Search NASA⌕ Search

SEARCH · Search NASA

Results for “Computational Human Performance Modeling”

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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Advocating for view and daylight in buildings: Next steps

With the exponential growth in population and commensurate increased density in urban cities, access to daylight and views to nature is being severely curtailed in buildings. In parallel, increasingly urgent demands to sharply reduce building energy use and associated greenhouse gas emissions are being made to mitigate climate change. There are many challenges and performance tradeoffs associated with the building facade (i.e., daylight and view versus solar and glare control); increased prioritization of health and well-being as a fundamental human requirement could adversely affect building energy-efficiency. Given the current state of knowledge on the effects of daylight and view on health and well-being in buildings, we identify critical needs in research, tools and technologies that if satisfied may enable more effective use of daylight and view in buildings within the constraints of climate change. Lack of knowledge regarding the complex causal mechanisms of window views on human factors is a severely limiting factor in forward progress. Current models and methods to derive bidirectional scattering distribution functions (BSDFs) will need to be modified. Developers of energy-efficient window technologies will need more guidance to shape product development. Advanced window technologies and integrated design can enable attainment of both health and well-being and net zero energy goals, but considerable work will be needed to make such options turnkey and broadly available.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2022

Improved Throughput and Analysis of Scratch Test Results via Automation and Machine Learning

A data analysis automation interface that incorporates machine learning (ML) has been developed to improve productivity, efficiency, and consistency in identifying and defining critical load values (or other values associated with optically identifiable characteristics) of a coating when a scratch test is performed. In this specific program, the machine learning component of the program has been trained to identify the Critical Load 2 (L C2 ) value by analyzing images of the scratch tracks created in each test. An optical examination of the scratch by a human operator is currently used to determine where this value occurs. However, the vagueness of the standard has led to varying interpretations and nonuniform usage by different operators at different laboratories where the test is implemented, resulting in multiple definitions of the desired parameter. Using a standard set of training and validation images to create the dataset, the critical load can be identified consistently amongst different laboratories using the automation interface without requiring the training of human operators. When the model was used in conjunction with an instrument manufacturer's scratch test software, the model produced accurate and repeatable results and defined L C2 values in as little as half of the time compared to a human operator. When combined with a program that automates other aspects of the scratch testing process usually conducted by a human operator, scratch testing and analysis can occur with little to no intervention from a human beyond initial setup and frees them to complete other work in the lab.

36 MATERIALS SCIENCE↗
DOE OSTI2015

Detection of microcalcifications by characteristic magnetic susceptibility effects using MR phase image cross-correlation analysis

Purpose: To develop and evaluate a new method for detecting calcium deposits using their characteristic magnetic susceptibility effects on magnetic resonance (MR) images at high fields and demonstrate its potential in practice for detecting breast microcalcifications. Methods: Characteristic dipole signatures of calcium deposits were detected in magnetic resonance phase images by computing the cross-correlation between the acquired data and a library of templates containing simulated phase patterns of spherical deposits. The influence of signal-to-noise ratio and various other MR parameters on the results were assessed using simulations and validated experimentally. The method was tested experimentally for detection of calcium fragments within gel phantoms and calcium-like inhomogeneities within chicken tissue at 7 T with optimized MR acquisition parameters. The method was also evaluated for detection of simulated microcalcifications, modeled from biopsy samples of malignant breast cancer, inserted in silico into breast magnetic resonance imaging (MRIs) of healthy subjects at 7 T. For both assessments of calcium fragments in phantoms and biopsy-based simulated microcalcifications in breast MRIs, receiver operator characteristic curve analyses were performed to determine the cross-correlation index cutoff, for achieving optimal sensitivity and specificity, and the area under the curve (AUC), for measuring the method’s performance. Results: The method detected calcium fragments with sizes of 0.14–0.79 mm, 1 mm calcium-like deposits, and simulated microcalcifications with sizes of 0.4–1.0 mm in images with voxel sizes between (0.2 mm){sup 3} and (0.6 mm){sup 3}. In images acquired at 7 T with voxel sizes of (0.2 mm){sup 3}–(0.4 mm){sup 3}, calcium fragments (size 0.3–0.4 mm) were detected with a sensitivity, specificity, and AUC of 78%–90%, 51%–68%, and 0.77%–0.88%, respectively. In images acquired with a human 7 T scanner, acquisition times below 12 min, and voxel sizes of (0.4 mm){sup 3}–(0.6 mm){sup 3}, simulated microcalcifications with sizes of 0.6–1.0 mm were detected with a sensitivity, specificity, and AUC of 75%–87%, 54%–87%, and 0.76%–0.90%, respectively. However, different microcalcification shapes were indistinguishable. Conclusions: The new method is promising for detecting relatively large microcalcifications (i.e., 0.6–0.9 mm) within the breast at 7 T in reasonable times. Detection of smaller deposits at high field may be possible with higher spatial resolution, but such images require relatively long scan times. Although mammography can detect and distinguish the shape of smaller microcalcifications with superior sensitivity and specificity, this alternative method does not expose tissue to ionizing radiation, is not affected by breast density, and can be combined with other MRI methods (e.g., dynamic contrast-enhanced MRI and diffusion weighted MRI), to potentially improve diagnostic performance.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

Deep active learning for classifying cancer pathology reports

Abstract Background Automated text classification has many important applications in the clinical setting; however, obtaining labelled data for training machine learning and deep learning models is often difficult and expensive. Active learning techniques may mitigate this challenge by reducing the amount of labelled data required to effectively train a model. In this study, we analyze the effectiveness of 11 active learning algorithms on classifying subsite and histology from cancer pathology reports using a Convolutional Neural Network as the text classification model. Results We compare the performance of each active learning strategy using two differently sized datasets and two different classification tasks. Our results show that on all tasks and dataset sizes, all active learning strategies except diversity-sampling strategies outperformed random sampling, i.e., no active learning. On our large dataset (15K initial labelled samples, adding 15K additional labelled samples each iteration of active learning), there was no clear winner between the different active learning strategies. On our small dataset (1K initial labelled samples, adding 1K additional labelled samples each iteration of active learning), marginal and ratio uncertainty sampling performed better than all other active learning techniques. We found that compared to random sampling, active learning strongly helps performance on rare classes by focusing on underrepresented classes. Conclusions Active learning can save annotation cost by helping human annotators efficiently and intelligently select which samples to label. Our results show that a dataset constructed using effective active learning techniques requires less than half the amount of labelled data to achieve the same performance as a dataset constructed using random sampling.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2025

Detection of Diversion in a Realistic Heat Pipe Microreactor Using Supervised Machine Learning

Microreactors (MRs) pose new challenges for international safeguards. Here, their small size and mass reproducibility make them ideal for deployment in greater numbers and in remote locations, making the job of safeguards inspectors more challenging. Machine learning (ML) is currently being applied to many fields to augment human performance and increase automation; in particular, ML could be used to provide insight for international inspectors to help detect the diversion of nuclear fuel from MR cores. Four ML model types (k-nearest neighbors, decision tree, random forest, and histogram-based gradient boosted ensemble) were trained on integrated flux and critical control drum angle data generated with Serpent 2 for a realistic heat pipe MR design, achieving nearly 100% binary classification accuracy of nominal and diversion core configurations by the end of 1 full power year for three of the four model types. Regression model variants were also trained, using the same input data, for predicting the number of fuel pins diverted. Root-mean-square errors below 5% of the total number of fuel pins were achieved by the 1 full power year mark for all models.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2024

Investigating genomic prediction strategies for grain carotenoid traits in a tropical/subtropical maize panel

Abstract Vitamin A deficiency remains prevalent on a global scale, including in regions where maize constitutes a high percentage of human diets. One solution for alleviating this deficiency has been to increase grain concentrations of provitamin A carotenoids in maize (Zea mays ssp. mays L.)—an example of biofortification. The International Maize and Wheat Improvement Center (CIMMYT) developed a Carotenoid Association Mapping panel of 380 inbred lines adapted to tropical and subtropical environments that have varying grain concentrations of provitamin A and other health-beneficial carotenoids. Several major genes have been identified for these traits, 2 of which have particularly been leveraged in marker-assisted selection. This project assesses the predictive ability of several genomic prediction strategies for maize grain carotenoid traits within and between 4 environments in Mexico. Ridge Regression-Best Linear Unbiased Prediction, Elastic Net, and Reproducing Kernel Hilbert Spaces had high predictive abilities for all tested traits (β-carotene, β-cryptoxanthin, provitamin A, lutein, and zeaxanthin) and outperformed Least Absolute Shrinkage and Selection Operator. Furthermore, predictive abilities were higher when using genome-wide markers rather than only the markers proximal to 2 or 13 genes. These findings suggest that genomic prediction models using genome-wide markers (and assuming equal variance of marker effects) are worthwhile for these traits even though key genes have already been identified, especially if breeding for additional grain carotenoid traits alongside β-carotene. Predictive ability was maintained for all traits except lutein in between-environment prediction. The TASSEL (Trait Analysis by aSSociation, Evolution, and Linkage) Genomic Selection plugin performed as well as other more computationally intensive methods for within-environment prediction. The findings observed herein indicate the utility of genomic prediction methods for these traits and could inform their resource-efficient implementation in biofortification breeding programs.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2020

A neural network for determination of latent dimensionality in Nonnegative Matrix Factorization

Non-negative Matrix Factorization (NMF) has proven to be a powerful unsupervised learning method for uncovering hidden features in complex and noisy datasets with applications in data mining, text recognition, dimension reduction, face recognition, anomaly detection, blind source separation, and many other fields. An important input for NMF is the latent dimensionality of the data, that is, the number of hidden features, K, present in the explored dataset. Unfortunately, and this quantity is rarely known a priori. The existing methods for determining latent dimensionality, such as Automatic Relevance Determination (ARD), are mostly heuristic and utilize different characteristics to estimate the number of hidden features. However, all of them require human presence to make a final determination of K. Here we utilize a supervised machine learning approach in combination with a recent method for model determination, called NMFk, to determine the number of hidden features automatically. NMFk performs a set of NMF simulations on an ensemble of matrices, obtained by bootstrapping the initial dataset, and estimates which K produces stable groups of latent features that reconstruct the initial dataset well. We then train a Multi-Layter Perceptron (MLP) classifier network to determine the correct number of latent features utilizing the statistics and characteristics of the NMF solution, obtained from NMFk. In order to train the MLP classifier, a training set of 58,660 matrices with predetermined latent features were factorized with NMFk. The MLP classifier in conjunction with NMFk maintains a greater than 95% success rate when applied to a held out test set. Additionally, when applied to two well-known benchmark datasets, the swimmer and MIT face data, NMFk/MLP correctly recovers the established number of hidden features. Finally, we compare the accuracy of our method to the ARD, AIC and Stability-based methods.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2016

SU-F-J-59: Assessment of Dose Response Distribution in Individual Human Tumor

Purpose: To fulfill precision radiotherapy via adaptive dose painting by number, voxel-by-voxel dose response or radio-sensitivity in individual human tumor needs to be determined in early treatment to guide treatment adaptation. In this study, multiple FDG PET images obtained pre- and weekly during the treatment course were utilized to determine the distribution/spectrum of dose response parameters in individual human tumors. Methods: FDG PET/CT images of 18 HN cancer patients were used in the study. Spatial parametric image of tumor metabolic ratio (dSUV) was created following voxel by voxel deformable image registration. Each voxel value in dSUV was a function of pre-treatment baseline SUV and treatment delivered dose, and used as a surrogate of tumor survival fraction (SF). Regression fitting with break points was performed using the LQ-model with tumor proliferation for the control and failure group of tumors separately. The distribution and spectrum of radiation sensitivity and growth in individual tumors were determined and evaluated. Results: Spectrum of tumor dose-sensitivity and proliferation in the controlled group was broad with α in tumor survival LQ-model from 0.17 to 0.8. It was proportional to the baseline SUV. Tlag was about 21∼25 days, and Tpot about 0.56∼1.67 days respectively. Commonly tumor voxels with high radio-sensitivity or larger α had small Tlag and Tpot. For the failure group, the radio-sensitivity α was low within 0.05 to 0.3, but did not show clear Tlag. In addition, tumor voxel radio-sensitivity could be estimated during the early treatment weeks. Conclusion: Dose response distribution with respect to radio-sensitivity and growth in individual human tumor can be determined using FDG PET imaging based tumor metabolic ratio measured in early treatment course. The discover is critical and provides a potential quantitative objective to implement tumor specific precision radiotherapy via adaptive dose painting by number.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2019

Scalable Contour Tree Computation by Data Parallel Peak Pruning

As data sets grow to exascale, automated data analysis and visualization are increasingly important, to intermediate human understanding and to reduce demands on disk storage via in situ analysis. Trends in architecture of high performance computing systems necessitate analysis algorithms to make effective use of combinations of massively multicore and distributed systems. One of the principal analytic tools is the contour tree, which analyses relationships between contours to identify features of more than local importance. Unfortunately, the predominant algorithms for computing the contour tree are explicitly serial, and founded on serial metaphors, which has limited the scalability of this form of analysis. While there is some work on distributed contour tree computation, and separately on hybrid GPU-CPU computation, there is no efficient algorithm with strong formal guarantees on performance allied with fast practical performance. Furthermore, we report the first shared SMP algorithm for fully parallel contour tree computation, with formal guarantees of O(lg V lg t) parallel steps and O(V lg V) work for data with V samples and t contour tree supernodes, and implementations with more than 30× parallel speed up on both CPU using TBB and GPU using Thrust and up 70× speed up compared to the serial sweep and merge algorithm.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

A data-centric weak supervised learning for highway traffic incident detection

Using the data from loop detector sensors for near-real-time detection of traffic incidents on highways is crucial to averting major traffic congestion. While recent supervised machine learning methods offer solutions to incident detection by leveraging human-labeled incident data, the false alarm rate is often too high to be used in practice. Specifically, the inconsistency in the human labeling of the incidents significantly affects the performance of supervised learning models. To that end, we focus on a data-centric approach to improve the accuracy and reduce the false alarm rate of traffic incident detection on highways. We develop a weak supervised learning workflow to generate high-quality training labels for the incident data without the ground truth labels, and we use those generated labels in the supervised learning setup for final detection. This approach comprises three stages. First, we introduce a data preprocessing and curation pipeline that processes traffic sensor data to generate high-quality training data through leveraging labeling functions, which can be domain knowledge-related or simple heuristic rules. Second, we evaluate the training data generated by weak supervision using three supervised learning models-random forest, k-nearest neighbors, and a support vector machine ensemble-and long short-term memory classifiers. The results show that the accuracy of all of the models improves significantly after using the training data generated by weak supervision. Third, we develop an online real-time incident detection approach that leverages the model ensemble and the uncertainty quantification while detecting incidents. Finally, we show that our proposed weak supervised learning workflow achieves a high incident detection rate (0.90) and low false alarm rate (0.08).

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

BeyondFingerprinting: AI-guided discovery of robust materials & processes

BeyondFingerprinting was a 2021-2024 Sandia Grand Challenge LDRD exploring the potential to develop new resilient materials and manufacturing processes by taking an artificial-intelligence (AI)-guided approach that integrates human-subject-matter expertise with algorithms enriched with physics-based constraints to unearth process-structure-property correlations. Such algorithms, trained on high-throughput experiments and simulations, are shown to serve as surrogate models that efficiently detect key “fingerprints” in materials data, prognose material performance, and guide effective process improvements. To accelerate broader adoption across mission areas, this AI-guided approach was demonstrated with three complex process-centric exemplars: electroplating, physical vapor deposition, and laser powder bed fusion. Together, these exemplars impact nearly every hardware component relevant to DOE and NNSA national security missions.

36 MATERIALS SCIENCE↗
DOE OSTI2019

Performance analysis and optimization for scalable deployment of deep learning models for country-scale settlement mapping on Titan supercomputer

Here, we present a scalable object detection workflow for detecting objects, such as settlements, from remotely sensed (RS) imagery. We have successfully deployed this workflow on Titan supercomputer and utilized it for the task of mapping human settlement at a country scale. The performance of various stages in the workflow was analyzed before making it operational. The workflow implemented various strategies to address issues such as suboptimal resource utilization and long-tail effects due to unbalanced image workload, data loss due to runtime failures, and maximum wall-time constraints imposed by Titan's job scheduling policy. A mean shift clustering–based static load balancing strategy was implemented, which partitions the image load such that each partition contained similar-sized images. Furthermore, a checkpoint-restart strategy was added in the workflow as a fault-tolerance mechanism to prevent the data losses due to unforeseen runtime failures. The performance of the above-mentioned strategies was observed in various scenarios, such as node failure, exceeding wall time, and successful completion. Using this workflow, we have examined an RS data set that has a spatial resolution of 0.31 m and is comprised of 685 675 km 2 of area of the Republic of Zambia in under six hours using 5426 nodes of the Titan supercomputer.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2016

Recent Advances in Analytical Electron Microscopy for Irradiated Materials

Analytical electron microscopy (AEM) in the scanning transmission electron microscope (STEM) has long been a staple of post-irradiation examination (PIE) and materials development campaigns. Recently, new advances in both STEM hardware and post-processing data-analysis software, along with ancillary advances such as in focused ion beam (FIB) sample preparation, has led to exciting advances in the quantity and quality of data that can be obtained. This talk will summarize recent experiments performed at Oak Ridge National Laboratory (ORNL), especially work on neutron-irradiated materials and fuels taking advantage of a newly-installed latest-generation AEM tool in the ORNL irradiated materials laboratories. ORNL installed a new FEI Talos F200X microscope, a 200 keV, high-brightness Schottky emission AEM system optimized for high-speed, high-efficiency X-ray mapping experiments. The F200X features the FEI 'SuperX' four-sector silicon drift detector (SDD) X-ray detector, which provides 0.9 sr solid angle of collection. By combining the high-brightness electron source, latest-generation electron optics, and fast computers and pulse processors, X-ray mapping speed is increased by a factor of 50-100 x compared to the previous-generation instruments. In addition to improvements in hardware, data mining via multivariate statistical analysis (MVSA) allows for maximum insight to the data obtained from the new hardware. MVSA data mining of the X-ray maps allows, for instance, unbiased computational identification of the phases present, and discovery of small phases or precipitates not expected by the human analyst. This talk will present examples of experiments performed using the most modern AEM tools and software. Silicon carbide is a major candidate for next-generation nuclear applications. Nano-infiltration by transient eutectic (NITE) SiC requires low sintering temperatures. X-ray mapping of the NITE structure find pockets of different oxides, such as Zr-Y-oxide, Y-Al-oxide, and yttria. MVSA data mining finds the regions where different oxides are present and indicates that Y additions preferentially wet grain boundaries, where Zr additions are preferentially concentrated at grain triple points. SiC coatings from AGR-I experiment TRISO particles were found to contain significant quantities of fission products (e.g., Pd) as well as actinides (U, Pu). MVSA data mining found complex precipitates and grain boundary films of the fission products and actinides deep (∼several microns) into the SiC layer. Advanced AEM methods, combined with MVSA data mining, makes possible both rapid and detailed PIE. Ferritic alloys, such as accident-tolerant FeCrAl grades, benefit greatly from advanced AEM examination. Neutron-irradiated FeCrAl precipitates the detrimental Cr-rich α' phase. Because the SuperX SDD system has four detectors, it is possible to obtain high X-ray count rates even on a highly tilted specimen, as might be needed to obtain STEM images of dislocation loops. Results on neutron-irradiated FeCrAl indicates that α' size and number density can be obtained by AEM, and MVSA results determine that minor elements (such as Si) has complex radiation-induced segregation behavior at grain boundaries. We have also developed and validated computer models of the AEM experiment, allowing different microstructures' X-ray map results to be simulated for different STEM instruments. This allows X-ray mapping to be more quantitative, such as by estimating what size nanoclusters in an oxide-dispersion strengthened steel would be visible or invisible in STEM. In summary, new advanced AEM technology provides a major leap forward in experiments possible in PIE of irradiated materials. (authors)

36 MATERIALS SCIENCE↗
DOE OSTI2020

Exploratory Investigation of Disengagements and Crashes in Autonomous Vehicles Under Mixed Traffic: An Endogenous Switching Regime Framework

Autonomous Vehicles (AVs) have a large potential to improve traffic safety but also pose some critical challenges. While AVs may help reduce crashes caused by human error, they still may experience failures of technologies and sensing, as well as decision-making errors in a mixed traffic environment. A disengagement refers to an AV transitioning control from autonomous systems to the trained test driver. The safety critical nature of disengagements makes it imperative to analyze disengagements and crashes together. In this study, we analyze both crashes and disengagements from real-world AV driving in California to fill the knowledge gap regarding the relationship between disengagements and crashes in a mixed traffic environment. A nested logit model was calibrated using three different outcomes: (1) disengagement with a crash, (2) disengagement with no crash, and (3) no disengagement with a crash. Furthermore, endogenous switching regime models were also calibrated to draw distinctions between the relation of disengagements and crashes while accounting for endogeneity effects. The results show that factors related to AV systems (such as software failures) and other roadway participants increase the propensity of a disengagement without a crash. Furthermore, AVs were observed to disengage less often as the technology matured over time. Marginal effects revealed an 8% decrease. The results thus suggest that disengagements are a part of AVs' safe performance and disengagement alerts may need to be triggered in order to avoid certain failures with current technology.

42 ENGINEERING↗
DOE OSTI2021

Semi-supervised Bayesian Low-shot Learning

Deep neural networks (NNs) typically outperform traditional machine learning (ML) approaches for complicated, non-linear tasks. It is expected that deep learning (DL) should offer superior performance for the important non-proliferation task of predicting explosive device configuration based upon observed optical signature, a task which human experts struggle with. However, supervised machine learning is difficult to apply in this mission space because most recorded signatures are not associated with the corresponding device description, or “truth labels.” This is challenging for NNs, which traditionally require many samples for strong performance. Semi-supervised learning (SSL), low-shot learning (LSL), and uncertainty quantification (UQ) for NNs are emerging approaches that could bridge the mission gaps of few labels and rare samples of importance. NN explainability techniques are important in gaining insight into the inferential feature importance of such a complex model. In this work, SSL, LSL, and UQ are merged into a single framework, a significant technical hurdle not previously demonstrated. Exponential Average Adversarial Training (EAAT) and Pairwise Neural Networks (PNNs) are chosen as the SSL and LSL methods of choice. Permutation feature importance (PFI) for functional data is used to provide explainability via the Variable importance Explainable Elastic Shape Analysis (VEESA) pipeline. A variety of uncertainty quantification approaches are explored: Bayesian Neural Networks (BNNs), ensemble methods, concrete dropout, and evidential deep learning. Two final approaches, one utilizing ensemble methods and one utilizing evidential learning, are constructed and compared using a well-quantified synthetic 2D dataset along with the DIRSIG Megascene.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2017

Cooling efficiency of a spot-type personalized air-conditioner

Here, this study defined Cooling Efficiency ( CE ) of a Spot-type Personalized Air-Conditioning (SPAC) device as the ratio of the additional sensible heat removal from human body induced by SPAC and the device's cooling capacity. CE enabled the investigation of SPAC performance on the occupant's sensible heat loss (Q s ) and thermal sensation by its quantitative relation with the change of PMV level ( ΔPMV ). Three round nozzles with the diameter of 0.08 m, 0.105 m, and 0.128 m, respectively, discharged air jets at airflow rates from 11.8 L s –1 to 59.0 L s –1 , toward the chest of a seated or standing human body with a clothing of 0.48 clo. This study developed a validated CFD model coupled with the Fanger's thermoregulation model, to calculate Q s in a room of 26 °C ventilated at a rate of 3 ACH. According to the results, Q s , CE and draft risk ( DR ) at face had significant positive linear correlation with the SPAC device's supply airflow rates (R2 >0.96), and a negative linear correlation for ΔPMV. With DR = 20% at face, CE was always under 0.3, and ΔPMV was around -1.0–1.1. Interestingly, both CE and ΔPMV had the least favorable values for the air jet produced by the nozzle with the diameter of 0.105 m independent of body posture. In conclusion, although SPAC could lead to additional Q s by sending air at a higher airflow rate from a smaller nozzle, the improvement in cooling efficiency and thermal sensation had a limit due to draft risk.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2024

Harnessing large language models’ zero-shot and few-shot learning capabilities for regulatory research

Abstract Large language models (LLMs) are sophisticated AI-driven models trained on vast sources of natural language data. They are adept at generating responses that closely mimic human conversational patterns. One of the most notable examples is OpenAI's ChatGPT, which has been extensively used across diverse sectors. Despite their flexibility, a significant challenge arises as most users must transmit their data to the servers of companies operating these models. Utilizing ChatGPT or similar models online may inadvertently expose sensitive information to the risk of data breaches. Therefore, implementing LLMs that are open source and smaller in scale within a secure local network becomes a crucial step for organizations where ensuring data privacy and protection has the highest priority, such as regulatory agencies. As a feasibility evaluation, we implemented a series of open-source LLMs within a regulatory agency’s local network and assessed their performance on specific tasks involving extracting relevant clinical pharmacology information from regulatory drug labels. Our research shows that some models work well in the context of few- or zero-shot learning, achieving performance comparable, or even better than, neural network models that needed thousands of training samples. One of the models was selected to address a real-world issue of finding intrinsic factors that affect drugs' clinical exposure without any training or fine-tuning. In a dataset of over 700 000 sentences, the model showed a 78.5% accuracy rate. Our work pointed to the possibility of implementing open-source LLMs within a secure local network and using these models to perform various natural language processing tasks when large numbers of training examples are unavailable.

Biochemistry & Molecular Biology↗
DOE OSTI2026

Hierarchical semi-Markov models with duration-aware dynamics for activity sequences

Residential electricity demand at granular scales is driven by what people do and for how long. Accurately forecasting this demand for applications like microgrid management and demand response therefore requires generative models for activities that can produce realistic daily activity sequences, capturing both the timing and duration of human behavior. This paper develops a generative model of human activity sequences using nationally representative time-use diaries at a 10-min resolution. We use this model to quantify which demographic factors are most critical for improving predictive performance. We propose a hierarchical semi-Markov framework that addresses two key modeling challenges. First, a time-inhomogeneous Markov router learns the patterns of “which activity comes next.” Second, a semi-Markov hazard component explicitly models activity durations, capturing “how long” activities realistically last. To ensure statistical stability when data are sparse, the model pools information across related demographic groups and time blocks. The entire framework is trained and evaluated using survey design weights to ensure our findings are representative of the U.S. population. On a held-out test set, we demonstrate that explicitly modeling durations with the hazard component provides a substantial and statistically significant improvement over purely Markovian models. Furthermore, our analysis reveals a clear hierarchy of demographic factors: Sex, Day-Type, and Household Size provide the largest predictive gains, while Region and Season, though important for energy calculations, contribute little to predicting the activity sequence itself. The result is an interpretable and robust generator of synthetic activity traces, providing a high-fidelity foundation for downstream energy systems modeling.

24 POWER TRANSMISSION AND DISTRIBUTION↗