Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “data classification”

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 523 records · Page 29
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Wintertime synoptic patterns of midlatitude boundary layer clouds over the western North Atlantic: Climatology and insights from in-situ ACTIVATE observations

The winter synoptic evolution of the western North Atlantic and its influence on the atmospheric boundary layer is described by means of a regime classification based on Self Organizing Maps applied to 12 year of data (2009-2020). The regimes are classified into categories according to daily 600-hPa geopotential height: dominant ridge, trough to ridge eastward transition (trough-ridge), dominant trough, and ridge to trough eastward transition (ridge-trough). A fifth synoptic regime resembles the winter climatological mean. Coherent changes in sea-level pressure and large-scale winds are in concert with the synoptic regimes: 1) the ridge regime is associated with a well-developed anticyclone; 2) the trough-ridge gives rise to a low pressure center over the ocean, ascents, and northerly winds over the coastal zone; 3) trough is associated with the eastward displacement of a cyclone, coastal subsidence, and northerly winds, all representative characteristics of cold-air outbreaks; 4) the ridge-trough regime features the development of an anticyclone and weak coastal winds. Low clouds are characteristic of the trough regime, with both trough and trough-ridge featuring synoptic maxima in cloud droplet number concentration (N d ). The N d increase is primarily observed near the coast, concomitant with strong surface heat fluxes exceeding by more than 400 W m -2 compared to fluxes further east. Five consecutive days of aircraft observations collected during the ACTIVATE campaign corroborates the climatological characterization, confirming the occurrence of high N d for days identified as trough. This study emphasizes the role of boundary-layer dynamics and aerosol activation and their roles in modulating cloud microphysics.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

An end-to-end trainable hybrid classical-quantum classifier

Abstract We introduce a hybrid model combining a quantum-inspired tensor network and a variational quantum circuit to perform supervised learning tasks. This architecture allows for the classical and quantum parts of the model to be trained simultaneously, providing an end-to-end training framework. We show that compared to the principal component analysis, a tensor network based on the matrix product state with low bond dimensions performs better as a feature extractor for the input data of the variational quantum circuit in the binary and ternary classification of MNIST and Fashion-MNIST datasets. The architecture is highly adaptable and the classical-quantum boundary can be adjusted according to the availability of the quantum resource by exploiting the correspondence between tensor networks and quantum circuits.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Dark Energy Survey Supernova 5YR Data Release

This data set was generated by the Dark Energy Survey (DES) team, and was used for the cosmological analysis in the DES key supernova cosmology paper (arXiv:2401.02929). When using this data please cite both the data release paper (Sanchez et al. 2024, ApJ, arXiv:2406.05046) and the DES key paper (Dark Energy Survey Collaboration 2024, ApJL, arXiv:2401.02929). At DES-SN5YR Github Repo (https://github.com/des-science/DES-SN5YR), we provide a comprehensive list of references for the different data products (SALT3 light-curve model, calibration, DCR corrections, SN classification, SN distances and uncertainties covariance matrix). If you have questions, post an issue on the DES-SN5YR Github repository and/or get in touch with the contact people indicated on the Github repository.

Contractor, Fermilab [Fermilab]↗
DOE OSTI2025

Mondo: integrating disease terminology across communities

Precision medicine aims to enhance diagnosis, treatment, and prognosis by integrating multimodal data at the point of care. However, challenges arise due to the vast number of diseases, differing methods of classification, and conflicting terminological coding systems and practices used to represent molecular definitions of disease. This lack of interoperability artificially constrains the potential for diagnosis, clinical decision support, care outcome analysis, as well as data linkage across research domains to support the development or repurposing of therapeutics. There is a clear and pressing need for a unified system for managing disease entities⁠—including identifiers, synonyms, and definitions. To address these issues, we created the Mondo disease ontology—a community-driven, open-source, unified disease classification system that harmonizes diverse terminologies into a consistent, computable framework. Mondo integrates key medical and biomedical terminologies, including Online Mendelian Inheritance in Man (OMIM), Orphanet, Medical Subject Headings (MeSH), National Cancer Institute Thesaurus (NCIt), and more, to provide a comprehensive and accurate representation of disease concepts with fully provenanced and attributed links back to the sources. Mondo can be used as the handle for curation of gene–disease associations utilized in diagnostic applications, research applications such as computational phenotyping, and in clinical coding systems in clinical decision support by pointing the clinician to the numerous knowledge resources linked to the Mondo identifier. Mondo's community-centric approach, stewarded by the Monarch Initiative's expertise in ontologies, ensures that the ontology remains adaptable to the evolving needs of biomedical research and clinical communities, as well as the knowledge providers.

biomedical informatics↗
DOE OSTI2022

Ensemble voting-based fault classification and location identification for a distribution system with microgrids using smart meter measurements

This study presents an ensemble learning approach for fault classification and location identification in a smart distribution network containing photovoltaics (PV)-based microgrid. Lack of available data points and the unbalanced nature of the distribution system make fault handling a challenging task for utilities. The proposed method uses event-driven voltage data from smart meters to classify and locate faults. The ensemble voting classifier is composed of three base learners; random forest, k-nearest neighbours, and artificial neural network. The fault location (FL) task has been formulated as a classification problem where the fault type is classified in the first step and based on the fault type, the faulty bus is identified. The method is tested on IEEE-123 bus system modified with added PV-based microgrid along with dynamic loading conditions and varying fault resistances from 0 to 20 Ω for both unbalanced and balanced fault types. A further sensitivity analysis has been done to test the robustness of the proposed method under various noise levels and data loss errors in the smart meter measurements. The ensemble method shows improved performance and robustness compared to some previously proposed FL methods. Finally, the proposed method has been experimentally validated on a real-time simulation-based testbed using a state-of-the-art digital real-time simulator, industry standard DNP3 communication protocol and a cpu-based control centre running the FL algorithm.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2020

Macro-physical Properties of Shallow Cumulus from Integrated ARM Observations (Final Report)

Fair-weather shallow cumuli (ShCu) play an important role in many climate-related processes. Irregular geometry of ShCu and their strong temporal and spatial variability make it challenging to observe ShCu holistically and to represent them correctly in climate models. To improve ShCu parameterizations, information on both vertically and horizontally resolved cloud properties is required. Commonly, the vertically resolved cloud properties are provided by zenith pointing lidar-radar observations with a very narrow field of view (FOV). Thus, these “pencil-beam” properties may not be representative of a larger surrounding area. Limited number of areal-averaged cloud properties, such as fractional sky cover (FSC), are offered typically by wide-FOV observations. The main goal of our project was to integrate advantages of the narrow-FOV (vertical structure of clouds) and wide-FOV (spatial arrangement of clouds) observations for an improved characterization of single-layer ShCu observed at the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) site for an 18-yr period (2000-2017). There are four major accomplishments of our project, First, an updated operational cloud classification for days with ShCu has been suggested and evaluated through a detailed comparison with the manually curated records. Our classification extends successfully the latest ARM cloud type Value Added Product (VAP) based on the Active Remote Sensing of Clouds (ARSCL) cloud product by incorporating both cloud fraction (CF) provided by narrow-FOV ceilometer data and FSC from wide-FOV images offered by a Total Sky Imager (TSI). Moreover, our classification allows one to identify impact of instrumentation changes at the SGP site, namely the transition to KAZRARSCL with the updated cloud radar, on the identification of periods with single-layer ShCu. Second, a new approach that resolves cloud area distributions for a given region (up to 4x4 km 2 ) has been suggested and cloud equivalent diameters (CEDs) have been estimated for the first time. These estimations have been performed over a wide range of cloud sizes (about 0.01–3.5 km) with high temporal resolution (30s) using wide-FOV TSI images and cloud base height (CBH) provided by complementary narrow-FOV lidar measurements. Our simple and computationally inexpensive approach offers a previously unavailable dataset for process studies in the convective boundary layer and evaluation of ShCu parameterizations in cloud-resolving models. Third, a long-term integrated record of ShCu macrophysical properties has been developed. The developed record represents the longest available compilation of events with ShCu and includes (i) a novel visualization of the spatial variability in cloud cover both along- and across-wind directions, (ii) updated estimates of narrow-FOV CF and wide-FOV FSC, (iii) updated narrow-FOV CBH, and (iv) complementary data, such as wind speed and direction from the 915-MHz Radar Wind Profiler (RWP) data. The developed record has been used successfully to assess conventional observational estimates of cloud cover and their sensitivity to the following two factors: (i) instrument-dependent cloud detection and data merging criteria and (ii) FOV configuration. Fourth, co-variability of the ShCu macrophysical properties and environmental parameters has been analyzed for a 3-yr period (2016-2018). Our initial analysis includes diurnal changes of FSCs obtained for clouds with small, moderate and large CEDs and several environmental parameters, such as lifted condensation level (LCL) and mixed layer height (zi). Preliminary results of our analysis suggest that the horizontal extent of ShCu is controlled substantially by the sign and magnitude of difference between these two parameters (zi-LCL): the CED tends to grow with increase of this difference (zi exceeds LCL). We have initiated relationships between the ShCu and key atmospheric parameters that control both the development and evolution of ShCu using our new data product, which combines effectively the advantages of narrow-FOV data offered by zenith pointing cloud radars and lidars and wide-FOV TSI images. While the latest instrumentation at the ARM sites may address these challenging relationships in the future, we believe that the historical ARM data at the SGP site has not yet been fully utilized. Overall, our data product can be used by researchers working on a wide range of climate-related projects. These projects may include (i) a comprehensive evaluation of outputs from the Large-Eddy Simulation (LES) and single-column models for their future improvement, (ii) the representativeness of “short-period” results obtained from the previous model and observational studies and (iii) the planning of future field campaigns with focus on improved understanding of the diurnal cycle of cumulus convection.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2023

Uncovering acoustic signatures of pore formation in laser powder bed fusion

Abstract We present a machine learning workflow to discover signatures in acoustic measurements that can be utilized to create a low-dimensional model to accurately predict the location of keyhole pores formed during additive manufacturing processes. Acoustic measurements were sampled at 100 kHz during single-layer laser powder bed fusion (LPBF) experiments, and spatio-temporal registration of pore locations was obtained from post-build radiography. Power spectral density (PSD) estimates of the acoustic data were then decomposed using non-negative matrix factorization with custom $$\varvec{k}$$ k -means clustering (NMF $$\varvec{k}$$ k ) to learn the underlying spectral patterns associated with pore formation. NMF $$\varvec{k}$$ k returned a library of basis signals and matching coefficients to blindly construct a feature space based on the PSD estimates in an optimized fashion. Moreover, the NMF $$\varvec{k}$$ k decomposition led to the development of computationally inexpensive machine learning models which are capable of quickly and accurately identifying pore formation with classification accuracy of supervised and unsupervised label learning greater than 95% and 90%, respectively. The intrinsic data compression of NMF k , the relatively light computational cost of the machine learning workflow, and the high classification accuracy makes the proposed workflow an attractive candidate for edge computing toward in-situ keyhole pore prediction in LPBF.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Analysis and Benchmarking of feature reduction for classification under computational constraints

Abstract Machine learning is most often expensive in terms of computational and memory costs due to training with large volumes of data. Current computational limitations of many computing systems motivate us to investigate practical approaches, such as feature selection and reduction, to reduce the time and memory costs while not sacrificing the accuracy of classification algorithms. In this work, we carefully review, analyze, and identify the feature reduction methods that have low costs/overheads in terms of time and memory. Then, we evaluate the identified reduction methods in terms of their impact on the accuracy, precision, time, and memory costs of traditional classification algorithms. Specifically, we focus on the least resource intensive feature reduction methods that are available in Scikit-Learn library. Since our goal is to identify the best performing low-cost reduction methods, we do not consider complex expensive reduction algorithms in this study. In our evaluation, we find that at quadratic-scale feature reduction, the classification algorithms achieve the best trade-off among competitive performance metrics. Results show that the overall training times are reduced 61%, the model sizes are reduced 6×, and accuracy scores increase 25% compared to the baselines on average with quadratic scale reduction.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Mobile Sensing for Wind Field Estimation in Wind Farms: Preprint

This paper introduces a novel approach for estimating the wind field over an entire wind farm using a mobile sensor to collect limited amounts of data. The proposed method estimates the boundary conditions of a simplified turbine wake model by computing the model sensitivity matrix and using a recursive least-squares algorithm to recover the model parameters from the wind field measurements. To address the fact that it is not practical to take measurements across the entire wind farm, the proposed method classifies each area on the map based on its sensitivity to parameter variations. This classification is then used to generate a suitable path for a mobile sensor, which is charged with collecting data for the recursive least-squares algorithm. The proposed framework can successfully estimate the model boundary conditions using just the measurements collected along the path of the mobile sensor. This preliminary result paves the way for using real-time wind field estimates for the coordinated control of all the turbines within a wind farm.

estimation↗
DOE OSTI2024

Topological Interpretability for Deep Learning

With the growing adoption of AI-based systems across everyday life, the need to understand their decision-making mechanisms is correspondingly increasing. The level at which we can trust the statistical inferences made from AI-based decision systems is an increasing concern, especially in high-risk systems such as criminal justice or medical diagnosis, where incorrect inferences may have tragic consequences. Despite their successes in providing solutions to problems involving real-world data, deep learning (DL) models cannot quantify the certainty of their predictions. These models are frequently quite confident, even when their solutions are incorrect. This work presents a method to infer prominent features in two DL classification models trained on clinical and non-clinical text by employing techniques from topological and geometric data analysis. We create a graph of a model's feature space and cluster the inputs into the graph's vertices by the similarity of features and prediction statistics. We then extract subgraphs demonstrating high-predictive accuracy for a given label. These subgraphs contain a wealth of information about features that the DL model has recognized as relevant to its decisions. We infer these features for a given label using a distance metric between probability measures, and demonstrate the stability of our method compared to the LIME and SHAP interpretability methods. This work establishes that we may gain insights into the decision mechanism of a DL model. This method allows us to ascertain if the model is making its decisions based on information germane to the problem or identifies extraneous patterns within the data.

Spannaus, Adam↗
DOE OSTI2026

Improved Spectral Classification of Local K/M Dwarfs in SDSS Surveys. I. Chemodynamic Validation and Metallicity Calibration

An examination of 109,276 spectra of low-mass stars in the Sloan Digital Sky Survey (SDSS) data archive, collected pre-2010, provides a broad collection of K/M (sub)dwarfs tracing the local (d ≲ 250 pc) population of the thin-disk, thick-disk, and halo, based on 3D kinematics. These populations have distinct metallicities and kinematics, which should be reflected in lower-mass members. One complication is measuring metallicities of M dwarfs from low-resolution spectroscopy or photometry alone remains a challenge, even with the availability of Gaia data. To better characterize physical parameters of low-mass stars observed by SDSS, we define a set of 536 improved, empirical K/M dwarf classification templates to more accurately determine spectral subtypes (K5.5–M8.5) and morphology classes (MCs) (0.5–12.5). We select the best-fit template to every archival SDSS spectrum in the range 5000–8000 Å based on minimum χ 2 value. We then use secondary metallicity estimators to calibrate the most likely [Fe/H] values corresponding to the assigned MC. We confirm that the most metal-rich K/M dwarfs have thin-disk kinematics, and we observe the known effect of Galactic radial metallicity migration, which we confirm is strongly imprinted in the chemodynamics of the local disk M dwarfs. Confirmation of these chemodynamic trends validates the precision of our improved templates for estimating [Fe/H] for M (sub)dwarfs, notably at the low-metallicity end (–3 < [Fe/H] < –1). Substructure in the chemodynamics of the most metal-poor K/M subdwarfs begins to distinguish between local Gaia-Enceladus and local in situ halo objects. This methodology opens the door to use the ubiquitous, local K/M (sub)dwarfs to trace local Galactic chemodynamics with the highest possible resolution.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Automating Bug Report Classification with Few Shot Learning

Orthogonal defect classification (ODC) is a method used to categorize software defects, providing valuable insights into the development process. This study focuses on automating the classification of software bug reports into different ODC defect types using few shot learning, a machine learning approach that requires minimal labeled data. Previous research has manually classified bug reports or used traditional machine learning algorithms like linear support vector machine, achieving limited success. Our approach uses few shot learning to improve classification accuracy and efficiency. The results show a harmonic mean of recall and precision (i.e., the F1 score) of around 0.6 which is a performance improvement over previous methods. The results highlight the potential benefit of few shot learning techniques and their application in enhancing the safety and reliability of nuclear digital instrumentation and control (DI&C) systems. Future work will explore incorporating advanced techniques to supplement the model's training data and achieve better results.

42 - ENGINEERING↗
DOE OSTI2024

Deep Learning for Fish Identification from Sonar Data (CRADA 481 Final Report)

In eastern regions of the United States, the American eel is a species of management and regulatory concern because of significant population declines, despite the species’ previous abundance in all tributaries of rivers flowing into the Atlantic Ocean. The American eel is also a candidate for listing under the U.S. Endangered Species Act. While hydropower construction and operation are only one of several factors contributing to this population decline, such a listing could impose additional regulatory challenges for a large number of hydropower projects. In this CRADA project, we improved technologies for identifying migrating eels with the goal of reducing the cost and time required for future American eel hydropower impact assessment and mitigation studies, while maintaining accuracy. We built on results from a previous FOA project (FOA# DE-FOA-0001662), led by the Electric Power Research Institute (EPRI), which developed a highly accurate, deep-learning method for identifying migrating eels from imaging sonar data. The current study aimed to further optimize this deep-learning model, originally designed for image classification, and to develop an object detection software capable of identifying fish from sonar videos in real time, enabling the detection of events like fish migrations and specific species, such as the American eel, at hydropower dams. The data conversion algorithms were packaged as software with a graphical user interface, and the software is evaluated by external collaborators. We focused on the American eel in this project and explored the transferability of the developed deep learning models to the sea lamprey, given the similar body shape and swimming behavior between the two species.

13 HYDRO ENERGY↗
DOE OSTI2020

Sampling Subjective Polygons for Patch-Based Deep Learning Land-Use Classification in Satellite Images

Model generalization remains a key challenge in the analysis of large amounts of heterogeneous satellite image data. One major limiting factor in developing generalizable models, in the context of supervised learning, is the lack of high quality training datasets. A model's capacity to perform well on new data is often inhibited by imbalance and bias in the data that was used for training. This is especially a problem when using convolutional neural networks to classify urban land-use in satellite images. Notable dataset imbalance issues in this application include land-use type imbalance and image scene imbalance. To begin understanding these dataset imbalance problems in more detail, we develop and test a number of sampling methods for generating training image datasets from subjective training polygons for urban land-use classification. We investigate sampling at different point densities as a means to reduce content repetition and therefore content imbalance and bias in the training image dataset.

Arndt, Jacob↗
DOE OSTI2021

ResNet and CycleGAN for pulse shape discrimination of He-4 detector pulses: Recovering pulses conventional algorithms fail to label unanimously

Pulse shape discrimination (PSD) capable detectors, such as He-4, that respond to neutron and gamma-ray 7 interactions have a threshold deposited energy value below which n/γ discrimination vanishes when using 8 conventional PSD algorithms. Recent attempts in applying supervised learning based artificial neural 9 networks for PSD use the pulses in the separated regions to train the networks so they can be used to classify 10 another set of separated pulses. In doing so, pulses previously indistinguishable are not recovered for 11 classification, which would have increased the number of neutron and gamma-ray pulses that could be used 12 for further analysis. Assuming the reason why conventional PSD algorithms have unseparated regions is 13 because the parameter space of the algorithms fail to capture the intrinsic (but subtle) distinguishing 14 behavior of some of the neutron and gamma-ray pulses, a cycle-consistent generative adversarial network 15 (CycleGAN) was trained to amplify those differences and extract well separated neutron and gamma-ray 16 clusters. Results show that, once the network is trained with pulses from separated and unseparated regions, 17 it was able to transform the pulses in the unseparated region to improve the PSD. Subsequent n/γ 18 classification was performed using deep residual network (ResNet) that takes pulses with 512 data points 19 as an input. Two different ResNets were explored – simple ResNet and modified ResNet which takes 20 segmented pulse inputs in the first layer and the corresponding time axis values in the last hidden layer. 21 The later approach enables the network to extract time correlated pulse features to enhance its ability to 22 capture the pulse behaviors relevant for PSD. Although it achieves slightly lower accuracy, 99.41% versus 23 99.89%, based on simply counting the number of correct n/γ labels assigned, compared to the simple 24 ResNet, the modified ResNets architecture was able to decreases the cross-entropy loss function by half, 25 which implies that the correct n/γ labels assigned are less likely to be accidental. PSD parameter 26 distributions based on n/γ classification by ResNet before and after transforming unseparated pulses using 27 CycleGAN show that by enhancing the separation between neutrons and gamma-rays, the transformation 28 helps improve the performance of classifier networks that are trained using labeled dataset. The 29 enhancement of neutron and gamma-ray separation by the CycleGAN increased the PSD figure of merit 30 (FOM) by up to 70% in some regions. Here, the results show that, if a given detector achieves clear separation 31 between neutron and gamma-ray pulses in any energy region, such neural network approaches can help 32 lower the energy threshold for the separation and increasing the number of neutron and gamma-ray pulses 33 that can be used for further analysis.

4He↗
DOE OSTI2022

J-PLUS: Support vector machine applied to STAR-GALAXY-QSO classification

Context. In modern astronomy, machine learning has proved to be efficient and effective in mining big data from the newest telescopes. Aims. In this study, we construct a supervised machine-learning algorithm to classify the objects in the Javalambre Photometric Local Universe Survey first data release (J-PLUS DR1). Methods. The sample set is featured with 12-waveband photometry and labeled with spectrum-based catalogs, including Sloan Digital Sky Survey spectroscopic data, the Large Sky Area Multi-Object Fiber Spectroscopic Telescope, and VERONCAT – the Veron Catalog of Quasars & AGN. The performance of the classifier is presented with the applications of blind test validations based on RAdial Velocity Extension, the Kepler Input Catalog, the Two Micron All Sky Survey Redshift Survey, and the UV-bright Quasar Survey. A new algorithm was applied to constrain the potential extrapolation that could decrease the performance of the machine-learning classifier. Results. The accuracies of the classifier are 96.5% in the blind test and 97.0% in training cross-validation. The F1-scores for each class are presented to show the balance between the precision and the recall of the classifier. We also discuss different methods to constrain the potential extrapolation.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

A Centralized AI Lakehouse Framework for Brain Tumor MRI Classification and Segmentation, University KPI Forecasting, and Water Potability Prediction

In many university and healthcare projects, models are built for very different data types such as tables, institutional time series, and medical images, but they are deployed as separate applications. In this work, that separation made testing and maintenance difficult because each module had its own pipeline and runtime requirements. This paper presents an integrated AI lakehouse-style implementation that runs three model pipelines inside one containerized backend. For medical imaging, we used MRI datasets from IEEE DataPort: a four-class classification set with 7012 images (5708 train/1304 test) and a segmentation set with 3063 image–mask pairs. The classification model (ResNet50 transfer learning) is evaluated using a proper train–validation–test protocol across multiple splits (80/10/10, 70/10/20, 60/10/30, and 10/30/60), achieving a test accuracy of 99.00% under the standard 80/10/10 split. Additionally, a patient-level evaluation is conducted using an external glioma dataset to provide a more realistic assessment without data leakage. The segmentation model (DeepLabV3-ResNet50) achieved 83.09% validation mIoU and 88.79% Dice score. For university KPI forecasting, we used annual IPEDS and NSF HERD data from 2010 to 2023 for three universities (BSU, EOU, and UAB). To examine the effect of preprocessing on forecasting performance, two case studies are conducted. In the first case, linear interpolation is applied to generate semester-level data. In the second case, the original annual data is used directly without interpolation. Random Forest regression and ARIMA models are evaluated using MAE, RMSE, MAPE, and R 2 . The results showed that interpolation improved apparent forecasting performance due to smoothing, while evaluation on the original annual data provided a more realistic assessment of model behavior. To further validate the framework on a larger dataset, an additional case study is conducted using a student dropout dataset. For water potability, we trained and compared multiple tabular classifiers on a large dataset (1,048,575 samples). A Random Forest model (100 trees, max depth 10) achieved 85.86% test accuracy and high recall for unsafe samples (0.8447). All modules are served via FastAPI and deployed together using Docker, with workflow automation routing requests to the correct endpoint. System-level benchmarking indicates that the backend maintains stable throughput and latency under concurrent requests.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Masked Particle Modeling on Sets: Towards Self-Supervised High Energy Physics Foundation Models

Abstract We propose masked particle modeling (MPM) as a self-supervised method for learning generic, transferable, and reusable representations on unordered sets of inputs for use in high energy physics (HEP) scientific data. This work provides a novel scheme to perform masked modeling based pre-training to learn permutation invariant functions on sets. More generally, this work provides a step towards building large foundation models for HEP that can be generically pre-trained with self-supervised learning and later fine-tuned for a variety of down-stream tasks. In MPM, particles in a set are masked and the training objective is to recover their identity, as defined by a discretized token representation of a pre-trained vector quantized variational autoencoder. We study the efficacy of the method in samples of high energy jets at collider physics experiments, including studies on the impact of discretization, permutation invariance, and ordering. We also study the fine-tuning capability of the model, showing that it can be adapted to tasks such as supervised and weakly supervised jet classification, and that the model can transfer efficiently with small fine-tuning data sets to new classes and new data domains.

Heinrich, Lukas (ORCID:0000000240487584)↗