Search NASA⌕ Search

SEARCH · Search NASA

Results for “Building Dataset”

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 703 records · Page 39

MABAL: a Novel Deep-Learning Architecture for Machine-Assisted Bone Age Labeling

Bone age assessment (BAA) is a commonly performed diagnostic study in pediatric radiology to assess skeletal maturity. The most commonly utilized method for assessment of BAA is the Greulich and Pyle method (Pediatr Radiol 46.9:1269–1274, 2016; Arch Dis Child 81.2:172–173, 1999) atlas. The evaluation of BAA can be a tedious and time-consuming process for the radiologist. As such, several computer-assisted detection/diagnosis (CAD) methods have been proposed for automation of BAA. Classical CAD tools have traditionally relied on hard-coded algorithmic features for BAA which suffer from a variety of drawbacks. Recently, the advent and proliferation of convolutional neural networks (CNNs) has shown promise in a variety of medical imaging applications. There have been at least two published applications of using deep learning for evaluation of bone age (Med Image Anal 36:41–51, 2017; JDI 1–5, 2017). However, current implementations are limited by a combination of both architecture design and relatively small datasets. The purpose of this study is to demonstrate the benefits of a customized neural network algorithm carefully calibrated to the evaluation of bone age utilizing a relatively large institutional dataset. In doing so, this study will aim to show that advanced architectures can be successfully trained from scratch in the medical imaging domain and can generate results that outperform any existing proposed algorithm. The training data consisted of 10,289 images of different skeletal age examinations, 8909 from the hospital Picture Archiving and Communication System at our institution and 1383 from the public Digital Hand Atlas Database. The data was separated into four cohorts, one each for male and female children above the age of 8, and one each for male and female children below the age of 10. The testing set consisted of 20 radiographs of each 1-year-age cohort from 0 to 1 years to 14–15+ years, half male and half female. The testing set included left-hand radiographs done for bone age assessment, trauma evaluation without significant findings, and skeletal surveys. A 14 hidden layer-customized neural network was designed for this study. The network included several state of the art techniques including residual-style connections, inception layers, and spatial transformer layers. Data augmentation was applied to the network inputs to prevent overfitting. A linear regression output was utilized. Mean square error was used as the network loss function and mean absolute error (MAE) was utilized as the primary performance metric. MAE accuracies on the validation and test sets for young females were 0.654 and 0.561 respectively. For older females, validation and test accuracies were 0.662 and 0.497 respectively. For young males, validation and test accuracies were 0.649 and 0.585 respectively. Finally, for older males, validation and test set accuracies were 0.581 and 0.501 respectively. The female cohorts were trained for 900 epochs each and the male cohorts were trained for 600 epochs. An eightfold cross-validation set was employed for hyperparameter tuning. Test error was obtained after training on a full data set with the selected hyperparameters. Using our proposed customized neural network architecture on our large available data, we achieved an aggregate validation and test set mean absolute errors of 0.637 and 0.536 respectively. To date, this is the best published performance on utilizing deep learning for bone age assessment. Our results support our initial hypothesis that customized, purpose-built neural networks provide improved performance over networks derived from pre-trained imaging data sets. We build on that initial work by showing that the addition of state-of-the-art techniques such as residual connections and inception architecture further improves prediction accuracy. This is important because the current assumption for use of residual and/or inception architectures is that a large pre-trained network is required for successful implementation given the relatively small datasets in medical imaging. Instead we show that a small, customized architecture incorporating advanced CNN strategies can indeed be trained from scratch, yielding significant improvements in algorithm accuracy. It should be noted that for all four cohorts, testing error outperformed validation error. One reason for this is that our ground truth for our test set was obtained by averaging two pediatric radiologist reads compared to our training data for which only a single read was used. This suggests that despite relatively noisy training data, the algorithm could successfully model the variation between observers and generate estimates that are close to the expected ground truth.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

SUSY searches at the Tevatron

The Tevatron collider has provided the CDF and D0 collaborations with large datasets as input to a rich program of physics beyond the standard model. The results presented here are from recent searches for SUSY particles using up to 6 fb{sup -1} of data. Supersymmetry (SUSY) [1] is one of the most favored theories beyond the standard model (SM). Each SM particle is associated to a sparticle whose spin differs by one half unit. This boson-fermion symmetry is obviously broken by some unknown mechanism. Even in the minimal supersymmetric extension of the SM (MSSM [2]) there are a large number of free parameters. To reduce this number one can introduce new assumptions on the symmetry breaking mechanism and build models based on minimal supergravity (as mSUGRA [3]) or on a Gauge Mediated Symmetry Breaking scenario (GMSB [4]), a top-down approach. Another possibility is to make phenomenological assumptions to reduce the number of particles accessible to the experiment while keeping some of the properties of the above models (bottom-up approach). As the sparticles are heavy, to produce them one has to make collisions at the highest center of mass energy. The Tevatron was the best place for discovery until the start of LHC. In the near term, Tevatron experiments and their large datasets remain competitive in areas like production of third generation squarks and of non-coloured sparticles. I will report on recent results from the CDF and D0 collaborations, assuming R-parity is conserved, i.e the sparticles are produced in pairs, and the lightest of them (LSP) is stable, neutral, weakly interacting, and detected as missing transverse energy, E{sub T}.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Deep learning to estimate permeability using geophysical data

Time-lapse electrical resistivity tomography (ERT) is a popular geophysical method to estimate three-dimensional (3D) permeability fields from electrical potential difference measurements. Traditional inversion and data assimilation methods are used to ingest this ERT data into hydrogeophysical models to estimate permeability. Due to ill-posedness and the curse of dimensionality, existing inversion strategies provide poor estimates and low resolution of the 3D permeability field. Recent advances in deep learning provide us with powerful algorithms to overcome this challenge. This paper presents a deep learning (DL) framework to estimate the 3D subsurface permeability from time-lapse ERT data. To test the feasibility of the proposed framework, we train DL-enabled inverse models on simulation data. Each measurement in both synthetic and field data is standardized by removing the mean and scaling the time-series to unit variance. This pre-processing step is necessary to bring simulation data closer to field observations. Subsurface process models based on hydrogeophysics are used to generate this synthetic data. Training performed on limited simulation data resulted in the DL model over-fitting. An advanced data augmentation based on mixup is implemented to generate additional training samples to overcome this issue. This mixup technique creates weakly labeled (low-fidelity) samples from strongly labeled (high-fidelity) data. The weakly labeled training data is then used to develop DL-enabled inverse models and reduce over-fitting. As both time-lapse ERT (1133048 features/realization) and 3D permeability (585453 features/realization) data samples are from a high-dimensional space, principal component analysis (PCA) is employed to reduce dimensionality. Encoded ERT and encoded permeability are generated using the trained PCA estimators. A deep neural network is then trained to map the encoded ERT to encoded permeability. This mixup training and unsupervised learning allowed us to build a fast and reasonably accurate DL-based inverse model under limited simulation data. Results show that proposed weak supervised learning can capture salient spatial features in the 3D permeability field. Quantitatively, the average mean squared error (in terms of the natural log) on the strongly labeled training, validation, and test datasets is less than 0.5. The R 2 -score (global metric) is greater than 0.75, and the percent error in each cell (local metric) is less than 10%. Finally, an added benefit in terms of computational cost is that the proposed DL-based inverse model is at least O(10 4 ) times faster than running a forward model once it is trained. Data generation, DL model training, and hyperparameter tuning to identify optimal neural network architectures utilized high-performance computing resources while the DL inference is performed on a standard laptop. Approximately, O(10 5 ) processor hours are used for generating data and DL tuning and training. We acknowledge that the data generation and DL model development are expensive. But once a DL model is trained, it can be re-used for inversion rapidly for the given system, with set physics and domain. Note that traditional inversion may require multiple forward model simulations (e.g., in the order of 10 to 1000), which are very expensive. This computational savings ≈ O(10 5 ) – O(10 7 )) makes the proposed DL-based inverse model attractive for subsurface imaging and real-time ERT monitoring applications due to fast and yet reasonably accurate estimations of permeability field.

58 GEOSCIENCES↗

Expanded analysis of machine learning models for nuclear transient identification using TPOT

Industries around the world are becoming more and more data driven. The nuclear field is no exception with several different applications being proposed. One popular area of research is the use of machine learning in transient detection. This paper seeks to build upon a previous study which made use of the AutoML package TPOT to train traditional machine learning models to classify transient events occurring with a reactor. Synthetic data was once again collected using a GPWR reactor simulator. Data on 12 different events was collected using 15 different initial conditions. Here, a dataset consisting of over 100,000 data points was compiled and used to train 7 different machine learning models using a pre-defined TPOT dictionary with 12 different preprocessing techniques. Three of the trained models were able to produce validation results in the 90s with the expanded dataset. Once the models were trained, it was possible to look into where during the simulation, misclassifications occurred. Using these three models, analysis was done to determine if TPOT could be used to train models that were effective if important features were missing. The results from this were positive with the newly trained models scoring close to the original models. Finally, to conclude this study, the three high performing models were retrained using different random states to see if there was any major variation when different states were used.

42 ENGINEERING↗

Storage events: distributed users, federation and beyond

For federated storage to work well, some knowledge from each storage system must exist outside that system, regardless of the use case. This is needed to allow coordinated activity; e.g., executing analysis jobs on worker nodes with good accessibility to the data.Currently, this is achieved by clients notifying central services of activity; e.g., a client notifies a replica catalogue after an upload. Unfortunately, this forces end users to use bespoke clients. It also forces clients to wait for asynchronous activities to finish.dCache provides an alternative approach: storage events. In this approach the storage systems (rather than the clients) become the coordinating service, notifying interested parties of key events.At DESY, we are investigating storage events along with Apache OpenWhisk and Kubernetes to build a "serverless" cloud, similar to AWS Lambda or Google Cloud Functions, for photon science use cases.Storage events are more generally useful: catalogues are notified whenever data is uploaded or delete, tape becomes more efficient because analysis can start immediately after the data is on disk, caches can be "smart" fetching new datasets preemptively.In this paper we will present work within dCache to support a new event-based interface, with which these and other use cases become more efficient.

Millar, A.Paul↗

Extended Fayans energy density functional: optimization and analysis

The Fayans energy density functional (EDF) has been very successful in describing global nuclear properties (binding energies, charge radii, and especially differences of radii) within nuclear density functional theory. In a recent study, supervised machine learning methods were used to calibrate the Fayans EDF. Building on this experience, in this work we explore the effect of adding isovector pairing terms, which are responsible for different proton and neutron pairing fields, by comparing a 13D model without the isovector pairing term against the extended 14D model. At the heart of the calibration is a carefully selected heterogeneous dataset of experimental observables representing ground-state properties of spherical even–even nuclei. To quantify the impact of the calibration dataset on model parameters and the importance of the new terms, we carry out advanced sensitivity and correlation analysis on both models. The extension to 14D improves the overall quality of the model by about 30%. The enhanced degrees of freedom of the 14D model reduce correlations between model parameters and enhance sensitivity.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Emerging Flexible Designs for Geospatial Multimodal Foundation Models

Foundation models are rapidly transforming Earth observation by enabling scalable pretraining across diverse unlabeled geospatial modalities. However, their architectural diversity—ranging from encoder-only to encoder-decoder and masked autoencoding paradigms—makes it challenging to assess performance trade-offs in a consistent manner. In this work, we present an apples-to-apples comparison of leading FM architectures designed for geospatial multimodal reasoning, with a particular focus on flexibility across varied spectral band configurations. We standardize pretraining using identical self-supervised learning objectives and training datasets, and evaluate all models under consistent parameterization on the GEOBench benchmark across classification and segmentation tasks. Our results offer new insights into the design trade-offs between model flexibility, modality alignment, and downstream task performance. By highlighting architectural strengths and limitations under controlled conditions, this study provides practical guidance for building next-generation geospatial foundation models capable of robust multimodal reasoning.

Ambrozio Dias, Philipe [ORNL] (ORCID:0000000194277↗

Rapid Evaluation Framework for the CMIP7 Assessment Fast Track

As Earth system models (ESMs) grow in complexity and in volume of output data, there is an increasing need for rapid, comprehensive evaluation of their scientific performance. The upcoming Assessment Fast Track for the Seventh Phase of the Coupled Model Intercomparison Project (CMIP7) will require expeditious response for model analyses designed to inform and drive integrated Earth system assessments. To meet this challenge, the Rapid Evaluation Framework (REF), a community-driven platform for benchmarking and performance assessment of ESMs, was designed and developed. The initial implementation of the REF, constructed to meet the near-term needs of the CMIP7 Assessment Fast Track, builds upon four disparate community evaluation and benchmarking tools that are coupled together using the Coordinated Model Evaluation Capabilities (CMEC) framework. The REF runs within a containerized workflow for portability and reproducibility and is aimed at generating and organizing diagnostics covering a variety of model variables. The REF leverages well documented observational datasets to provide assessments of model fidelity across a collection of diagnostics. All diagnostics were identified and selected with community involvement and consultation. Operational integration with the Earth System Grid Federation (ESGF) will permit automated execution of the REF for selected diagnostics as soon as model output data are published on ESGF by the originating modeling centers. The REF is designed to be portable across a range of current computational platforms to facilitate use by modeling centers for assessing the evolution of model versions or gauging the relative performance of CMIP simulations before being published on ESGF. When integrated into production simulation workflows, results from the REF provide immediate quantitative feedback that allows model developers and scientists to quickly identify model biases and performance issues. After the REF is released to the community, its subsequent development and support will be prioritized by an international consortium of scientists and engineers, enabling a broader impact across Earth science disciplines. For instance, the REF will facilitate improvements to models and will enhance confidence in model projections through process-based selection of models based on their performance with respect to observations. Production of reproducible diagnostics and community-based assessments are key features of the REF. Furthermore, providing interoperability with existing evaluation packages assures that contributions from previous community efforts will be available for use in future model intercomparison projects.

Hoffman, Forrest [ORNL] (ORCID:0000000158024134)↗

‪A Novel Methodology for Longitudinal Studies of Home Thermal Comfort Perception and Behavior

Human-building interactions significantly influence building energy consumption and affect peak energy demand. For example, heating and cooling contribute 46% of daily peak residential energy demand. Grid-interactive efficient buildings (GEBs) can potentially increase energy-demand flexibility and accelerate the adoption of renewables. However, traditional demand response (DR) programs focused on shedding peak loads disregard the human-building interactions leading to occupant thermal frustration. Specifically, they do not model occupants’ ability to override thermostat controls, nor how indoor environmental conditions and sociocultural factors affect the timing and magnitude of overrides. Studies have found 20% of occupants override thermostat setpoints during DR events longer than 6 hours yet lack detail about the motivation that might guide more successful efforts. Balancing energy-demand flexibility with occupant thermal comfort requires understanding dynamic occupant behavior, the underlying psychophysiological drivers, in the context of homes. This paper presents methods for scalable longitudinal studies of residential occupant behavior dynamics to inform the development of psychophysiological occupant-centric building models. Smart sensors were installed to measure local environmental conditions in 20 homes in two regions of the United States. Just-in-time ecological momentary assessments (EMAs) provided qualitative data on occupant thermal comfort and local environmental conditions not captured in Internet-of-Things (IoT) based studies or existing datasets. Participant interviews provided insight into environmental attitudes, mental models, and the role of economics in comfort and behavior, which in turn may affect thermostat interactions. Based on these data, interventions in the next phase of the study will collect data and monitor occupant behavior during simulated DR events.

Kane, Michael↗

DEEPEN 3D PFA Favorability Models and 2D Favorability Maps at Newberry Volcano

DEEPEN stands for DE-risking Exploration of geothermal Plays in magmatic ENvironments. Part of the DEEPEN project involved developing and testing a methodology for a 3D play fairway analysis (PFA) for multiple play types (conventional hydrothermal, superhot EGS, and supercritical). This was tested using new and existing geoscientific exploration datasets at Newberry Volcano. This GDR submission includes images, data, and models related to the 3D favorability and uncertainty models and the 2D favorability and uncertainty maps. The DEEPEN PFA Methodology is based on the method proposed by Poux et al. (2020), which uses the Leapfrog Geothermal software with the Edge extension to conduct PFA in 3D. This method uses all available data to build a 3D geodata model which can be broken down into smaller blocks and analyzed with advanced geostatistical methods. Each data set is imported into a 3D model in Leapfrog and divided into smaller blocks. Conditional queries can then be used to assign each block an index value which conditionally ranks each block's favorability, from 0-5 with 5 being most favorable, for each model (e.g., lithologic, seismic, magnetic, structural). The values between 0-5 assigned to each block are referred to as index values. The final step of the process is to combine all the index models to create a favorability index. This involves multiplying each index model by a given weight and then summing the resulting values. The DEEPEN PFA Methodology follows this approach, but split up by the specific geologic components of each play type. These components are defined as follows for each magmatic play type: 1. Conventional hydrothermal plays in magmatic environments: Heat, fluid, and permeability 2. Superhot EGS plays: Heat, thermal insulation, and producibility (the ability to create and sustain fractures suitable for and EGS reservoir) 3. Supercritical plays: Heat, supercritical fluid, pressure seal, and producibility (the proper permeability and pressure conditions to allow production of supercritical fluid) More information on these components and their development can be found in Kolker et al., 2022. For the purposes of subsurface imaging, it is easier to detect a permeable fluid-filled reservoir than it is to detect separate fluid and permeability components. Therefore, in this analysis, we combine fluid and permeability for conventional hydrothermal plays, and supercritical fluid and producibility for supercritical plays. More information on this process is described in the following sections. We also project the 3D favorability volumes onto 2D surfaces for simplified joint interpretation, and we incorporate an uncertainty component. Uncertainty was modeled using the best approach for the dataset in question, for the datasets where we had enough information to do so. Identifying which subsurface parameters are the least resolved can help qualify current PFA results and focus future efforts in data collection. Where possible, the resulting uncertainty models/indices were weighted using the same weights applied to the respective datasets, and summed, following the PFA methodology above, but for uncertainty. There are two different versions of the Leapfrog model and associated favorability models: - v1.0: The first release in June 2023 - v2.1: The second release, with improvements made to the earthquake catalog (included additional identified events, removed duplicate events), to the temperature model (fixed a deep BHT), and to the index models (updated the seismicity-heat source index models for supercritical and EGS, and the resistivity-insulation index models for all three play types). Also uses the jet color map rather than the magma color map for improved interpretability. - v2.1.1: Updated to include v2.0 uncertainty results (see below for uncertainty model versions) There are two different versions of the associated uncertainty models: - v1.0: The first release in June 2023 - v2.0: The se...

15 GEOTHERMAL ENERGY↗

Digital archives, big data and image-based culturomics for social impact assessment: Opportunities and challenges

Social impact assessment (SIA) is well-established but uses conventional approaches that have become less effective in recent decades, particularly in relation to declining survey response rates and a lack of youth engagement. Images from digital archives and social media sources are poised to advance the research and practice of SIA by transcending text-based methods with insights into changing landscapes, and human engagement with them. This viewpoint describes progress, challenges and cautions toward the development of such tools (defined as culturomics), using hydroelectricity cases to illustrate potential approaches. These tools build on foundational work in a range of disciplines, including the humanities and computer science. We describe necessary advances in machine learning, image digitization, and data aggregation and visualization techniques, as well as ways to ensure that such tools are carefully tested, applied and interpreted. Challenges include the automation, acquisition and management of datasets, and using these tools appropriately and equitably. Critically, culturomics of any kind must not be used as a replacement for engagement with people, but as complementary to inclusive stakeholder engagement. - Highlights: •Image-based digital archives and social media present opportunities for new SIA tools. •Application-ready big data approaches are emerging across many fields. •SIA scholars, practitioners and stakeholders should engage with culturomics. •Challenges include automation, digitization, interpretation and justice. •Culturomics of any kind must not replace engagement with people.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Integrated deep learning framework for unstable event identification and disruption prediction of tokamak plasmas

Abstract The ability to identify underlying disruption precursors is key to disruption avoidance. In this paper, we present an integrated deep learning (DL) based model that combines disruption prediction with the identification of several disruption precursors like rotating modes, locked modes, H-to-L back transitions and radiative collapses. The first part of our study demonstrates that the DL-based unstable event identifier trained on 160 manually labeled DIII-D shots can achieve, on average, 84% event identification rate of various frequent unstable events (like H-L back transition, locked mode, radiative collapse, rotating MHD mode, large sawtooth crash), and the trained identifier can be adapted to label unseen discharges, thus expanding the original manually labeled database. Based on these results, the integrated DL-based framework is developed using a combined database of manually labeled and automatically labeled DIII-D data, and it shows state-of-the-art (AUC = 0.940) disruption prediction and event identification abilities on DIII-D. Through cross-machine numerical disruption prediction studies using this new integrated model and leveraging the C-Mod, DIII-D, and EAST disruption warning databases, we demonstrate the improved cross-machine disruption prediction ability and extended warning time of the new model compared with a baseline predictor. In addition, the trained integrated model shows qualitatively good cross-machine event identification ability. Given a labeled dataset, the strategy presented in this paper, i.e. one that combines a disruption predictor with an event identifier module, can be applied to upgrade any neural network based disruption predictor. The results presented here inform possible development strategies of machine learning based disruption avoidance algorithms for future tokamaks and highlight the importance of building comprehensive databases with unstable event information on current machines.

plasma instabilities↗

Locating Equitable Solar Opportunities by Census Tract: A Guide to the Screening Tool for Equitable Adoption and Deployment of Solar (STEADy Solar)

The Screening Tool for Equitable Adoption and DeploYment of Solar (STEADy Solar) is a database and mapping tool that indicates locations that may be eligible for the Investment Tax Credit bonus adders defined in the 2022 Inflation Reduction Act (IRA). The tool combines publicly available information on demographics, solar technical potential, solar economics (modeled net present value), building counts by use-type, and eligibility for tax credit adders. It can be used by states, municipalities, community-based organizations, developers, and researchers to identify sites where solar projects may be economical and where federal incentives may be available to support equitable adoption of solar. This report describes the STEADy dataset and presents high level insights from the data.

census tract↗

Fast Bayesian optimization of Needle-in-a-Haystack problems using zooming memory-based initialization (ZoMBI)

Abstract Needle-in-a-Haystack problems exist across a wide range of applications including rare disease prediction, ecological resource management, fraud detection, and material property optimization. A Needle-in-a-Haystack problem arises when there is an extreme imbalance of optimum conditions relative to the size of the dataset. However, current state-of-the-art optimization algorithms are not designed with the capabilities to find solutions to these challenging multidimensional Needle-in-a-Haystack problems, resulting in slow convergence or pigeonholing into a local minimum. In this paper, we present a Zooming Memory-Based Initialization algorithm, entitled ZoMBI, that builds on conventional Bayesian optimization principles to quickly and efficiently optimize Needle-in-a-Haystack problems in both less time and fewer experiments. The ZoMBI algorithm demonstrates compute time speed-ups of 400× compared to traditional Bayesian optimization as well as efficiently discovering optima in under 100 experiments that are up to 3× more highly optimized than those discovered by similar methods.

36 MATERIALS SCIENCE↗

Bringing high performance computing to the biologist’s workbench: approaches, applications and challenges

Data-intensive and high-performance computing are poised to significantly impact the future of biological research which is increasingly driven by the prevalence of high-throughput experimental methodologies for genome sequencing, transcriptomics, proteomics, and other areas. Large centers such as NIH’s National Center for Biotechnology Information (NCBI), The Institute for Genomic Research (TIGR), and the DOE’s Joint Genome Institute (JGI) Integrated Microbial Genome (IMG) have made extensive use of multiprocessor architectures to deal with some of the challenges of processing, storing and curating exponentially growing genomic and proteomic datasets—enabling end users to rapidly access a growing public data source, as well as utilize analysis tools transparently on high-performance computing resources. Applying this computational power to single-investigator analysis, however, often relies on users to provide their own computational resources, forcing them to endure the learning curve of porting, building, and running software on multiprocessor architectures. Solving the next generation of large-scale biology challenges using multiprocessor machines—from small clusters to emerging petascale machines—can most practically be realized if this learning curve can be minimized through a combination of workflow management, data management and resource allocation as well as intuitive interfaces and compatibility with existing common data formats.

97 MATHEMATICS AND COMPUTING↗

PDF Entity Annotation Tool (PEAT)

While different text mining approaches – including the use of Artificial Intelligence (AI) and other machine based methods - continue to expand at a rapid pace, the tools used by researchers to create the labeled datasets required for training, modeling, and evaluation remain rudimentary. Labeled datasets contain the target attributes the machine is going to learn; for example, training an algorithm to delineate between images of a car or truck would generally require a set of images with a quantitative description of the underlying features of each vehicle type. Development of labeled textual data that can be used to build natural language machine learning models for scientific literature is not currently integrated into existing manual workflows used by domain experts. Published literature is rich with important information, such as different types of embedded text, plots, and tables that can all be used as inputs to train ML/natural language processing (NLP) models, when extracted and prepared in machine readable formats. Currently, both normalized data extraction of use to domain experts and extraction to support development of ML/NLP models are labor intensive and cumbersome manual processes. Automatic extraction of data and information from formats such as PDFs that are optimized for layout and human readability, not machine readability. The PDF (Portable Document Format) Entity Annotation Tool (PEAT) was developed with the goal of allowing users to annotate publications within their current print format, while also allowing those annotations to be captured in a machine-readable format. One of the main issues with traditional annotation tools is that they require transforming the PDF into plain text to facilitate the annotation process. While doing so lessens the technical challenges of annotating data, the user loses all structure and provenance that was inherent in the underlying PDF. Also, textual data extraction from PDFs can be an error prone process. Challenges include identifying sequential blocks of text and a multitude of document formats (multiple columns, font encodings, etc.). As a result of these challenges, using existing tools for development of NLP/ML models directly from PDFs is difficult because the generated outputs are not interoperable. We created a system that allows annotations to be completed on the original PDF document structure, with no plain text extraction. The result is an application that allows for easier and more accurate annotations. In addition, by including a feature that grants the user the ability to easily create a schema, we have developed a system that can be used to annotate text for different domain-centric schemas of relevance to subject matter experts. Different knowledge domains require distinct schemas and annotation tags to support machine learning.

97 MATHEMATICS AND COMPUTING↗

Double-differential Measurements of Mesonless Charged-current Muon Neutrino Interactions on Argon with Final-State Protons using the MicroBooNE Detector

This note reports the first double-differential measurements of charged-current $ν_µ$ scattering on argon leading to final states containing zero mesons and one or more protons. This event topology (hereafter abbreviated as CC0$πNp$) is the most common at the neutrino energies typically produced by the Fermilab Booster Neutrino Beam. A detailed understanding of neutrino-argon scattering in the CC0$πNp$ channel is therefore crucial for the success of the precision neutrino oscillation analyses planned for the Short-Baseline Neutrino (SBN) program. This remains true for the upcoming Deep Underground Neutrino Experiment (DUNE), but the higher mean neutrino energy used there will ensure that more inelastic reaction modes, such as single pion production, will also play a major role. The analysis described in this note builds on a previous MicroBooNE study of the CC0$πNp$ channel which obtained the first single-differential cross-section measurements on an argon target. Since that foundational work, significant improvements have been made to MicroBooNE’s simulation software, event reconstruction algorithms, and procedure for calculating systematic uncertainties. When combined with a larger dataset (corresponding to a beam exposure of 6.79 × 10 20 protons-on-target versus 1.60 × 10 20 in Ref. [3]), these enhancements allow the important CC0$πNp$ channel to be studied in more detail. This note begins with a description of the data and simulation samples used as input to the analysis. Section 3 then defines the CC0$πNp$ signal event topology, and Section 4 describes a set of selection criteria designed to identify these events in MicroBooNE data. Binning schemes are then defined in Section 5 for two double-differential measurements of event rates. The first of these considers the momentum and scattering cosine of the outgoing muon, while the second reports the same observables for the leading proton, i.e., the final-state proton with the largest momentum. After a discussion of systematic uncertainties in Section 6, the note concludes by comparing the predictions of MicroBooNE Monte Carlo (MC) simulations to the measured double-differential distributions. These results will form the basis for a future extraction of flux-averaged double-differential CC0$πNp$ cross sections that will be immediately comparable to the theoretical predictions of multiple neutrino event generators. The selection described herein may also be used to study various other observables in CC0$πNp$ events, including those which are sensitive to correlations between leptonic and hadronic kinematics in the final state.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

The ASHRAE Great Energy Predictor III competition: Overview and results

In late 2019, ASHRAE hosted the Great Energy Predictor III (GEPIII) machine learning competition on the Kaggle platform. This launch marked the third energy prediction competition from ASHRAE and the first since the mid-1990s. In this updated version, the competitors were provided with over 20 million points of training data from 2,380 energy meters collected for 1,448 buildings from 16 sources. This competition’s overall objective was to find the most accurate modeling solutions for the prediction of over 41 million private and public test data points. Furthermore, the competition had 4,370 participants, split across 3,614 teams from 94 countries who submitted 39,403 predictions. In addition to the top five winning workflows, the competitors publicly shared 415 reproducible online machine learning workflow examples (notebooks), including over 40 additional, full solutions. This paper gives a high-level overview of the competition preparation and dataset, competitors and their discussions, machine learning workflows and models generated, winners and their submissions, discussion of lessons learned, and competition outputs and next steps. The most popular and accurate machine learning workflows used large ensembles of mostly gradient boosting tree models, such as LightGBM. Similar to the first predictor competition, preprocessing of the data sets emerged as a key differentiator.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗