Results for “Benchmark dataset” — Page 19

DOE OSTI2020

Development of Gamma Background Radiation Digital Twin with Machine Learning Algorithms: Application of Unsupervised Machine Learning to Detection of Anomalies and Nuisances in Gamma Background Radiation Environmental Screening Data

Environmental screening of gamma radiation consists of detecting weak nuisance and anomaly signal in the presence of strong and highly varying background. In a typical scenario, a mobile detector-spectrometer continuously measures gamma radiation spectra in short, e.g., one-second, signal acquisition intervals. The measurement data is a 2D matrix, where one dimension is gamma ray energy, and the other dimension is the number of measurements or total time. In principle, gamma radiation sources can be detected and identified from the measured data by their unique spectral lines. Detecting sources from data measured in a search scenario is difficult due to the highly varying background because of naturally occurring radioactive material (NORM), and low signal-to-noise ratio (S/N) of spectral signal measured during one-second acquisition intervals. The objective of this work is to explore unsupervised machine learning (ML) algorithms for development of a digital twin of gamma radiation background, and for detection and identification of weak nuisances and anomalies events in the presence of highly fluctuating background. In one segment of work, we developed a gamma background estimation model using a Longshort term memory (LSTM) network for one-step CPS time series prediction. The LSTM model was validated with two data sets of measurements from two independent NaI detectors positioned on a mobile platform. The data sets contained background radiation only and no orphan isotope sources. The LSTM model was constructed and tested using data from one of the detectors. Performance of the LSTM model was validate through one-step prediction of CPS time series of another NaI detector without re-training. This approach allows to create a digital twin for nuclear background estimation. Using LSTM, it could be possible to detect a source through subtraction of the estimated counts from the measured background. In another segment of work, we investigated detection of gamma emitting sources in the presence of complex background using unsupervised machine learning. Spectral lines of isotopes are difficult to observe in one-second measurements. Averaging over the entire measurement campaign data set reveals spectral lines of most common background isotopes. Spectral lines of orphan sources, which might appear only in a few measurements during the campaign, will be washed out if averaging is performed over the entire measurement data set. The approach we have explored consists of extracting one-second measurements containing weak spectral features through data clustering. Averaging one-second spectra in a cluster should reveal the presence of anomaly sources. We created two ML models using K-means clustering and Neural Network Self-organizing Map (SOM). Performance of these ML models was benchmarked using search data. One data set contained 137 Cs source, and another dataset contained 131 I source.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2021

Detection of Anomalies in Gamma Background Radiation Data with K-Means and Self-Organizing Map Clustering Algorithms (Consortium on Nuclear Security Technologies (CONNECT) Q1 Report)

Environmental screening of gamma radiation consists of detecting weak nuisance and anomaly signal in the presence of strong and highly varying background. In a typical scenario, a mobile detector-spectrometer continuously measures gamma radiation spectra in short, e.g., one-second, signal acquisition intervals. The measurement data is a 2D matrix, where one dimension is gamma ray energy, and the other dimension is the number of measurements or total time. In principle, gamma radiation sources can be detected and identified from the measured data by their unique spectral lines. Detecting sources from data measured in a search scenario is difficult due to the highly varying background because of naturally occurring radioactive material (NORM), and low signal-to-noise ratio (S/N) of spectral signal measured during one-second acquisition intervals. The objective of this work is to explore unsupervised machine learning (ML) algorithms for detection and identification of weak nuisances and anomalies events in the presence of highly fluctuating background. The challenge is that spectral lines of isotopes are difficult to observe in one-second measurements. Averaging over the entire measurement campaign data set reveals spectral lines of most common background isotopes. Spectral lines of orphan sources, which might appear only in a few measurements during the campaign, will be washed out if averaging is performed over the entire measurement data set. The approach we have explored consists of extracting one-second measurements containing weak spectral features through data clustering. Averaging one-second spectra in a cluster should reveal the presence of anomaly sources. We created two ML models using K-means clustering and Neural Network Self-organizing Map (SOM). Performance of these ML models was benchmarked using search data. One data set contained 137 Cs source, and another dataset contained 131 I source.

61 RADIATION PROTECTION AND DOSIMETRY↗

DOE OSTI2023

An Inventory of AI-ready Benchmark Data for US Fires, Heatwaves, and Droughts

Extreme weather events, including fires, heatwaves, and droughts, have significant impacts on earth, environmental, and energy systems. Mechanistic and predictive understanding, as well as probabilistic risk assessment of these extreme weather events, are crucial for detecting, planning for, and responding to these extremes. Records of extreme weather events provide an important data source for understanding present and future extremes, but the existing data needs preprocessing before it can be used for analysis. Moreover, there are many nonstandard metrics defining the levels of severity or impacts of extremes. In this study, we compile a comprehensive benchmark data inventory of extreme weather events, including fires, heatwaves, and droughts. The dataset covers the period from 2001 to 2020 with a daily temporal resolution and a spatial resolution of 0.5°×0.5° (~55km×55km) over the continental United States (CONUS), and a spatial resolution of 1km × 1km over the Pacific Northwest (PNW) region, together with the co-located and relevant meteorological variables. By exploring and summarizing the spatial and temporal patterns of these extremes in various forms of marginal, conditional, and joint probability distributions, we gain a better understanding of the characteristics of climate extremes. The resulting AI/ML-ready data products can be readily applied to ML-based research, fostering and encouraging AI/ML research in the field of extreme weather. This study can contribute significantly to the advancement of extreme weather research, aiding researchers, policymakers, and practitioners in developing improved preparedness and response strategies to protect communities and ecosystems from the adverse impacts of extreme weather events.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2023

An Inventory of AI-ready Benchmark Data for US Fires, Heatwaves, and Droughts

Extreme weather events, including fires, heatwaves, and droughts, have significant impacts on earth, environmental, and energy systems. Mechanistic and predictive understanding, as well as probabilistic risk assessment of these extreme weather events, are crucial for detecting, planning for, and responding to these extremes. Records of extreme weather events provide an important data source for understanding present and future extremes, but the existing data needs preprocessing before it can be used for analysis. Moreover, there are many nonstandard metrics defining the levels of severity or impacts of extremes. In this study, we compile a comprehensive benchmark data inventory of extreme weather events, including fires, heatwaves, and droughts. The dataset covers the period from 2001 to 2020 with a daily temporal resolution and a spatial resolution of 0.5°×0.5° (~55km×55km) over the continental United States (CONUS), and a spatial resolution of 1km × 1km over the Pacific Northwest (PNW) region, together with the co-located and relevant meteorological variables. By exploring and summarizing the spatial and temporal patterns of these extremes in various forms of marginal, conditional, and joint probability distributions, we gain a better understanding of the characteristics of climate extremes. The resulting AI/ML-ready data products can be readily applied to ML-based research, fostering and encouraging AI/ML research in the field of extreme weather. This study can contribute significantly to the advancement of extreme weather research, aiding researchers, policymakers, and practitioners in developing improved preparedness and response strategies to protect communities and ecosystems from the adverse impacts of extreme weather events. Usage Notes We presented a long term (2001-2020) and comprehensive data inventory of historical extreme events with daily temporal resolution covering the separate spatial extents of CONUS (0.5°×0.5°) and PNW(1km×1km) for various applications and studies. The dataset with 0.5°×0.5° resolution for CONUS can be used to help build more accurate climate models for the entire CONUS, which can help in understanding long-term climate trends, including changes in the frequency and intensity of extreme events, predicting future extreme events as well as understanding the implications of extreme events on society and the environment. The data can also be applied for risk accessment of the extremes. For example, ML/AI models can be developed to predict wildfire risk or forecast HWs by analyzing historical weather data, and past fires or heateave , allowing for early warnings and risk mitigation strategies. Using this dataset, AI-driven risk assessment models can also be built to identify vulnerable energy and utilities infrastructure, imrpove grid resilience and suggest adaptations to withstand extreme weather events. The high-resolution 1km×1km dataset ove PNW are advantageous for real-time, localized and detailed applications. It can enhance the accuracy of early warning systems for extreme weather events, helping authorities and communities prepare for and respond to disasters more effectively. For example, ML models can be developed to provide localized HW predictions for specific neighborhoods or cities, enabling residents and local emergency services to take targeted actions; the assessment of drought severity in specific communities or watersheds within the PNW can help local authorities manage water resources more effectively.

Lin, Xinming↗

DOE OSTI2020

AlphaBuilding - Synthetic Buildings Operation Dataset

This is a synthetic building operation dataset which includes HVAC, lighting, miscellaneous electric loads (MELs) system operating conditions, occupant counts, environmental parameters, end-use and whole-building energy consumptions at 10-minute intervals. The data is created with 1395 annual simulations using the U.S. DOE detailed medium-sized reference office building, and 30 years' historical weather data in three typical climates including Miami, San Francisco, and Chicago. Three energy efficiency levels of the building and systems are considered. Assumptions regarding occupant movements, occupants' diverse temperature preferences, lighting, and MELs are adopted to reflect realistic building operations. A semantic building metadata schema - BRICK, is used to store the building metadata. The dataset is saved in a 1.2 TB of compressed HDF5 file. This dataset can be used in various applications, including building energy and load shape benchmarking, energy model calibration, evaluation of occupant and weather variability and their influences on building performance, algorithm development and testing for thermal and energy load prediction, model predictive control, policy development for reinforcement learning based building controls.

AlphaBuilding↗

DOE OSTI2026

Machine learning inversion of interatomic force constants from single-crystal inelastic neutron scattering

Atomic vibrations govern many macroscopic properties of materials, but experiments to comprehensively probe them remain challenging. Inelastic neutron scattering (INS) is a powerful technique to map phonon dispersions in crystals, especially when leveraging modern time-of-flight (ToF) spectrometers with large detectors. However, efficiently and robustly extracting interatomic force constants (FCs) parameterizing phonon dynamics from experimental spectra remains a bottleneck due to the complexity and high dimensionality of ToF INS datasets. Here, we present a machine learning approach for the direct inversion of FCs from single-crystal INS measurements. The framework leverages synthetic training data generated using universal machine-learned force fields and an efficient physics-based forward model. We benchmark two neural architectures–one emphasizing structured latent representation learning and the other direct, supervised spectral regression–across simulated datasets for two materials under idealized and noisy conditions. The latent-representation model is subsequently applied to experimental single-crystal INS data on germanium. The model is shown to reproduce FCs derived from both first-principles simulations and from iterative optimization, and furthermore achieves reliable inference even from sparse, single-orientation measurements representing short data acquisitions. Analysis of the learned latent space reveals semantically continuous and physically interpretable encodings that support strong cross-domain generalization. By bridging theoretical and experimental domains, we establish a path toward rapid inversion of experimental spectra and data-driven interpretation of temperature-dependent lattice dynamics.

42 ENGINEERING↗

DOE OSTI2025

D–MOPH–25: diverse MOF–molecule pairs for Henry’s constants prediction

Computational methods like grand-canonical Monte Carlo simulations and machine learning (ML) have accelerated metal–organic frameworks (MOF) exploration but are typically limited to a narrow range of adsorbates due to data availability and force field constraints. In this study, we introduce a dataset of diverse MOF–molecule pairs for Henry’s constant prediction, D–MOPH–25, which systematically explores a diverse chemical space by combining 113 molecular adsorbates with over 5000 MOF structures through an active learning process. D–MOPH–25 constitutes the most diverse adsorbate dataset used in any ML study of molecular adsorption in MOFs to date. Our workflow builds a benchmark for predicting Henry’s constants at 300 K, leveraging conformal prediction for uncertainty quantification. Assessment through Shannon entropy and uniform manifold approximation and projection confirms the comprehensiveness of D–MOPH–25 while highlighting the importance of robust classification to filter out unphysical data points in regression tasks. Although future enhancements in model architecture and sampling criteria could improve predictive performance, our dataset already spans the target space using only 2.31% of total possibilities. This comprehensive dataset facilitates assessment of model generalizability across adsorbate species and can establish a foundation for high-throughput MOF screening and ML-driven separation processes.

active learning↗

DOE OSTI2021

EpitopeVec: linear epitope prediction using deep protein sequence embeddings

Motivation: B-cell epitopes (BCEs) play a pivotal role in the development of peptide vaccines, immuno-diagnostic reagents and antibody production, and thus in infectious disease prevention and diagnostics in general. Experimental methods used to determine BCEs are costly and time-consuming. Therefore, it is essential to develop computational methods for the rapid identification of BCEs. Although several computational methods have been developed for this task, generalizability is still a major concern, where cross-testing of the classifiers trained and tested on different datasets has revealed accuracies of 51–53%. Results: We describe a new method called EpitopeVec, which uses a combination of residue properties, modified antigenicity scales, and protein language model-based representations (protein vectors) as features of peptides for linear BCE predictions. Extensive benchmarking of EpitopeVec and other state-of-the-art methods for linear BCE prediction on several large and small datasets, as well as cross-testing, demonstrated an improvement in the performance of EpitopeVec over other methods in terms of accuracy and area under the curve. As the predictive performance depended on the species origin of the respective antigens (viral, bacterial and eukaryotic), we also trained our method on a large viral dataset to create a dedicated linear viral BCE predictor with improved cross-testing performance.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2024

OReole-FM: successes and challenges toward billion-parameter foundation models for high-resolution satellite imagery

While the pretraining of Foundation Models (FMs) for remote sensing (RS) imagery is on the rise, models remain restricted to a few hundred million parameters. Scaling models to billions of parameters has been shown to yield unprecedented benefits including emergent abilities, but requires data scaling and computing resources typically not available outside industry R&D labs. In this work, we pair high-performance computing resources including Frontier supercomputer, America's first exascale system, and high-resolution optical RS data to pretrain billion-scale FMs. Our study assesses performance of different pretrained variants of vision Transformers across image classification, semantic segmentation and object detection benchmarks, which highlight the importance of data scaling for effective model scaling. Moreover, we discuss construction of a novel TIU pretraining dataset, model initialization, with data and pretrained models intended for public release. By discussing technical challenges and details often lacking in the related literature, this work is intended to offer best practices to the geospatial community toward efficient training and benchmarking of larger FMs.

Ambrozio Dias, Philipe↗

DOE OSTI2021

The LHC Olympics 2020 a community challenge for anomaly detection in high energy physics

A new paradigm for data-driven, model-agnostic new physics searches at colliders is emerging, and aims to leverage recent breakthroughs in anomaly detection and machine learning. In order to develop and benchmark new anomaly detection methods within this framework, it is essential to have standard datasets. To this end, we have created the LHC Olympics 2020, a community challenge accompanied by a set of simulated collider events. Participants in these Olympics have developed their methods using an R&D dataset and then tested them on black boxes: datasets with an unknown anomaly (or not). Furthermore, methods made use of modern machine learning tools and were based on unsupervised learning (autoencoders, generative adversarial networks, normalizing flows), weakly supervised learning, and semi-supervised learning. This paper will review the LHC Olympics 2020 challenge, including an overview of the competition, a description of methods deployed in the competition, lessons learned from the experience, and implications for data analyses with future datasets as well as future colliders.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2022

Bay Area Regional Energy: Network Integrated Commercial Retrofits (BRICR) Project. Final Report

The BRICR project applied large-scale building energy modeling concepts with the aim of reducing the cost of energy efficiency targeting, design, and project development, and measurement of energy savings for energy efficiency programs implemented by local governments that serve small and medium commercial buildings (SMB). The project leveraged the services and resources of existing local government energy programs serving disadvantaged and hard-to-reach SMB customers. In contrast to programs run by utilities, local government programs generally do not have direct access to energy billing records for an entire class of customers in a geographic area, which prior research demonstrated useful for large-scale building energy model baseline development and calibration. , However, local governments are rich in public records that offer important clues about physical attributes and uses that, along with behavior, determine energy use. Relying only on public records, BRICR demonstrated development of credible baseline energy models for 3,792 office, retail, and hotel buildings. Publicly disclosed annual energy use data from a local energy benchmarking program and anonymized data from the Building Performance Database, the nation’s largest dataset about energy-related characteristics of buildings, were utilized to validate and calibrate energy models via an innovative method comparing distributions of energy intensity by fuel type for portfolios of buildings of similar size, vintage, and use. Portfolio calibration does not provide certainty that an energy model fits an individual building; the method is useful when billing data is not accessible – a common situation for researchers, energy service providers and ESCOs, local governments, and any party other than a utility. A software component was developed, the BRICR gem, which automates simulation when relevant data is added or edited by the user to a file saved in the standardized BuildingSync XML schema for energy audit data. The component was demonstrated as a simplified means to generate a mass of energy models corresponding to public records containing basic attributes such as building scale, location, use, year built, and aspect ratio in combination with building energy code prototype data corresponding to use and vintage. The component was also demonstrated as a simplified means to automate energy simulation when attributes are revised; the intention was to enable iterative improvement of the baseline model and energy savings estimates for common energy conservation measures as users revise relevant attributes based on their observations. In the context of institutional change and uncertainty for the participating local government energy programs, 13 whole building retrofits were completed. Impacts were measured by applying the CalTRACK2.0 methods to standardize measurement of normalized metered energy consumption. The GRIDMeter methods of stratified sampling and individual load shape analysis were applied to adjust for impacts of the effect of COVID-19 on retrofitted buildings in the context of all local buildings of similar size and use. Excluding impacts of the pandemic, retrofitted buildings demonstrated between 1.6% and 25.1% reduction in energy use. The project contributed use cases and feedback that helped inform evolution of the software tools and data formats that were combined for the first time in the BRICR project.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

DOE OSTI2021

Structural- and Functional-Informed Machine Learning for Protein Function Prediction

In this project we aimed to extend methods for protein function prediction to include structural prediction data, and benchmark methods against existing tools. We proposed to apply the method to large metagenome datasets, and develop approaches to examine activity-based protein profiling results for protein function-structure patterns. Nitrogen cycle protein families were previously identified and are used here to provide a proof-of-principle for use of structure prediction in protein function classification.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2020

SFCOMPO Database of Spent Nuclear Fuel Assay Data – the Next Frontier

SFCOMPO is the world’s largest database for measured spent nuclear fuel assay data. An international effort coordinated by the Nuclear Energy Agency (NEA) resulted in a significant expansion of the database and its release online in 2017 as a downloadable application. The SFCOMPO Technical Review Group (TRG) was recently formed under the direction of NEA’s Nuclear Science Committee/Working Party on Nuclear Criticality Safety and was mandated to maintain and further coordinate the development of SFCOMPO. This TRG is currently focused on (1) critical evaluation of the experimental assay data by independent experts and (2) development of benchmarks and benchmark models that can be applied to validate burnup codes. This will improve the quality and documentation of the experimental datasets and enable their use by the international community to support code validation for design and safety analysis of spent nuclear fuel transportation, storage, and repository applications. It follows the precedent and draws on the experience gained from similar NEA efforts in the International Reactor Physics Experiment Evaluation Project and the International Criticality Safety Benchmark Experiment Project. Ongoing SFCOMPO evaluations have served as a test bed to develop templates for documenting evaluations, develop review guidance, improve approaches for a global uncertainty analysis, and devise a strategy focused on providing practical information of highest value to the user community. The current effort, status, and associated challenges are discussed.

Ilas, Germina↗

DOE OSTI2024

I/O performance studies of analysis workloads on production and dedicated resources at CERN

The recent evolutions of the analysis frameworks and physics data formats of the LHC experiments provide the opportunity of using central analysis facilities with a strong focus on interactivity and short turnaround times, to complement the more common distributed analysis on the Grid. In order to plan for such facilities, it is essential to know in detail the performance of the combination of a given analysis framework, of a specific analysis and of the installed computing and storage resources. This contribution describes performance studies performed at CERN, using the EOS disk-based storage, either directly or through an XCache instance, from both batch resources and highperformance compute nodes which could be used to build an analysis facility. A variety of benchmarks, both synthetic and based on real-world physics analyses and their corresponding input datasets, are utilized. In particular, the RNTuple format from the ROOT project is put to the test and compared to the latest version of the TTree format, and the impact of caches is assessed. In addition, we assessed the difference in performance between the use of storage system specific protocols, like XRootd, and FUSE. The results of this study are intended to be a valuable input in the design of analysis facilities, at CERN and elsewhere.

Sciabà, Andrea↗

DOE OSTI2021

Attend and Decode: 4D fMRI Task State Decoding Using Attention Models

Source code for Brain Attend and Decode paper. Functional magnetic resonance imaging (fMRI) is a neuroimaging modality that captures the blood oxygen level in a subject's brain while the subject either rests or performs a variety of functional tasks under different conditions. Given fMRI data, the problem of inferring the task, known as task state decoding, is challenging due to the high dimensionality (hundreds of million sampling points per datum) and complex spatio-temporal blood flow patterns inherent in the data. In this work, we propose to tackle the fMRI task state decoding problem by casting it as a 4D spatiotemporal classification problem. We present a novel architecture called Brain Attend and Decode (BAnD), that uses residual convolutional neural networks for spatial feature extraction and self-attention mechanisms for temporal modeling. We achieve significant performance gain compared to previous works on a 7-task benchmark from the large-scale Human Connectome Project-Young Adult (HCP-YA) dataset. We also investigate the transferability of BAnD's extracted features on unseen HCP tasks, either by freezing the spatial feature extraction layers and retraining the temporal model, or finetuning the entire model. The pre-trained features from BAnD are useful on similar tasks while finetuning them yields competitive results on unseen tasks/conditions.

Ng, BrendaM.↗

DOE OSTI2024

Established Regulatory Limits for Surface Water and Groundwater

The regulatory limits for surface water and groundwater contained in this database comprise chemical-specific preliminary remediation goals (PRGs) based on potential Applicable or Relevant and Appropriate Requirements (ARARs). These values should be used in conjunction with PRGs based on risk assessment to ensure that the PRGs for a site meet the residual risk requirements for protection of human health and the environment in the National Oil and Hazardous Substance Pollution Contingency Plan (NCP). In some cases, the values presented in this database may be To-Be-Considered (TBC) Guidance because these values are superseded by site-specific state requirements. ARARs are available for the Federal government, California, Georgia, Illinois, Kentucky, New Jersey, New Mexico, Nevada, New York, Ohio, South Carolina, Tennessee, Washington, and Wisconsin. The dataset provides maximum contaminant levels for drinking water and water quality criteria for fresh and marine environments based on acute and chronic exposures. Users can select ARAR sources, drinking water standards and water quality standards, individual chemicals, and retrieve results in tabular or spreadsheet formats for analysis. ARARs are derived from authoritative sources, including federal and state government agencies. The dataset supports environmental risk assessments, regulatory decision-making, and environmental planning, with tools for benchmarking against promulgated standards. This structured approach ensures a robust evaluation of environmental risks tailored to regulatory needs.

Stewart, Debra [Oak Ridge National Laboratory (ORN↗

DOE OSTI2024

Chemical-specific Parameters Dataset

The chemical-specific parameters dataset is searchable for physicochemical information for multiple chemicals simultaneously. After selecting chemicals of interest and the desired parameters, the RAIS will generate a table containing the values, chosen according to an established hierarchy. Results can be downloaded in Excel format. Over 40 parameters are available, including melting point, boiling point, density, density, vapor pressure, water solubility, and Henry’s Law constants. Thirteen primary sources are used to populate the dataset of chemical-specific parameters. These values should be used in cancer risk and noncancer hazard assessments for the calculation of preliminary remediation goals (PRGs), hazard characterization, and transport modeling. Users can select up to 1000 chemicals per query. The dataset supports environmental risk assessments, regulatory decision-making, and environmental planning with tools for benchmarking against risk-based standards. This structured approach ensures a robust evaluation of environmental risks tailored to regulatory needs.

Dolislager, Fred [Oak Ridge National Laboratory (O↗

DOE OSTI2024

Toxicity Values for Chemicals

The toxicity values for chemicals contained in this dataset comprise acute, subchronic, and chronic exposure durations. Cancer slope factors, inhalation unit risk, reference dose, and reference concentrations are available. These values should be used in cancer risk and noncancer hazard assessments for the calculation of preliminary remediation goals (PRGs) and hazard characterization. Users can select toxicity values from 10 combinations of exposure durations and cancer/noncancer toxicity values and select up to 1000 chemicals per query. The dataset supports environmental risk assessments, regulatory decision-making, and environmental planning with tools for benchmarking against risk-based standards. This structured approach ensures a robust evaluation of environmental risks tailored to regulatory needs.

Stewart, Debra [Oak Ridge National Laboratory (ORN↗