledge from calorimetric equations into the RL framework for enhanced sensor and human error anomaly detection. We evaluate the performance of PIRL-AD against a non-physics informed RL benchmark and a support vector machine (SVM) on data collected from a forced flow loop testbed. Experimental results suggest that PIRL-AD outperforms other baselines on a range of anomalous datasets that include both sensor and human-induced anomalies across key performance metrics, statistically outperforming the RL and SVM benchmarks with respect to geometric mean (respectively, 92.96% vs. 91.06% vs. 83.01%) and F1-score (respectively, 89.23% vs. 86.98% vs. 77.01%). Furthermore, the findings suggest the potential of physics-integrated reinforcement learning models for enhanced anomaly detection performance in NPPs.

Reinforcement learning↗
DOE OSTI2026

SMEFT ATLAS: The landscape beyond the Standard Model

The Standard Model Effective Field Theory (SMEFT) based on the unbroken gauge group SU(3) 𝐶 ⊗SU(2) 𝐿 ⊗U(1) 𝑌 and containing only particles of the Standard Model (SM) has developed in the last decade to a mature field. It is the framework to be used in the energy gap from scales sufficiently higher than the electroweak scale up to the lowest energy scale at which new particles show up. We summarize the present status of this theory with a particular emphasize on its role in the indirect search for new physics (NP). While flavour physics of both quarks and leptons is the main topic of our review, we also discuss electric dipole moments, anomalous magnetic moments (𝑔−2) 𝜇,𝑒 , 𝑍-pole observables, Higgs observables and high-𝑝𝑇 scattering processes within the SMEFT. We group the observables into ten classes and list for each class the most relevant operators and the corresponding renormalization group equations (RGEs). We exhibit the correlations between different classes implied both by the operator mixing and the SU(2) 𝐿 gauge symmetry. Our main goal is to provide an insight into the complicated operator structure of this framework which hopefully will facilitate the identification of valid ultraviolet completions behind possible anomalies observed in future data. Numerous colourful charts, and 85 tables, while representing rather complicated RG evolution from the NP scale down to the electroweak scale, beautify the involved SMEFT landscape. Over 1000 references to the literature underline the importance and the popularity of this field. We discuss both top-down and bottom-up approaches as well as their interplay. This allows us eventually to present an atlas of different landscapes beyond the SM that includes heavy gauge bosons and scalars, vector-like quarks and leptons and leptoquarks.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Coincident learning for beam-based rf station fault identification using phase information at the SLAC linac coherent light source

Anomalies in radio-frequency (rf) stations can result in unplanned downtime and performance degradation in linear accelerators such as SLAC’s Linac Coherent Light Source (LCLS). Detecting these anomalies is challenging due to the complexity of accelerator systems, high data volume, and scarcity of labeled fault data. Prior work identified faults using beam-based detection, combining rf amplitude and beam position monitor data. Due to the simplicity of the rf amplitude data, classical methods are sufficient to identify faults, but the recall is constrained by the low-frequency and asynchronous characteristics of the data. In this work, we leverage high-frequency, time-synchronous rf phase data to enhance anomaly detection in the LCLS accelerator. Due to the complexity of phase data, classical methods fail, and we instead train deep neural networks within the Coincident Anomaly Detection (CoAD) framework. We find that applying CoAD to phase data detects nearly 3 times as many anomalies as when applied to amplitude data, while achieving broader coverage across rf stations. Furthermore, the rich structure of phase data enables us to cluster anomalies into distinct physical categories. Through the integration of auxiliary system status bits, we link clusters to specific fault signatures, providing additional granularity for uncovering the root cause of faults. We also investigate interpretability via Shapley values, confirming that the learned models focus on the most informative regions of the data and providing insight for cases where the model makes mistakes. This work demonstrates that phase-based anomaly detection for rf stations improves both diagnostic coverage and root cause analysis in accelerator systems and that deep neural networks are essential for effective analysis.

Accelerator Physics (physics.acc-ph)↗
DOE OSTI2025

Braxton Marlatt Intern Poster

The Internet of Things (IoT) encompasses a vast network of interconnected devices embedded with software, sensors, and network connectivity, enabling data collection and exchange. While IoT technology revolutionizes various industries, it also introduces significant security challenges. This research focuses on enhancing IoT security through the implementation of Zero Trust Architecture concepts, specifically targeting the Network and Device pillars of the Cybersecurity and Infrastructure Security Agency’s Zero Trust Maturity Model. By generating Codified Attack Surfaces (CAS) using custom Structured Threat Information eXpression bundles, this project aims to provide enhanced visibility into network communications, detect vulnerabilities in device firmware, and improve the overall security posture for IoT devices and networks. The methodology involves defining custom STIX schema and objects, collecting data from intra-IoT traffic, external network traffic, and firmware analysis, and automating the conversion and correlation of this data into STIX bundles. The automated generation of attack surfaces offers comprehensive insights into activity, vulnerabilities, and anomalies within an IoT environment, enabling proactive threat identification and mitigation.

24 - POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2026

Search for resonances decaying to an anomalous jet and a Higgs boson in proton–proton collisions at $\sqrt{s}=13\,\text {Te}\hspace{-.08em}\text {V}$

This paper presents a search for new physics through the process where a massive particle, X, decays into a Higgs boson and a second particle, Y. The Higgs boson subsequently decays into a bottom quark–antiquark pair, which is reconstructed as a single large-radius jet. The decay products of Yare also assumed to produce a single large-radius jet. The identification of the Yparticle is enhanced by computing the anomaly score of its candidate jet using an autoencoder, which measures deviations from typical quark- or gluon-induced jets. This allows a simultaneous search for multiple Ydecay scenarios within a single analysis. In the main benchmark process, Yis a scalar particle that decays into a Wboson pair. Two other scalar Ydecay processes are also considered as benchmarks: decays to a light quark–antiquark pair, and decays to a top quark–antiquark pair. A fourth benchmark process considers Yas a hadronically decaying top quark, arising from the decay of a vector-like quark into a top quark and a Higgs boson. Data recorded by the CMS experiment at a center-of-mass energy of 13 TeV in 2016–2018, corresponding to an integrated luminosity of 138 fb -1 , are analyzed. The search covers Xmasses between 1.4 and 3.0 TeV and Ymasses between 90 and 400 TeV, with all simulated signals produced in the narrow-width approximation. No significant excess above the standard model background expectation is observed. The most stringent upper limits to date are placed on benchmark signal cross sections for various masses of X and Y particles

Hayrapetyan, A. [Yerevan Physics Institute]↗
DOE OSTI2026

A Semi-Supervised Learning Method for the Identification of Bad Exposures in Large Imaging Surveys

As the data volume of astronomical imaging surveys rapidly increases, traditional methods for image anomaly detection, such as visual inspection by human experts, are becoming impractical. We introduce a machine-learning-based approach to detect poor-quality exposures in large imaging surveys, with a focus on the DECam Legacy Survey (DECaLS) in regions of low extinction (i.e., E ( B − V ) < 0.04 ). Our semi-supervised pipeline integrates a vision transformer (ViT), trained via self-supervised learning (SSL), with a k-Nearest Neighbor (kNN) classifier. We train and validate our pipeline using a small set of labeled exposures observed by surveys with the Dark Energy Camera (DECam). A clustering-space analysis of where our pipeline places images labeled in good and bad categories suggests that our approach can efficiently and accurately determine the quality of exposures. Applied to new imaging being reduced for DECaLS Data Release 11, our pipeline identifies 780 problematic exposures, which we subsequently verify through visual inspection. Being highly efficient and adaptable, our method offers a scalable solution for quality control in other large imaging surveys.

Luo, Yufeng (ORCID:0000000246230683)↗
NASA NTRS2010

Ares I-X Ground Diagnostic Prototype

The automation of pre-launch diagnostics for launch vehicles offers three potential benefits: improving safety, reducing cost, and reducing launch delays. The Ares I-X Ground Diagnostic Prototype demonstrated anomaly detection, fault detection, fault isolation, and diagnostics for the Ares I-X first-stage Thrust Vector Control and for the associated ground hydraulics while the vehicle was in the Vehicle Assembly Building at Kennedy Space Center (KSC) and while it was on the launch pad. The prototype combines three existing tools. The first tool, TEAMS (Testability Engineering and Maintenance System), is a model-based tool from Qualtech Systems Inc. for fault isolation and diagnostics. The second tool, SHINE (Spacecraft Health Inference Engine), is a rule-based expert system that was developed at the NASA Jet Propulsion Laboratory. We developed SHINE rules for fault detection and mode identification, and used the outputs of SHINE as inputs to TEAMS. The third tool, IMS (Inductive Monitoring System), is an anomaly detection tool that was developed at NASA Ames Research Center. The three tools were integrated and deployed to KSC, where they were interfaced with live data. This paper describes how the prototype performed during the period of time before the launch, including accuracy and computer resource usage. The paper concludes with some of the lessons that we learned from the experience of developing and deploying the prototype.

Machine Learning↗
NASA NTRS2010

Ares I-X Ground Diagnostic Prototype

Automating prelaunch diagnostics for launch vehicles offers three potential benefits. First, it potentially improves safety by detecting faults that might otherwise have been missed so that they can be corrected before launch. Second, it potentially reduces launch delays by more quickly diagnosing the cause of anomalies that occur during prelaunch processing. Reducing launch delays will be critical to the success of NASA's planned future missions that require in-orbit rendezvous. Third, it potentially reduces costs by reducing both launch delays and the number of people needed to monitor the prelaunch process. NASA is currently developing the Ares I launch vehicle to bring the Orion capsule and its crew of four astronauts to low-earth orbit on their way to the moon. Ares I-X will be the first unmanned test flight of Ares I. It is scheduled to launch on October 27, 2009. The Ares I-X Ground Diagnostic Prototype is a prototype ground diagnostic system that will provide anomaly detection, fault detection, fault isolation, and diagnostics for the Ares I-X first-stage thrust vector control (TVC) and for the associated ground hydraulics while it is in the Vehicle Assembly Building (VAB) at John F. Kennedy Space Center (KSC) and on the launch pad. It will serve as a prototype for a future operational ground diagnostic system for Ares I. The prototype combines three existing diagnostic tools. The first tool, TEAMS (Testability Engineering and Maintenance System), is a model-based tool that is commercially produced by Qualtech Systems, Inc. It uses a qualitative model of failure propagation to perform fault isolation and diagnostics. We adapted an existing TEAMS model of the TVC to use for diagnostics and developed a TEAMS model of the ground hydraulics. The second tool, Spacecraft Health Inference Engine (SHINE), is a rule-based expert system developed at the NASA Jet Propulsion Laboratory. We developed SHINE rules for fault detection and mode identification. The prototype uses the outputs of SHINE as inputs to TEAMS. The third tool, the Inductive Monitoring System (IMS), is an anomaly detection tool developed at NASA Ames Research Center and is currently used to monitor the International Space Station Control Moment Gyroscopes. IMS automatically "learns" a model of historical nominal data in the form of a set of clusters and signals an alarm when new data fails to match this model. IMS offers the potential to detect faults that have not been modeled. The three tools have been integrated and deployed to Hangar AE at KSC where they interface with live data from the Ares I-X vehicle and from the ground hydraulics. The outputs of the tools are displayed on a console in Hangar AE, one of the locations from which the Ares I-X launch will be monitored. In a previous publication, we discussed how we selected the three tools based primarily on their ability to be certified for human spaceflight and described our plans for the prototype. This abstract is due October 23, 2009, and the Ares I-X launch is currently scheduled for October 27, 2009. If this abstract is accepted, then the full paper will describe how the prototype performed before the launch. It will include an analysis of the prototype's accuracy, including false-positive rates, false-negative rates, and receiver operating characteristics (ROC) curves. It will also include a description of the prototype's computational requirements, including CPU usage, main memory usage, and disk usage. If the prototype detects any faults during the prelaunch period then the paper will include a description of those faults. Similarly, if the prototype has any false alarms then the paper will describe them and will attempt to explain their causes. Also, the paper will describe the three tools and how they are used in the prototype. It will include a description of the TEAMS models of the Ares I-X first-stage TVC and associated ground hydraulics and how we adapted the TVC model for use in real-time diagnostics. It will describe the SHINE rules used for fault detection and mode identification and the software architecture that interfaces the various pieces of existing software that are part of the prototype to one another. It will describe how we selected the sensor values and commands that were used to train the IMS model and how we optimized the number of clusters in the IMS model. It will include screen shots of the graphical display that we developed in Java to display the outputs of the three tools. Because Ares I-X data was not yet available to us while we were developing the prototype, we used historical data from the Space Shuttle's Solid Rocket Booster (SRB) TVCs and the associated ground hydraulics to train IMS and to test the entire prototype. Because most of the failure modes that we modeled have never occurred in the Shuttle we inserted simulated failures into the Shuttle data. The Ares I-X first-stage TVC is very similar to the SRB TVC and we expect the data will be very similar. After the launch, we will determine how similar the data actually is and report how any differences in the data affected the diagnostic accuracy of the prototype. Finally, although we did not get the prototype certified, we designed it in a way that it could be certified and wrote a preliminary certification plan. The paper will include a brief summary of how we considered the need for certification in the design of the prototype, how we tested the prototype before deploying it to Hangar AE, and how we would propose to get it certified if it were deployed as an operational system. The paper will conclude with a description of some of the challenges we faced and some of the lessons learned in developing and deploying the prototype.

International Space Station↗
DOE OSTI2026

Wasserstein normalized autoencoder for anomaly detection

A novel anomaly detection algorithm is presented. The Wasserstein normalized autoencoder (WNAE) is a normalized probabilistic model that minimizes the Wasserstein distance between the learned probability distribution—a Boltzmann distribution where the energy is the reconstruction error of the autoencoder (AE)—and the distribution of the training data. This algorithm has been developed and applied to the identification of semivisible jets—conical sprays of visible standard model (SM) particles and invisible dark matter states—with the CMS experiment at the CERN LHC. Trained on jets of particles from simulated SM processes, the WNAE is shown to learn the probability distribution of the input data in a fully unsupervised fashion, such that it effectively identifies new physics jets as anomalies. The model exhibits stable, convergent training and recovers strong classification performance for a wide range of signals against the selected background process, for which a standard AE fails because of outlier reconstruction. In addition, the model improves upon standard normalized autoencoders while remaining fully agnostic to the signal. The WNAE directly tackles the problem of outlier reconstruction, a common failure mode of autoencoders in anomaly detection tasks.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗