Search NASA⌕ Search

SEARCH · Search NASA

Results for “synthetic 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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Reconstruction of the Density Power Spectrum from Quasar Spectra using Machine Learning

We describe a novel end-to-end approach using Machine Learning to reconstruct the power spectrum of cosmological density perturbations at high redshift from observed quasar spectra. State-of-the-art cosmological simulations of structure formation are used to generate a large synthetic dataset of line-of-sight absorption spectra paired with 1-dimensional fluid quantities along the same line-of-sight, such as the total density of matter and the density of neutral atomic hydrogen. With this dataset, we build a series of data-driven models to predict the power spectrum of total matter density. We are able to produce models which yield reconstruction to accuracy of about 1% for wavelengths $k \leq 2 h Mpc^{-1}$, while the error increases at larger $k$. We show the size of data sample required to reach a particular error rate, giving a sense of how much data is necessary to reach a desired accuracy. This work provides a foundation for developing methods to analyse very large upcoming datasets with the next-generation observational facilities.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2026

Probing the matter-dominated expansion with multi-redshift Lyman-$α$ BAO from DESI DR2

We present a multi-redshift Baryon Acoustic Oscillations (BAO) analysis of the DESI Data Release 2 (DR2) Lyman-$α$ (Ly$α$) forest, splitting the forest auto-correlation and its cross-correlation with quasars into three redshift bins. We obtain BAO measurements at effective redshifts $z_{\rm eff} = 2.13$, $2.40$, and $2.81$ with $\sim2.0$--$2.5\%$ precision per bin in the radial and transverse directions, corresponding to $\sim1.1$--$1.2\%$ precision for the isotropic BAO measurement. Using the same data products and modeling framework as the DESI DR2 Ly$α$ BAO analysis, we validate the pipeline on $400$ synthetic datasets and find unbiased BAO recovery with well-calibrated uncertainties. The measurements show an increase in the isotropic dilation parameter $D_V/r_d$ from $30.26\pm0.39$ to $32.22\pm0.47$ and in the Alcock-Paczyński parameter $D_M/D_H$ from $3.96\pm0.15$ to $5.63^{+0.22}_{-0.24}$. The Hubble distance $D_H/r_d$ decreases from $9.40\pm0.20$ to $7.22\pm0.17$, providing a direct measurement of the expansion history consistent with $Λ$CDM and the expected matter-dominated scaling, with $H(z)\propto(1+z)^n$ giving $n=1.34\pm0.16$. The redshift split also provides a self-consistent measurement of clustering evolution: the Ly$α$ forest bias evolves as $(1+z)^γ$ with $γ_α=3.05\pm0.16$, the RSD parameter has a redshift evolution described by $γ_β=-0.97\pm0.26$, and the quasar bias evolves with $γ_Q=1.56\pm0.23$, consistent with independent quasar clustering measurements. Combining these three-bin BAO measurements with DESI DR2 galaxy and quasar BAO measurements yields cosmological constraints consistent with the single-bin Ly$α$ BAO analysis in flat $Λ$CDM and $w_0w_a$CDM and improves curvature constraints by $\sim12\%$ in $Λ$CDM$+Ω_\mathrm{K}$.

Herrera-Alcantar, Hiram K. [Paris, Inst. Astrophys↗
DOE OSTI2022

Physics-Based Method for Generating Fully Synthetic IV Curve Training Datasets for Machine Learning Classification of PV Failures

Classification machine learning models require high-quality labeled datasets for training. Among the most useful datasets for photovoltaic array fault detection and diagnosis are module or string current-voltage (IV) curves. Unfortunately, such datasets are rarely collected due to the cost of high fidelity monitoring, and the data that is available is generally not ideal, often consisting of unbalanced classes, noisy data due to environmental conditions, and few samples. In this paper, we propose an alternate approach that utilizes physics-based simulations of string-level IV curves as a fully synthetic training corpus that is independent of the test dataset. In our example, the training corpus consists of baseline (no fault), partial soiling, and cell crack system modes. The training corpus is used to train a 1D convolutional neural network (CNN) for failure classification. The approach is validated by comparing the model’s ability to classify failures detected on a real, measured IV curve testing corpus obtained from laboratory and field experiments. Results obtained using a fully synthetic training dataset achieve identical accuracy to those obtained with use of a measured training dataset. When evaluating the measured data’s test split, a 100% accuracy was found both when using simulations or measured data as the training corpus. When evaluating all of the measured data, a 96% accuracy was found when using a fully synthetic training dataset. The use of physics-based modeling results as a training corpus for failure detection and classification has many advantages for implementation as each PV system is configured differently, and it would be nearly impossible to train using labeled measured data.

Hopwood, Michael W. (ORCID:0000000161901767)↗
DOE OSTI2025

Using Neural Networks to Identify Mixture Components in Hyperspectral Reflectance Data

Neural networks have been employed to identify materials of interest from hyperspectral data (generally imagery) based on their unique spectral signatures. This approach assumes that there is a single material that is standing out from the rest of the spectrum to be identified. However, pixels often contain more than one material, or a material of interest may itself be a mixture of multiple materials. Neural networks are only as good as the data used to train them, and it takes a great deal of work in the laboratory to identify, make, and measure all potential mixtures of interest. Thus, researchers often calculate synthetic spectra using algorithms with varying degrees of fidelity to the physics that govern the interactions between light and multiple materials. In this work, we have (1) adapted a neural network designed to identify mixture components from Raman spectroscopy to work with visible to near‐infrared reflectance data and (2) tested three common mixture algorithms to determine the most accurate and least computationally expensive method to build synthetic training datasets. With our initial test dataset, we have achieved accuracies of > 90% and found that the synthetic training dataset produced using the Hapke mixture model provides the best results.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2023

Characterizing Quantum Classifier Utility in Natural Language Processing Workflows

Quantum Natural Language Processing (QNLP) develops natural language processing (NLP) models for deployment on quantum computers. We explore feature and data prototype selection techniques to address challenges posed by encoding high dimensional features. Our study builds quantum circuit classifiers that includes classical feature pre-processing, quantum embedding and quantum model training. The quantum models are built on 4 or 6 qubits and the quantum neural network (QNN) uses the established bricklayer design. We compare the dependence of model performance (in terms of accuracy and F1 scores) on feature length, embedding gates and parameterized unitary design. We compare the performance of quantum machine learning models to classical convolution neural network model (CNN) on binary and multi-class classification tasks using two datasets of synthetic features and labels. The first is the ECP-CANDLE P3B3 dataset a corpus of synthetically generated cancer pathology reports. The second dataset is extracted from well-known benchmark dataset (MADELON) - features are generated with a combination of informative, repeated and uninformative features. Both datasets are used for binary classification and multi-class classification with 3 classes. We observe robust, accurate performance from all models on the binary classification tasks, but multiclass classification is a challenge for the quantum models-there is a notable decrease in accuracy when using 3 classes. Overall the performance is comparable in terms of recall and accuracy between QNNs and CNNs, even with large datasets. These results provide a point of comparison between quantum and classical models on real-world datasets.

Hamilton, Kathleen↗
DOE OSTI2023

pnnl/mlxrd

A multi-task neural network for the simultaneous identification of multiple compounds from XRD data obtained in a synchrotron-based hydrothermal fluid system. To enhance the alignment between the training and test datasets, random noise is introduced into the synthetic training dataset, and minor labeled experimental data is incorporated. Additionally, we employ a weighted cross-entropy loss function to address data imbalances and ensure smoother model training.

Li, Ang↗
DOE OSTI2022

Synthetic Data Generation Using Machine Learning

Robust machine learning techniques for image analysis require a substantial amount of data to yield confident results. In the nuclear domain, data scarcity is a substantial challenge because there are so few facilities worldwide. This research focuses on being able alleviate the data scarcity problem by generating synthetic data to bridge the gap between large and small datasets. This work achieves that goal using a Generative Adversarial Network (GAN) architectural approach, by training a model on real-world data and expands that small dataset through synthetic data amendments. Model performance is impacted by the size of the real-world dataset and the number of training epochs utilized. This means that 1) It is important to develop your GAN to be optimized with the specific data type, and 2) approaches taken when training the GAN should be specialized to encompass important aspects of the dataset that it is generating. By taking a step to improve dataset sizes in this way, the gap between models trained by parties with significant amount of data and those without access to large data, closes, allowing for robust analyses of satellite imagery for nuclear domain applications.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2021

2021 Smoky Mountains Conference Data Challenge Synthetic-to-Real Domain Adaptation for Autonomous Driving Dataset

The dataset is comprised of both real and synthetic images from a vehicle's forward-facing camera. Each camera image is accompanied by a corresponding pixel-level semantic segmentation image (all files are .png files). In total, the dataset contains 5600 images in the training/validation set and 1400 images in the testing set. The training dataset contains mostly synthetic RGB images collected with a wide range of weather and lighting conditions using the CARLA simulator [1]. In addition, the training data also includes a small pre-selected subset of data from the Cityscapes training dataset – which is comprised of RGB-segmentation image pairs from driving scenarios in various European cities [2]. The testing data is split into three sets. The first set contains synthetic CARLA images with weather/lighting conditions that were not present in the training set. The second set is a subset of the Cityscapes testing dataset. Finally, the third set is an unknown testing set which will not be revealed to the participants until after the submission deadline. [1] Dosovitskiy, A., Ros, G., Codevilla, F., Lopez, A., and Koltun, V. (2017, October). CARLA: An open urban driving simulator. In Conference on robot learning (pp. 1-16). PMLR. [2] Cordts, M., Omran, M., Ramos, S., Rehfeld, T., Enzweiler, M., Benenson, R., ... and Schiele, B. (2016). The cityscapes dataset for semantic urban scene understanding. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3213-3223).

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2021

An ontology to represent synthetic building occupant characteristics and behavior

Since the introduction of the occupant behavior Drivers-Needs-Actions-Systems (DNAS) framework in 2013, researchers have used the framework or further developed it based on their case studies, which include efforts to collect new data on occupant behaviors. The effort is often costly for the relatively few new data points added. Problems emerge when the already collected data do not meet the modelers' interoperability requirements. Previous studies addressed this issue by developing more sophisticated ontologies that enable integration with other datasets and synthetic data methodologies that would meet unique research applications. This paper presents an extension of the DNAS framework for the representation of synthetic occupant data to support various applications and use cases across the building life cycle. An agent-based modeling application is one of our motivations that requires more elaborate characteristics of an occupant-agent or a group-of-agent. The extension, built upon a review of the literature, introduces new elements to the framework that fall into five categories, including socio-economic, geographical location, activities, subjective values, and individual and collective adaptive actions. On-going research includes identifying occupant datasets and developing data fusion methods to generate synthetic occupants, as well as to demonstrate its applications in agent-based modeling coupled with building performance simulation.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗
NASA NTRSDate not supplied

Optimizing a Small RNAseq Analysis Pipeline for NASA GeneLab Using Open-Source Tools and Libraries

Small RNA sequencing (small RNAseq) is a powerful tool for studying the regulation of gene expression in various organisms. Small RNAseq has been leveraged in space biology research to study how expression of small RNAs, e.g. micro RNAs (miRNAs), small interfering RNAs (siRNAs), and piwi-interacting RNAs (piRNAs), change upon exposure to the space environment. NASA GeneLab currently hosts small RNAseq raw data derived from space-relevant experiments on the Open Science Data Repository (OSDR). To maximize the accessibility of these data to the scientific community, in addition to hosting raw data, which is only interpretable by bioinformaticians, GeneLab plans to process all small RNAseq datasets and make those processed data available to the scientific community via the OSDR. In this study, we present the development of the GeneLab standardized pipeline for processing small RNAseq datasets. Using human, plant, and synthetic small RNAseq datasets, we interrogate various open-source software and publicly available databases to evaluate their accuracy and reproducibility in each step of the pipeline. For quality control and adapter detection and trimming, we evaluated TrimGalore!, FASTX, SeqKit, and DNApi methods to optimize alignment to reference genomes. We compared BWA, Bowtie, and Bowtie2 to determine the optimal alignment tool. For each alignment tool we also assessed various reference databases, including Ensembl reference genomes and different types of small RNA reference databases, including genome, hairpin, and miRNA references from the miRbase and MirGeneDB databases. To quantify the aligned data, we compared SAMtools, HTSeq, and RSEM for counting alignment events from each alignment tool used. Finally, we evaluated various tools, including DESeq2 and EdgeR, for data normalization and subsequent differential expression analysis. We will present the results from our comparative analyses for each pipeline step and propose a consensus pipeline for processing small RNAseq data derived from various organisms exposed to the space environment.

SmallRNAseq, NASA GeneLab, quality control, adapte↗
DOE OSTI2021

Deep-Learning-Derived Evaluation Metrics Enable Effective Benchmarking of Computational Tools for Phosphopeptide Identification

Tandem mass spectrometry (MS/MS)-based phosphoproteomics is a powerful technology for global phosphorylation analysis. However, applying four computational pipelines to a typical mass spectrometry (MS)-based phosphoproteomic dataset from a human cancer study, we observed a large discrepancy among the reported phosphopeptide identification and phosphosite localization results, underscoring a critical need for benchmarking. While efforts have been made to compare performance of computational pipelines using data from synthetic phosphopeptides, evaluations involving real application data have been largely limited to comparing the numbers of phosphopeptide identifications due to the lack of appropriate evaluation metrics. We investigated three deep learning-derived features as potential evaluation metrics: phosphosite probability, Delta RT and spectral similarity. Predicted phosphosite probability is computed by MusiteDeep, which provides high accuracy as previously reported; Delta RT is defined as the absolute retention time (RT) difference between RTs observed and predicted by AutoRT; and spectral similarity is defined as the Pearson’s correlation coefficient between spectra observed and predicted by pDeep2. Using a synthetic peptide dataset, we found that both Delta RT and spectral similarity provided excellent discrimination between correct and incorrect peptide-spectrum matches (PSMs) both when incorrect PSMs involved wrong peptide sequences and even when incorrect PSMs were caused by only incorrect phosphosite localization. Based on these results, we used all the three deep learning-derived features as evaluation metrics to compare different computational pipelines on diverse set of phosphoproteomic datasets and showed their utility in benchmarking performance of the pipelines. The benchmark metrics demonstrated in this study will enable users to select computational pipelines and parameters for routine analysis of phosphoproteomics data and will offer guidance for developers to improve computational methods.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Modeling Air Handling Units to Create a Diverse Fault Dataset for FDD Innovation: Lessons Learned and Recommendations

As energy management and information systems (e.g., automated fault detection and diagnostics [AFDD] tools) become more prevalent in the commercial building stock, it is important to determine the effectiveness of these technologies by benchmarking their performance. The authors have been working to develop the largest publicly available dataset of HVAC fault datasets for performance benchmarking applications, covering the most common HVAC systems and designs including chiller plants, rooftop packaged units, dual duct air handling unit and single duct air handling units. This study covers the development, modeling, and validation of a synthetic fault dataset for the air handling unit (AHU), one of the most common HVAC configurations found in the commercial building stock. Despite this being a common system, real-world time series data are scarce and usually do not span a wide range of weather conditions. Due to this limitation, two detailed AHU models, which included the single duct AHU and dual duct AHU developed in the Modelica language and HVACSIM+ were employed to carry out annual simulations of numerous common sensor faults, mechanical faults, and control sequence faults. The fault inclusive data were then validated by comparing fault effects on system performance to expected symptoms. We summarize the nature of each fault and their impacts under different weather and operation conditions. We report some lessons learnt during the efforts of validating the high volumes of the FDD data sets. Finally, we highlight considerations for FDD developers that may want to use this dataset to assess their algorithms’ performance and their improvement over time.

Casillas, Armando↗
DOE OSTI2022

Inference-Optimized AI and High Performance Computing for Gravitational Wave Detection at Scale

We introduce an ensemble of artificial intelligence models for gravitational wave detection that we trained in the Summit supercomputer using 32 nodes, equivalent to 192 NVIDIA V100 GPUs, within 2 h. Once fully trained, we optimized these models for accelerated inference using NVIDIA TensorRT. We deployed our inference-optimized AI ensemble in the ThetaGPU supercomputer at Argonne Leadership Computer Facility to conduct distributed inference. Using the entire ThetaGPU supercomputer, consisting of 20 nodes each of which has 8 NVIDIA A100 Tensor Core GPUs and 2 AMD Rome CPUs, our NVIDIA TensorRT-optimized AI ensemble processed an entire month of advanced LIGO data (including Hanford and Livingston data streams) within 50 s. Our inference-optimized AI ensemble retains the same sensitivity of traditional AI models, namely, it identifies all known binary black hole mergers previously identified in this advanced LIGO dataset and reports no misclassifications, while also providing a 3X inference speedup compared to traditional artificial intelligence models. We used time slides to quantify the performance of our AI ensemble to process up to 5 years worth of advanced LIGO data. In this synthetically enhanced dataset, our AI ensemble reports an average of one misclassification for every month of searched advanced LIGO data. We also present the receiver operating characteristic curve of our AI ensemble using this 5 year long advanced LIGO dataset. This approach provides the required tools to conduct accelerated, AI-driven gravitational wave detection at scale.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Development of a Annual Air Handling Unit Fault Dataset for FDD Tools: Lessons Learned and Considerations for FDD Developers

As energy management and information systems (e.g., automated fault detection and diagnostics [AFDD] tools) become more prevalent in the commercial building stock, it is important to determine the effectiveness of these technologies by benchmarking their performance. The authors have been working to develop the largest publicly available dataset of HVAC fault data for performance benchmarking applications, covering the most common HVAC systems and designs including chiller plants, rooftop packaged units, dual duct air handling units and single duct air handling units. This study covers the development, modeling, and validation of a synthetic fault dataset for a single duct air handling unit (AHU), one of the most common HVAC configurations found in the commercial building stock. Despite this being a common system, real-world time series data are scarce and usually do not span a wide range of weather conditions. Due to this limitation, a detailed AHU model was employed to carry out annual simulations of numerous common sensor and mechanical faults, which were then validated by comparing their effects on system performance to expected symptoms. We summarize the nature of each fault and their impacts under different weather and operation conditions. Finally, we highlight considerations for FDD developers that may want to use this dataset to assess their algorithms’ performance and their improvement over time.

Casillas, Armando↗
DOE OSTI2026

wa-hls4ml: A Benchmark and Surrogate Models for hls4ml Resource and Latency Estimation

As machine learning (ML) is increasingly implemented in hardware to address real-time challenges in scientific applications, the development of advanced toolchains has significantly reduced the time required to iterate on various designs. These advancements have solved major obstacles, but also exposed new challenges. For example, processes that were not previously considered bottlenecks, such as hardware synthesis, are becoming limiting factors in the rapid iteration of designs. To mitigate these emerging constraints, multiple efforts have been undertaken to develop an ML-based surrogate model that estimates resource usage of ML accelerator architectures. We introduce wa-hls4ml, a benchmark for ML accelerator resource and latency estimation, and its corresponding initial dataset of over 680,000 fully connected and convolutional neural networks, all synthesized using hls4ml and targeting Xilinx FPGAs. The benchmark evaluates the performance of resource and latency predictors against several common ML model architectures, primarily originating from scientific domains, as exemplar models, and the average performance across a subset of the dataset. Additionally, we introduce GNN- and transformer-based surrogate models that predict latency and resources for ML accelerators. We present the architecture and performance of the models and find that the models generally predict latency and resources for the 75% percentile within several percent of the synthesized resources on the synthetic test dataset.

Hawks, Benjamin [Fermilab] (ORCID:0000000157000288↗
DOE OSTI2025

wa-hls4ml: A Benchmark and Surrogate Models for hls4ml Resource and Latency Estimation

As machine learning (ML) is increasingly implemented in hardware to address real-time challenges in scientific applications, the development of advanced toolchains has significantly reduced the time required to iterate on various designs. These advancements have solved major obstacles, but also exposed new challenges. For example, processes that were not previously considered bottlenecks, such as hardware synthesis, are becoming limiting factors in the rapid iteration of designs. To mitigate these emerging constraints, multipleefforts have been undertaken to develop an ML-based surrogate model that estimates resource usage of synthesized ML accelerator architectures. We introduce wa-hls4ml, a benchmark for ML accelerator resource and latency estimation, and its corresponding initial dataset of over 680 000 fully connected and convolutional neural networks, all synthesized using hls4ml and targeting Xilinx FPGAs. The benchmark evaluates the performance of resource and latency predictors against several common ML model architectures, primarily originating from scientific domains, as exemplar models, and the average performance across a subset of the dataset. Additionally, we introduce GNN- and transformer-based surrogate models that predict latency and resources for ML accelerators. We present the architecture and performance of the models and find that the models generally predict latency and resources for the 75% percentile within several percent of the synthesized resources on the synthetic test dataset.

Hawks, Benjamin G. [Fermilab]↗
DOE OSTI2021

Novel estimation of aerosol processes with particle size distribution measurements: a case study with the TOMAS algorithm v1.0.0

Atmospheric aerosol microphysical processes are a significant source of uncertainty in predicting climate change. Specifically, aerosol nucleation, emissions, and growth rates, which are simulated in chemical transport models to predict the particle size distribution, are not understood well. However, long-term size distribution measurements made at several ground-based sites across Europe implicitly contain information about the processes that created those size distributions. This work aims to extract that information by developing and applying an inverse technique to constrain aerosol emissions as well as nucleation and growth rates based on hourly size distribution measurements. We developed an inverse method based upon process control theory into an online estimation technique to scale aerosol nucleation, emissions, and growth so that the model–measurement bias in three measured aerosol properties exponentially decays. The properties, which are calculated from the measured and predicted size distributions, used to constrain aerosol nucleation, emission, and growth rates are the number of particles with a diameter between 3 and 6 nm, the number with a diameter greater than 10 nm, and the total dry volume of aerosol ( N 3–6 , N 10 , V dry ), respectively. In this paper, we focus on developing and applying the estimation methodology in a zero-dimensional “box” model as a proof of concept before applying it to a three-dimensional simulation in subsequent work. The methodology is first tested on a dataset of synthetic and perfect measurements that span diverse environments in which the true particle emissions, growth, and nucleation rates are known. The inverse technique accurately estimates the aerosol microphysical process rates with an average and maximum error of 2 % and 13 %, respectively. Next, we investigate the effect that measurement noise has on the estimated rates. The method is robust to typical instrument noise in the aerosol properties as there is a negligible increase in the bias of the estimated process rates. Finally, the methodology is applied to long-term datasets of in situ size distribution measurements in western Europe from May 2006 through June 2007. At Melpitz, Germany, and Hyytiälä, Finland, the average diurnal profiles of estimated 3 nm particle formation rates are reasonable, having peaks near noon local time with average peak values of 1 and 0.15 cm –3 s –1 , respectively. The normalized absolute error in estimated N 3–6 , N 10 , and V dry at three European measurement sites is less than 15 %, showing that the estimation framework developed here has potential to decrease model–measurement bias while constraining uncertain aerosol microphysical processes.

54 ENVIRONMENTAL SCIENCES↗