Search NASA⌕ Search

SEARCH · Search NASA

Results for “evaluation datasets”

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

Boosting Noise2Inverse via enhanced model selection for denoising computed tomography data

Synchrotron-based x-ray tomographic imaging enables the examination of the internal structure of materials at high spatial and temporal resolution. Experimental constraints can impose dose and time limits on the measurements, introducing a higher level of noise and artifacts in the reconstructed images. Deep learning has emerged as a powerful tool to remove noise from reconstructed images. Recently, the Noise2Inverse method was designed specifically for denoising reconstructed images without requiring paired noisy and clean images. This method creates multiple statistically independent reconstructions used to pair the data in which training involves transforming one reconstruction into the other, and vice versa. Originally designed to be used after a fixed number of epochs, we see in practice that this approach may not produce the optimal model and may unnecessarily waste computational resources. Therefore, we propose an alternative method of identifying the best model during training that aligns with the Noise2Inverse method. During validation, we compare the model output of the multiple reconstructions among each other. We hypothesize that the best model is the one that produces images with the highest similarity, implying a convergence in the predicted material properties and absorption values. To compare model outputs, we consider the absolute error, square error, structural similarity index (SSIM), peak signal-to-noise ratio (PSNR), and cosine similarity. We evaluate our method on two simulated tomography datasets and two, real-world, low-contrast, high-energy x-ray tomography datasets. We show our approach is more effective at determining the best model, up to an increase of 12.50% and 12.53% in SSIM and PSNR, respectively, while only requiring a fifth of the training time compared to the original approach.

CT↗

Understanding important features of deep learning models for segmentation of high-resolution transmission electron microscopy images

Cutting edge deep learning techniques allow for image segmentation with great speed and accuracy. However, application to problems in materials science is often difficult since these complex models may have difficultly learning meaningful image features that would enable extension to new datasets. In situ electron microscopy provides a clear platform for utilizing automated image analysis. In this work, we consider the case of studying coarsening dynamics in supported nanoparticles, which is important for understanding, for example, the degradation of industrial catalysts. By systematically studying dataset preparation, neural network architecture, and accuracy evaluation, we describe important considerations in applying deep learning to physical applications, where generalizable and convincing models are required. With a focus on unique challenges that arise in high-resolution images, we propose methods for optimizing performance of image segmentation using convolutional neural networks, critically examining the application of complex deep learning models in favor of motivating intentional process design.

36 MATERIALS SCIENCE↗

Hybrid feature selection approach for power transformer fault diagnosis based on support vector machine and genetic algorithm

To further improve fault diagnosis accuracy, a new hybrid feature selection approach combined with a genetic algorithm (GA) and support vector machine (SVM) is presented in this study. Adaptive synthetic technique and arctangent transformation method are adopted to improve the statistical property of the training set (IEC TC10 dataset). Five filter methods based on different evaluation metrics are employed to rank 48 input features derived from dissolved gas analysis (DGA). Then, feature combination methods are applied to aggregate feature ranks and form a lower-dimension candidate feature subset. The GA-SVM model is implemented to optimise parameters and select optimal feature subsets. 5-fold cross-validation accuracy of the GA-SVM is used to evaluate fault diagnosis capability of feature subsets and finally, a novel subset is determined as the optimal feature subset. Accuracy comparison manifests the superiority of the optimal feature subsets over that of conventional approaches. Besides, generalisation and robustness of the optimal subset are validated by testing DGA samples from the local power utility. Results indicate that the optimal feature subset obtained by the proposed method can significantly improve the accuracies of power transformer fault diagnosis.

Kari, Tusongjiang↗

Primordial features due to a step in the inflaton potential

Certain oscillatory features in the primordial scalar power spectrum are known to provide a better fit to the outliers in the cosmic microwave background data near the multipole moments of l = 22 and 40. These features are usually generated by introducing a step in the popular, quadratic potential describing the canonical scalar field. Such a model will be ruled out, if the tensors remain undetected at a level corresponding to a tensor-to-scalar ratio of, say, r ≅ 0.1. In this work, in addition to the popular quadratic potential, we investigate the effects of the step in a small field model and a tachyon model. With possible applications to future datasets (such as PLANCK) in mind, we evaluate the tensor power spectrum exactly, and include its contribution in our analysis. We compare the models with the WMAP (five as well as seven-year), the QUaD and the ACBAR data. As expected, a step at a particular location and of a suitable magnitude and width is found to improve the fit to the outliers (near l = 22 and 40) in all these cases. We point out that, if the tensors prove to be small (say, r∼<0.01), the quadratic potential and the tachyon model will cease to be viable, and more attention will need to be paid to examples such as the small field models.

79 ASTRONOMY AND ASTROPHYSICS↗

The WorkQueue project: A task queue for the CMS workload management system

We present the development and first experience of a new component (termed WorkQueue) in the CMS workload management system. This component provides a link between a global request system (Request Manager) and agents (WMAgents) which process requests at compute and storage resources (known as sites). These requests typically consist of creation or processing of a data sample (possibly terabytes in size). Unlike the standard concept of a task queue, the WorkQueue does not contain fully resolved work units (known typically as jobs in HEP). This would require the WorkQueue to run computationally heavy algorithms that are better suited to run in the WMAgents. Instead the request specifies an algorithm that the WorkQueue uses to split the request into reasonable size chunks (known as elements). An advantage of performing lazy evaluation of an element is that expanding datasets can be accommodated by having job details resolved as late as possible. The WorkQueue architecture consists of a global WorkQueue which obtains requests from the request system, expands them and forms an element ordering based on the request priority. Each WMAgent contains a local WorkQueue which buffers work close to the agent, this overcomes temporary unavailability of the global WorkQueue and reduces latency for an agent to begin processing. Elements are pulled from the global WorkQueue to the local WorkQueue and into the WMAgent based on the estimate of the amount of work within the element and the resources available to the agent. WorkQueue is based on CouchDB, a document oriented NoSQL database. The WorkQueue uses the features of CouchDB (map/reduce views and bi-directional replication between distributed instances) to provide a scalable distributed system for managing large queues of work. The project described here represents an improvement over the old approach to workload management in CMS which involved individual operators feeding requests into agents. This new approach allows for a system where individual WMAgents are transient and can be added or removed from the system as needed.

Ryu, S.↗

Predicting the von Neumann entanglement entropy using a graph neural network

Calculating the von Neumann entanglement entropy from experimental data is challenging due to its dependence on the complete wavefunction, forcing reliance on approximations such as classical mutual information (MI). We propose a machine learning approach using a graph neural network to predict the von Neumann entropy directly from experimentally accessible bitstrings. We test this approach on a Rydberg ladder system and achieve a mean absolute error of $3.6\,\times 10^{-3}$ when evaluating within the training range on a dataset with entropy values ranging from 0 to 1.9. The model achieves a mean absolute percentage error of 1.44% and outperforms MI-based bounds. When tested beyond the training range, the model maintains reasonable accuracy. Furthermore, we demonstrate that fine-tuning the model with small datasets significantly improves performance on data outside the original training range.

graph neural networks↗

Sigma: Strain-level inference of genomes from metagenomic analysis for biosurveillance

Motivation: Metagenomic sequencing of clinical samples provides a promising technique for direct pathogen detection and characterization in biosurveillance. Taxonomic analysis at the strain level can be used to resolve serotypes of a pathogen in biosurveillance. Sigma was developed for strain-level identification and quantification of pathogens using their reference genomes based on metagenomic analysis. Results: Sigma provides not only accurate strain-level inferences, but also three unique capabilities: (i) Sigma quantifies the statistical uncertainty of its inferences, which includes hypothesis testing of identified genomes and confidence interval estimation of their relative abundances; (ii) Sigma enables strain variant calling by assigning metagenomic reads to their most likely reference genomes; and (iii) Sigma supports parallel computing for fast analysis of large datasets. In conclusion, the algorithm performance was evaluated using simulated mock communities and fecal samples with spike-in pathogen strains. Availability and Implementation: Sigma was implemented in C++ with source codes and binaries freely available at http://sigma.omicsbio.org.

59 BASIC BIOLOGICAL SCIENCES↗

Single-Net Continual Learning with Progressive Segmented Training

There is an increasing need of continual learning in dynamic systems, such as the self-driving vehicle, the surveillance drone, and the robotic system. Such a system requires learning from the data stream, training the model to preserve previous information and adapt to a new task, and generating a single-headed vector for future inference. Different from previous approaches with dynamic structures, this work focuses on a single network and model segmentation to prevent catastrophic forgetting. Leveraging the redundant capacity of a single network, model parameters for each task are separated into two groups: one important group which is frozen to preserve current knowledge, and secondary group to be saved (not pruned) for a future learning. A fixed-size memory containing a small amount of previously seen data is further adopted to assist the training. Without additional regularization, the simple yet effective approach of Progressive Segmented Training (PST) successfully incorporates multiple tasks and achieves the state-of-the-art accuracy in the single-head evaluation on CIFAR-10 and CIFAR-100 datasets. Moreover, the segmented training significantly improves computation efficiency in continual learning at the edge.

Du, Xiaocong↗

Behind-the-Meter Solar Generation Disaggregation using Consumer Mixture Models

To facilitate deep penetration of solar energy in smart grids, we need high observability of solar generation at the edges of the grid. Current advanced metering infrastructures (AMI) only monitor the aggregated measurements from net-metered households, but disaggregated consumption and solar generation components are required for grid optimizations. We propose an unsupervised disaggregation model for disaggregating solar generation from AMI measurements without the need of training data. The model requires only AMI measurements from consumers in a region and the solar irradiance as input, and models the consumption of consumers by neighboring households without rooftop photovoltaics (PV) to perform the disaggregation. We evaluate our results on a real life dataset from Austin, Texas. We show that our model is able to disaggregate consumption and solar generation measurements with 42.24% and 31.67% less mean squared error, respectively, in comparison to a baseline technique that uses supervised learning. This shows that our model is capable of disaggregating historical data even if the dataset has no training data and only contains minimal exogenous data.

14 SOLAR ENERGY↗

A novel partial volume effects correction technique integrating deconvolution associated with denoising within an iterative PET image reconstruction

Purpose: Partial volume effect (PVE) plays an important role in both qualitative and quantitative PET image accuracy, especially for small structures. A previously proposed voxelwise PVE correction method applied on PET reconstructed images involves the use of Lucy–Richardson deconvolution incorporating wavelet-based denoising to limit the associated propagation of noise. The aim of this study is to incorporate the deconvolution, coupled with the denoising step, directly inside the iterative reconstruction process to further improve PVE correction. Methods: The list-mode ordered subset expectation maximization (OSEM) algorithm has been modified accordingly with the application of the Lucy–Richardson deconvolution algorithm to the current estimation of the image, at each reconstruction iteration. Acquisitions of the NEMA NU2-2001 IQ phantom were performed on a GE DRX PET/CT system to study the impact of incorporating the deconvolution inside the reconstruction [with and without the point spread function (PSF) model] in comparison to its application postreconstruction and to standard iterative reconstruction incorporating the PSF model. The impact of the denoising step was also evaluated. Images were semiquantitatively assessed by studying the trade-off between the intensity recovery and the noise level in the background estimated as relative standard deviation. Qualitative assessments of the developed methods were additionally performed on clinical cases. Results: Incorporating the deconvolution without denoising within the reconstruction achieved superior intensity recovery in comparison to both standard OSEM reconstruction integrating a PSF model and application of the deconvolution algorithm in a postreconstruction process. The addition of the denoising step permitted to limit the SNR degradation while preserving the intensity recovery. Conclusions: This study demonstrates the feasibility of incorporating the Lucy–Richardson deconvolution associated with a wavelet-based denoising in the reconstruction process to better correct for PVE. Future work includes further evaluations of the proposed method on clinical datasets and the use of improved PSF models.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Scalable Regression Tree Learning on Hadoop using OpenPlanet

As scientific and engineering domains attempt to effectively analyze the deluge of data arriving from sensors and instruments, machine learning is becoming a key data mining tool to build prediction models. Regression tree is a popular learning model that combines decision trees and linear regression to forecast numerical target variables based on a set of input features. Map Reduce is well suited for addressing such data intensive learning applications, and a proprietary regression tree algorithm, PLANET, using MapReduce has been proposed earlier. In this paper, we describe an open source implement of this algorithm, OpenPlanet, on the Hadoop framework using a hybrid approach. Further, we evaluate the performance of OpenPlanet using realworld datasets from the Smart Power Grid domain to perform energy use forecasting, and propose tuning strategies of Hadoop parameters to improve the performance of the default configuration by 75% for a training dataset of 17 million tuples on a 64-core Hadoop cluster on FutureGrid.

Yin, Wei↗

Mastering HPC Runtime Prediction: From Observing Patterns to a Methodological Approach

The continual expansion of high-performance computing (HPC) brings with it an increasing need for efficiency. Heavy investment in energy, hardware, and software infrastructure to support peta- and exascale computing requires the optimization of existing systems and, wherever possible, the discernment and adoption of best-practices towards these goals. Such is the case for runtime prediction. When a job is submitted to an HPC system, an estimate of its runtime is provided by the user in the form of "requested wallclock". Error in this user-provided estimate can lead to jobs being prematurely killed by the scheduler, increased wait time on the queue, and decreased system utilization. More than fifteen years of research has been directed at mitigating these effects by using data-driven runtime predictions. Codified here is a set of commonalities and insights emerging from this body of work, which we present as recommendations and best practices. These practices are combined into a methodological approach described and evaluated on an 11-million-job dataset from the National Renewable Energy Laboratory's petascale HPC system, Eagle. This dataset and the accompanying codebase have been released to the public domain for the benefit of the wider HPC research community.

high performance computing↗

Learning Universal Authorship Representations

This code contains all the utilities required to reproduce the results of our EMNLP 2021 paper "Learning Universal Authorship Representations". It contains the utilities required for download the datasets, training our model, and performing all evaluations necessary for reproducing the results in the paper. Here's the abstract of our work: Determining whether two documents were composed by the same author, also known as authorship verification, has traditionally been tackled using statistical methods. Recently, authorship representations learned using neural networks have been found to outperform alternatives, particularly in large-scale settings involving hundreds of thousands of authors. But do such representations learned in a particular domain transfer to other domains? Or are these representations inherently entangled with domain-specific features? To study these questions, we conduct the first large-scale study of cross-domain transfer for authorship verification considering zero-shot transfers involving three disparate domains: Amazon reviews, fanfiction short stories, and Reddit comments. We find that although a surprising degree of transfer is possible between certain do- mains, it is not so successful between others. We examine properties of these domains that influence generalization and propose simple but effective methods to improve transfer.

Rivera Soto, RafaelA↗

Appending High-Resolution Elevation Data to GPS Speed Traces for Vehicle Energy Modeling and Simulation

Accurate and reliable global positioning system (GPS)-based vehicle use data are highly valuable for many transportation, analysis, and automotive considerations. Model-based design, real-world fuel economy analysis, and the growing field of autonomous and connected technologies (including predictive powertrain control and self-driving cars) all have a vested interest in high-fidelity estimation of powertrain loads and vehicle usage profiles. Unfortunately, road grade can be a difficult property to extract from GPS data with consistency. In this report, we present a methodology for appending high-resolution elevation data to GPS speed traces via a static digital elevation model. Anomalous data points in the digital elevation model are addressed during a filtration/smoothing routine, resulting in an elevation profile that can be used to calculate road grade. This process is evaluated against a large, commercially available height/slope dataset from the Navteq/Nokia/HERE Advanced Driver Assistance Systems product. Results will show good agreement with the Advanced Driver Assistance Systems data in the ability to estimate road grade between any two consecutive points in the contiguous United States.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Information Theoretic Measures for Visual Analytics: The Silver Ticket?

In the context of text-based analysis workflows, we propose that an effective analytic tool facilitates triage by a) enabling users to identify and set aside irrelevant content (i.e., reduce the complexity of information in a dataset) and b) develop a working mental model of which items are most relevant to the question at hand. This LDRD funded research developed a dataset that is enabling this team to evaluate propose normalized compression distance (NCD) as a task, user, and context-insensitive measure of categorization outcomes (Shannon entropy is reduced as order is imposed). Effective analytics tools help people impose order, reducing complexity in measurable ways. Our concept and research was documented in a paper accepted to the ACM conference "Beyond Time and Error: Novel Methods in Information Visualization Evaluation", part of the IEEE VisWeek Conference, Baltimore, MD, October 16-21, 2016. The paper is included as an appendix to this report.

97 MATHEMATICS AND COMPUTING↗

A Retrospective Analysis of Distributed Solar Interconnection Timelines and Related State Mandates

Most distributed residential photovoltaic (PV) systems must secure an interconnection agreement prior to operation that ensures the local electrical system will operate safely within the broader electrical grid. PV installers generally submit an interconnection application to the local utility that provides all the necessary details the utility needs to evaluate the proposed system. To encourage timely execution of this process, many states’ utility commissions have established mandates that limit the maximum number of days allowed for utility review and approval of interconnection for certain small, often residential applications. In this paper, we derive the median and range of cycle times for the pre-installation approval phase of the interconnection process (i.e., from application submission to approval by the utility) across 24 U.S. states, using a data set of approximately 170,000 projects. We evaluate the percentage of projects that are approved within respective state-mandated timelines from 2017– 2019. We further evaluate how timelines have evolved since 2012 for a subset of five states: Arizona, California, Colorado, New Jersey, and New York. The analyses are divided into two size ranges: systems that are ≤10 kilowatts (kW), and systems that are 11–50 kW.

14 SOLAR ENERGY↗

Cross-Scale Land-Atmosphere Experiment (CSLAEX) (Final Report)

It is now well recognized that land-atmosphere interactions play an important role in both weather and climate. These interactions take place over a wide range of spatial and temporal scales. Most of the current formulations of the momentum, heat and moisture transport laws are informed by observations with a limited temporal or spatial scale, and hence may have limited applicability. The objective of this work is to improve the multi-scale representation of the near-surface heat transfer in the ground and in the atmosphere using a combination of modeling and field observations. Our knowledge and predictive capacity in land-atmosphere coupling is limited by our capacity to observe the full-scale spectrum at play. In order to bridge this fundamental gap in our observation capacity I will work to: create the first long-term Integrated Cross-Scale land-Atmosphere Experiment (CSLAEX) since I will work to: measure the components of the surface energy balance at a spatial scale relevant to remote sensing and meteorological applications (5km), along with observations of the nested subscale processes with Distributed Temperature Sensors (DTS). This proposal aims to fill the gaps in modeling and observing land-atmosphere interactions across scales via a novel Cross-Scale Land-Atmosphere Experiment (CSLAEX) to be implemented as part of the Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) site near Lamont, Oklahoma. The proposed research effort aims at exploring land-atmosphere processes across multiple spatial scales and to investigate the effect of landscape heterogeneity using Distributed Temperature Sensor measurements. The novel instrumentation of the Surface Energy Balance and simultaneous observations of the atmospheric state (boundary layer, cloud, convection) available at the DOE ARM SGP site will be used to create CSLAEX. This dataset will be used to frame and evaluate new transport laws and parameterizations for the boundary layer and the land surface, which can be implemented into the Weather and Research Forecasting (WRF) model and Community Atmospheric Model (CAM). The result of this study should impact multiple fields: meteorology, climate, and hydrology.

54 ENVIRONMENTAL SCIENCES↗

DetSuM: Detector Surrogate Model for LArTPC

In large neutrino experiments such as the Deep Underground Neutrino Experiment (DUNE), estimating detector response uncertainties typically requires simulation samples that consume substantial computing resources and time. To mitigate this challenge, we present DetSuM, an uncertainty-aware surrogate model designed to capture the detector response variations with reduced computing load compared to full simulations. This poster describes the construction and evaluation of DetSuM using simulation and reconstruction datasets in a rare-event search at DUNE. We assess DetSuM's ability to predict key detector-response variations and their associated uncertainties, discuss current limitations, and outline improvements to extend its validity in systematics studies of DUNE physics.

Li, Aobo [UC, San Diego]↗