Search NASA⌕ Search

SEARCH · Search NASA

Results for “EM Algorithms”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2014

A study on fast digital discrimination of neutron and gamma-ray for improvement neutron emission profile measurement

Neutron and γ-ray (n-γ) discrimination with a digital signal processing system has been used to measure the neutron emission profile in magnetic confinement fusion devices. However, a sampling rate must be set low to extend the measurement time because the memory storage is limited. Time jitter decreases a discrimination quality due to a low sampling rate. As described in this paper, a new charge comparison method was developed. Furthermore, automatic n-γ discrimination method was examined using a probabilistic approach. Analysis results were investigated using the figure of merit. Results show that the discrimination quality was improved. Automatic discrimination was applied using the EM algorithm and k-means algorithm.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2018

Latent Laplacian Maximum Entropy Discrimination for Detection of High-Utility Anomalies

Data-driven anomaly detection methods suffer from the drawback of detecting all instances that are statistically rare, irrespective of whether the detected instances have realworld significance or not. In this paper, we are interested in the problem of specifically detecting anomalous instances that are known to have high real-world utility, while ignoring the low-utility statistically anomalous instances. To this end, we propose a novel method called Latent Laplacian Maximum Entropy Discrimination (LatLapMED) as a potential solution. This method uses the EM algorithm to simultaneously incorporate the Geometric Entropy Minimization principle for identifying statistical anomalies, and the Maximum Entropy Discrimination principle to incorporate utility labels, in order to detect highutility anomalies. Here, we apply our method in both simulated and real datasets to demonstrate that it has superior performance over existing alternatives that independently pre-process with unsupervised anomaly detection algorithms before classifying.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Gamma-Ray Point-Source Localization and Sparse Image Reconstruction using Poisson Likelihood

Gamma-ray imaging trys to reconstruct the spatial and intensity distribution of gamma-emitting radionuclides from a set of measurements. Wholly, this problem is solved by discretizing the spatial dimensions and employing the maximum likelihood expectation maximization (ML-EM) algorithm, with or without some form of regularization. While the generality of this formulation enables use in a wide variety of scenarios, it is susceptible to overfitting, limited by the discretization of spatial coordinates, and can be computationally expensive. We present a novel approach to 3D gamma-ray image reconstruction for scenarios where sparsity may be assumed, for example radiological source search. In this work we first formulate a point-source localization (PSL) approach as an optimization problem where both position and source intensity are continuous variables. We then extend and generalize this formulation to an iterative algorithm called additive point-source localization (APSL) for sparse parametric image reconstruction. A set of simulated source search scenarios using a single non-directional detector are considered, finding improved image accuracy and computational efficiency with APSL over traditional grid-based methods.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2019

Anomaly Detection in Partially Observed Traffic Networks

This paper addresses the problem of detecting anomalous activity in traffic networks where the network is not directly observed. Given knowledge of what the node-tonode traffic in a network should be, any activity that differs significantly from this baseline would be considered anomalous. We propose a Bayesian hierarchical model for estimating the traffic rates and detecting anomalous changes in the network. The probabilistic nature of the model allows us to perform statistical goodness-of-fit tests to detect significant deviations from a baseline network. We show that due to the more defined structure of the hierarchical Bayesian model, such tests perform well even when the empirical models estimated by the EM algorithm are misspecified. Finally, we apply our model to both simulated and real datasets to demonstrate its superior performance over existing alternatives.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

SU-E-T-51: Bayesian Network Models for Radiotherapy Error Detection

Purpose: To develop a probabilistic model of radiotherapy plans using Bayesian networks that will detect potential errors in radiation delivery. Methods: Semi-structured interviews with medical physicists and other domain experts were employed to generate a set of layered nodes and arcs forming a Bayesian Network (BN) which encapsulates relevant radiotherapy concepts and their associated interdependencies. Concepts in the final network were limited to those whose parameters are represented in the institutional database at a level significant enough to develop mathematical distributions. The concept-relation knowledge base was constructed using the Web Ontology Language (OWL) and translated into Hugin Expert Bayes Network files via the the RHugin package in the R statistical programming language. A subset of de-identified data derived from a Mosaiq relational database representing 1937 unique prescription cases was processed and pre-screened for errors and then used by the Hugin implementation of the Estimation-Maximization (EM) algorithm for machine learning all parameter distributions. Individual networks were generated for each of several commonly treated anatomic regions identified by ICD-9 neoplasm categories including lung, brain, lymphoma, and female breast. Results: The resulting Bayesian networks represent a large part of the probabilistic knowledge inherent in treatment planning. By populating the networks entirely with data captured from a clinical oncology information management system over the course of several years of normal practice, we were able to create accurate probability tables with no additional time spent by experts or clinicians. These probabilistic descriptions of the treatment planning allow one to check if a treatment plan is within the normal scope of practice, given some initial set of clinical evidence and thereby detect for potential outliers to be flagged for further investigation. Conclusion: The networks developed here support the use of probabilistic models into clinical chart checking for improved detection of potential errors in RT plans.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2014

TH-E-BRF-09: Gaussian Mixture Model Analysis of Radiation-Induced Parotid-Gland Injury: An Ultrasound Study of Acute and Late Xerostomia in Head-And-Neck Radiotherapy

Purpose: Xerostomia (dry mouth), secondary to parotid-gland injury, is a distressing side-effect in head-and-neck radiotherapy (RT). This study's purpose is to develop a novel ultrasound technique to quantitatively evaluate post-RT parotid-gland injury. Methods: Recent ultrasound studies have shown that healthy parotid glands exhibit homogeneous echotexture, whereas post-RT parotid glands are often heterogeneous, with multiple hypoechoic (inflammation) or hyperechoic (fibrosis) regions. We propose to use a Gaussian mixture model to analyze the ultrasonic echo-histogram of the parotid glands. An IRB-approved clinical study was conducted: (1) control-group: 13 healthy-volunteers, served as the control; (2) acutetoxicity group − 20 patients (mean age: 62.5 ± 8.9 years, follow-up: 2.0±0.8 months); and (3) late-toxicity group − 18 patients (mean age: 60.7 ± 7.3 years, follow-up: 20.1±10.4 months). All patients experienced RTOG grade 1 or 2 salivary-gland toxicity. Each participant underwent an ultrasound scan (10 MHz) of the bilateral parotid glands. An echo-intensity histogram was derived for each parotid and a Gaussian mixture model was used to fit the histogram using expectation maximization (EM) algorithm. The quality of the fitting was evaluated with the R-squared value. Results: (1) Controlgroup: all parotid glands fitted well with one Gaussian component, with a mean intensity of 79.8±4.9 (R-squared>0.96). (2) Acute-toxicity group: 37 of the 40 post-RT parotid glands fitted well with two Gaussian components, with a mean intensity of 42.9±7.4, 73.3±12.2 (R-squared>0.95). (3) Latetoxicity group: 32 of the 36 post-RT parotid fitted well with 3 Gaussian components, with mean intensities of 49.7±7.6, 77.2±8.7, and 118.6±11.8 (R-squared>0.98). Conclusion: RT-associated parotid-gland injury is common in head-and-neck RT, but challenging to assess. This work has demonstrated that the Gaussian mixture model of the echo-histogram could quantify acute and late toxicity of the parotid glands. This study provides meaningful preliminary data from future observational and interventional clinical research.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

Improved dosimetry for targeted radionuclide therapy using nonrigid registration on sequential SPECT images

Purpose: Voxel-level and patient-specific 3D dosimetry for targeted radionuclide therapy (TRT) typically involves serial nuclear medicine scans. Misalignment of the images can result in reduced dosimetric accuracy. Since the scans are typically performed over a period of several days, there will be patient movement between scans and possible nonrigid organ deformation. This work aims to implement and evaluate the use of nonrigid image registration on a series of quantitative SPECT (QSPECT) images for TRT dosimetry. Methods: A population of 4D extended cardiac torso phantoms, comprised of three In-111 Zevalin biokinetics models and three anatomical variations, was generated based on the patient data. The authors simulated QSPECT acquisitions at five time points. At each time point, individual organ and whole-body deformation between scans were modeled by translating/rotating organs and the body up to 5°/voxels, keeping ≤5% difference in organ volume. An analytical projector was used to generate realistic noisy projections for a medium energy general purpose collimator. Projections were reconstructed using OS-EM algorithm with geometric collimator detector response, attenuation, and scatter corrections. The QSPECT images were registered using organ-based nonrigid image registration method. The cumulative activity in each voxel was obtained by integrating the activity over time. Dose distribution images were obtained by convolving the cumulative activity images with a Y-90 dose kernel. Dose volume histograms (DVHs) for organs-of-interest were analyzed. Results: After nonrigid registration, the mean differences in organ doses compared to the case without misalignment were improved from (−15.50 ± 5.59)% to (−2.12 ± 1.05)% and (−7.28 ± 2.30)% to (−0.23 ± 0.71)% for the spleen and liver, respectively. For all organs, the cumulative DVHs showed improvement after nonrigid registration and the normalized absolute error of differential DVHs ranged from 6.79% to 22.70% for liver and 26.00% to 39.70% for spleen with different segmentation methods. Conclusions: These results demonstrated that nonrigid registration of sequential QSPECT images is feasible for TRT and improves the accuracy of 3D dosimetry.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2016

A Scintillation-Fiber Based Cosmic-ray Muon Tomography System for Imaging Dry Storage Cask

Because of the growth of the nuclear power industry in the United States and the policy decision to ban reprocessing of commercial spent nuclear fuel, the spent fuel inventory at commercial reactor sites has been increasing. Spent nuclear fuel presents extra complexity of security and proliferation issues. These fuel assemblies are practically inaccessible for inspection purposes, as reopening a DSC would require special facilities and would be tremendously expensive. Cosmic-ray muon Tomography has been studied in the past for various applications and was demonstrated to be a good non-destructive assay technique for high-Z materials, such as spent nuclear fuel assemblies in dry storage casks (DSC). Compared to conventional imaging techniques, muon imaging has some advantages due to its negligible health risk and high penetrating power. In this paper, we report our simulation study on imaging a DSC using muon Tomography technique. The Point of Closest Approach (POCA) and the Maximum-Likelihood Expectation- Maximization (ML-EM) algorithms are implemented and evaluated. A scaled-down prototype muon tomography system is proposed and studied. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2022

Structures via Reasoning - Applying AI to Cryo Electron Microscopy to Reveal Structural Variability

There have been breakthroughs of latest cryo electron microscopy (cryo-EM) data analysis algorithms to classify cryo-EM image data. However, most of these cryo-EM reconstruction methods have focused on classifying distinctly different biomolecule structures. Here, we present our approaches of deep learning to differentiate homologous structures that are distinguishable only with inner morphological differences. We succeeded supervised classification of these subtly different homologues. However, we could not differentiate them with unsupervised methods. Here we discuss what further approaches are likely needed for successful unsupervised classification.

47 OTHER INSTRUMENTATION↗
DOE OSTI2026

Computationally efficient Bayesian estimation of graphical networks for omics data

Graphical networks are useful, widely-used modeling approaches to represent complex biological processes with biological measurements generated by platforms such as mass spectrometry. Bayesian analyses of graphical networks for omics data have several advantages over their frequentist counterparts, such as the inclusion of prior knowledge in the estimation of models. However, Bayesian approaches to date have only been feasible for data with a couple hundred biomolecules due to prohibitive computational time, but omics data often contains tens of thousands of biomolecules. Here, we present and illustrate a more computationally efficient approach named BPlane (Bayesian PseudoLikelihood-based Algorithm for Network Estimation) to extend Bayesian modeling capabilities for larger-sized datasets, such as most untargeted proteomics data. Via simulation, we demonstrate that BPlane produces substantial computational savings over a current state-of-the-art Bayesian algorithm while maintaining competitive edge detection accuracy. On a SARS-CoV2 proteomics data with 7000 proteins, the competing algorithm takes three times as long to complete the first iteration as BPlane takes to converge after over 100 iterations.

EM algorithm↗
DOE OSTI2024

Neutron source reconstruction using a generalized expectation–maximization algorithm on one-dimensional neutron images from the Z facility

Magnetized Liner Inertial Fusion experiments have been performed at the Z facility at Sandia National Laboratories. These experiments use deuterium fuel, which produces 2.45 MeV neutrons on reaching thermonuclear conditions. To study the spatial structure of neutron production, the one-dimensional imager of neutrons diagnostic was fielded to record axial resolved neutron images. In this diagnostic, neutrons passing through a rolled edge aperture form an image on a CR-39-based solid state nuclear track detector. In this report we present a modified generalized expectation–maximization algorithm to reconstruct an axial neutron emission profile of the stagnated fusion plasma. We validate the approach by comparing the reconstructed neutron emission profile to an x-ray emission profile provided by a time-integrated pinhole camera.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

Estimating Sparse Direct Effects in Multivariate Regression With the Spike-and-Slab LASSO

The multivariate regression interpretation of the Gaussian chain graph model simultaneously parametrizes (i) the direct effects of p predictors on q outcomes and (ii) the residual partial covariances between pairs of outcomes. We introduce a new method for fitting sparse versions of these models with spike-and-slab LASSO (SSL) priors. We develop an Expectation Conditional Maximization algorithm to obtain sparse estimates of the p × q matrix of direct effects and the q × q residual precision matrix. Our algorithm iteratively solves a sequence of penalized maximum likelihood problems with self-adaptive penalties that gradually filter out negligible regression coefficients and partial covariances. Because it adaptively penalizes individual model parameters, our method is seen to outperform fixed-penalty competitors on simulated data. We establish the posterior contraction rate for our model, buttressing our method’s excellent empirical performance with strong theoretical guarantees. Using our method, we estimated the direct effects of diet and residence type on the composition of the gut microbiome of elderly adults.

EM algorithm↗
DOE OSTI2013

An Expectation-Maximization Method for Calibrating Synchronous Machine Models

The accuracy of a power system dynamic model is essential to its secure and efficient operation. Lower confidence in model accuracy usually leads to conservative operation and lowers asset usage. To improve model accuracy, this paper proposes an expectation-maximization (EM) method to calibrate the synchronous machine model using phasor measurement unit (PMU) data. First, an extended Kalman filter (EKF) is applied to estimate the dynamic states using measurement data. Then, the parameters are calculated based on the estimated states using maximum likelihood estimation (MLE) method. The EM method iterates over the preceding two steps to improve estimation accuracy. The proposed EM method’s performance is evaluated using a single-machine infinite bus system and compared with a method where both state and parameters are estimated using an EKF method. Sensitivity studies of the parameter calibration using EM method are also presented to show the robustness of the proposed method for different levels of measurement noise and initial parameter uncertainty.

EM Algorithms↗
DOE OSTI2022

Gaussian Mixture Model Solvers for the Boltzmann Equation

This report documents our experience constructing a numerical method for the collisional Boltzmann equation that is capable of accurately capturing the collisionless through strongly collisional limits. We explore three different functional representations and present a detailed account of a numerical method based on a spatially dependent Gaussian mixture model (GMM). The Kullback-Leibler divergence is used as a closeness measure and various expectation maximization (EM) solution algorithms are implemented to find a compact representation in velocity space for distribution functions that exhibit significant non-Maxwellian character. We discuss issues that appear with this representation over a range of Knudsen numbers for a prototypical test problem and demonstrate that the strongly collisional limit recovers a solution to Euler's equations. Looking forward, this approach is broadly applicable to the non-relativistic and relativistic collisional Vlasov equations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Comparison of Optimal Energy Management Strategies Using Dynamic Programming, Model Predictive Control, and Constant Velocity Prediction

Due to the recent advancements in autonomous vehicle technology, future vehicle velocity predictions are becoming more robust which allows fuel economy (FE) improvements in hybrid electric vehicles through optimal energy management strategies (EMS). A real- world highway drive cycle (DC) and a controls-oriented 2017 Toyota Prius Prime model are used to study potential FE improvements. We proposed three important metrics for comparison: (1) perfect full drive cycle prediction using dynamic programming, (2) 10-second prediction horizon model predictive control (MPC), and (3) 10-second constant velocity prediction. These different velocity predictions are put into an optimal EMS derivation algorithm to derive optimal engine torque and engine speed. The results show that the constant velocity prediction algorithm outperformed the baseline control strategy but underperformed the MPC strategy with an average 1.58% and 2.45% of FE improvement with highway and city-highway DC. Also, using a 10-second prediction window MPC strategy provided FE improvement results close to the full drive cycle prediction case. MPC has the potential to achieve 60%-65% and 70% - 80% of global FE improvement over highway and city-highway DC respectively.

Patel, Amol Arvind↗
DOE OSTI2020

Comparison of Optimal Energy Management Strategies Using Dynamic Programming, Model Predictive Control, and Constant Velocity Prediction

Due to the recent advancements in autonomous vehicle technology, future vehicle velocity predictions are becoming more robust which allows fuel economy (FE) improvements in hybrid electric vehicles through optimal energy management strategies (EMS). A realworld highway drive cycle (DC) and a controls-oriented 2017 Toyota Prius Prime model are used to study potential FE improvements. We proposed three important metrics for comparison: (1) perfect full drive cycle prediction using dynamic programming, (2) 10-second prediction horizon model predictive control (MPC), and (3) 10-second constant velocity prediction. These different velocity predictions are put into an optimal EMS derivation algorithm to derive optimal engine torque and engine speed. The results show that the constant velocity prediction algorithm outperformed the baseline control strategy but underperformed the MPC strategy with an average 1.58% and 2.45% of FE improvement with highway and city-highway DC. Also, using a 10-second prediction window MPC strategy provided FE improvement results close to the full drive cycle prediction case. MPC has the potential to achieve 60%-65% and 70% - 80% of global FE improvement over highway and city-highway DC respectively.

Patel, Amol Arvind↗
DOE OSTI2017

What does fault tolerant Deep Learning need from MPI?

Deep Learning (DL) algorithms have become the {\em de facto} Machine Learning (ML) algorithm for large scale data analysis. DL algorithms are computationally expensive -- even distributed DL implementations which use MPI require days of training (model learning) time on commonly studied datasets. Long running DL applications become susceptible to faults -- requiring development of a fault tolerant system infrastructure, in addition to fault tolerant DL algorithms. This raises an important question: {\em What is needed from MPI for designing fault tolerant DL implementations?} In this paper, we address this problem for permanent faults. We motivate the need for a fault tolerant MPI specification by an in-depth consideration of recent innovations in DL algorithms and their properties, which drive the need for specific fault tolerance features. We present an in-depth discussion on the suitability of different parallelism types (model, data and hybrid); a need (or lack thereof) for check-pointing of any critical data structures; and most importantly, consideration for several fault tolerance proposals (user-level fault mitigation (ULFM), Reinit) in MPI and their applicability to fault tolerant DL implementations. We leverage a distributed memory implementation of Caffe, currently available under the Machine Learning Toolkit for Extreme Scale (MaTEx). We implement our approaches by extending MaTEx-Caffe for using ULFM-based implementation. Our evaluation using the ImageNet dataset and AlexNet neural network topology demonstrates the effectiveness of the proposed fault tolerant DL implementation using OpenMPI based ULFM.

Amatya, Vinay C.↗
DOE OSTI2016

Parallel FETI‐DP algorithm for efficient simulation of large‐scale EM problems

Summary An efficient parallelization of the dual‐primal finite‐element tearing and interconnecting (FETI‐DP) algorithm is presented for large‐scale electromagnetic simulations. As a nonoverlapping domain decomposition method, the FETI‐DP algorithm formulates a global interface problem, whose iterative solution is accelerated with a solution of a global corner problem. To achieve a good load balance for parallel computation, the original computational domain is decomposed into subdomains with similar sizes and shapes. The subdomains are then distributed to processors based on their close proximity to minimize inter‐processor communication. The parallel generalized minimal residual method, enhanced with the iterative classical Gram‐Schmidt orthogonalization scheme to reduce global communication, is adopted to solve the global interface problem with a fast convergence rate. The global corner‐related coarse problem is solved iteratively with a parallel communication‐avoiding biconjugate gradient stabilized method to minimize global communication, and its convergence is accelerated by a diagonal preconditioner constructed from the coarse system matrix. To alleviate neighboring communication overhead, the non‐blocking communication approach is employed in both generalized minimal residual and communication‐avoiding biconjugate gradient stabilized iterative solutions. Three numerical examples are presented to demonstrate the accuracy, scalability, and capability of the proposed parallel FETI‐DP algorithm for electromagnetic modeling of general objects and antenna arrays. Copyright © 2016 John Wiley & Sons, Ltd.

Zhang, Kedi↗