Search NASA⌕ Search

SEARCH · Search NASA

Results for “maximum likelihood estimation”

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 73 records · Page 4

A Locally Adapting Technique for Edge Detection using Image Segmentation

Rapid growth in the field of quantitative digital image analysis is paving the way for scientific researchers to make precise measurements about objects in an image. To compute quantities from an image such as the density of compressed materials or the velocity of a shockwave, object boundaries must first be determined. Images containing regions that each have a spatial trend in intensity are of particular interest here. For edge detection, we present a supervised, statistical image segmentation method that incorporates spatial information to locate boundaries between regions with overlapping intensity histograms, specifically for images where the regions are known but precise boundary locations are unknown. The segmentation of a pixel is determined by comparing its intensity to distributions from nearby pixel intensities, and a gradient of the segmented image indicates edge locations. Because of the statistical nature of the algorithm, we use maximum likelihood estimation to quantify uncertainty about each boundary. We demonstrate the success of this algorithm at locating boundaries and providing uncertainty bands on a radiograph of a multicomponent cylinder and on an optical image of a laser-induced shockwave.

97 MATHEMATICS AND COMPUTING↗

Bayesian forecasting and uncertainty quantifying of stream flows using Metropolis–Hastings Markov Chain Monte Carlo algorithm

This paper presents a Bayesian approach using Metropolis-Hastings Markov Chain Monte Carlo algorithm and applies this method for daily river flow rate forecast and uncertainty quantification for Zhujiachuan River using data collected from Qiaotoubao Gage Station and other 13 gage stations in Zhujiachuan watershed in China. The proposed method is also compared with the conventional maximum likelihood estimation (MLE) for parameter estimation and quantification of associated uncertainties. While the Bayesian method performs similarly in estimating the mean value of daily flow rate, it performs over the conventional MLE method on uncertainty quantification, providing relatively narrower reliable interval than the MLE confidence interval and thus more precise estimation by using the related information from regional gage stations. As a result, the Bayesian MCMC method might be more favorable in the uncertainty analysis and risk management.

58 GEOSCIENCES↗

Position and Velocity Tracking in Cellular Networks Using the Kalman Filter

Access to the right information anytime, anywhere is becoming the new driving force for the information technology revolution. The 'right' information's relevance is based on the user's profile and his/her current geographical position and/or time. Location Based Service (LBS) is an innovative technology that provides information or makes information available based on the geographical location of the mobile user. Analysts predict that LBSs will lead to new applications, generating billions of US dollars worldwide (Leite, 2001; Searle, 2001). The need for an efficient and accurate mobile station (MS) positioning system is growing day by day. The ability to pinpoint the location of an individual has an obvious and vital value in the context of emergency services (Chan, 2003; Olama et al., 2008). Pinpointing the location of people and other valuable assets also opens the door to a new world of previously unimagined information services and m-commerce probabilities. For example, availability of services like 'Where is the nearest ATM?', 'Check traffic conditions on the highway on my route', 'Find a parking lot nearby', as well as answers to 'Where is my advisor?', and 'Where is my car?' will be an everyday rule in our lives (Charalambous & Panayiotou, 2004). A technology independent LBS architecture can be considered as comprised by three main parts (Girodon, 2002): A user requesting information, a mobile network operator and its partners, and several content providers (e.g. data, maps). The subscriber requests a personalized service dependant on his geographic location. The system will ask the Location Services Manager (which is in charge of handling requests, i.e., send/receive to the Location Calculator and the Content Providers) to pinpoint the location of the mobile. The Location Services Manager (LSM), using the Location Calculator, will ask the Content Provider (CP) to supply qualified information according to the mobile's geographical position. The LSM will eventually receive the answer from the CP and send it to the mobile, performing the essential data translations. Fig. 1 outlines the precedent concept. For effective provision of LBS, one has to provide an accurate location, as well as suitable information for users required by the corresponding service, with minimal expenditure. Thus, there are three main technology issues that have to be resolved for LBS: positioning technology, application technology, and location services (Dru & Saada, 2001). A very important technology is of course the positioning technology, the way to find out the location of a mobile device accurately. Due to the unique characteristics of the cellular environment, it is a great challenge to locate the user precisely. However, in many cases, application technology and location services are important consideration of LBS. Application technology manages the geographic information and delivers the customer requests to the appropriate service provider, thus it constitutes the communication system involved. LBS uses the geographic information to provide geographically sensitive information and services. Location-based applications and services are not sensitive to the type of location technology that is used - they merely rely on reasonably accurate geographic coordinates (Chan, 2003). This chapter is structured as follows: In Section 2, we describe the use and applications of LBSs. The current location determination technologies and standards are presented in Section 3. In Section 4 we describe the mathematical models used for the location and velocity estimation algorithms. An initial attempt for MS location estimation via received signal level using the maximum likelihood estimation (MLE) approach and triangulation is presented in Sections 5. Since the former approach lacks acceptable accuracy for demanding services as numerical results reveal, the extended Kalman filter (EKF) approach, which is the main topic in this chapter, is introduced in Sections 6. In Section 7 we present numerical results. Section 8 provides concluding remarks.

99 GENERAL AND MISCELLANEOUS↗

Autonomous Correction of Sensor Data Applied to Building Technologies Utilizing Statistical Processing Methods

Autonomous detection and correction of potentially missing or corrupt sensor data is a essential concern in building technologies since data availability and correctness is necessary to develop accurate software models for instrumented experiments. Therefore, this paper aims to address this problem by using statistical processing methods including: (1) least squares; (2) maximum likelihood estimation; (3) segmentation averaging; and (4) threshold based techniques. Application of these validation schemes are applied to a subset of data collected from Oak Ridge National Laboratory s (ORNL) ZEBRAlliance research project, which is comprised of four single-family homes in Oak Ridge, TN outfitted with a total of 1,218 sensors. The focus of this paper is on three different types of sensor data: (1) temperature; (2) humidity; and (3) energy consumption. Simulations illustrate the threshold based statistical processing method performed best in predicting temperature, humidity, and energy data.

building technologies.↗

Investigation of methodology for uncertainty quantification of model parameters

Quantifying the uncertainty contributors of best-estimate thermal hydraulic code is getting more and more attention in safety analysis of nuclear industry during recent decades. Yet for uncertainties of certain model parameters which may not be directly measured, the quantification process is usually subjective and inaccurate. This paper investigates the statistical methodologies in order to get the probability density function (pdf) of model parameters more objectively based on observed experimental responses. The simplification of mathematical model is described for the parameter estimation, and solutions based on Bayesian theorem and Maximum Likelihood Estimation (MLE) are classified and demonstrated for both deterministic approach and stochastic approach. As an application, the uncertainties of model parameters related to reflood phenomena implemented in RELAP5 code are quantified. It is indicated that the statistical method which is independent of best-estimate code is feasible and easy to apply. And through the check of uncertainty propagation, it demonstrates that the uncertainty bands can envelope most of the experiment measurement with advantage of accuracy. (authors)

42 ENGINEERING↗

Asteroid models from photometry and complementary data sources

I discuss inversion methods for asteroid shape and spin reconstruction with photometry (lightcurves) and complementary data sources such as adaptive optics or other images, occultation timings, interferometry, and range-Doppler radar data. These are essentially different sampling modes (generalized projections) of plane-of-sky images. An important concept in this approach is the optimal weighting of the various data modes. The maximum compatibility estimate, a multi-modal generalization of the maximum likelihood estimate, can be used for this purpose. I discuss the fundamental properties of lightcurve inversion by examining the two-dimensional case that, though not usable in our three-dimensional world, is simple to analyze, and it shares essentially the same uniqueness and stability properties as the 3-D case. After this, I review the main aspects of 3-D shape representations, lightcurve inversion, and the inclusion of complementary data.

79 ASTRONOMY AND ASTROPHYSICS↗

Development of control quality factor for HVAC control loop performance assessment I—Methodology (ASHRAE RP-1587)

This article is the second paper from the ASHRAE research project RP-1587, focusing on the methodology of obtaining a control quality factor (CQF). This article presents a development of two CQFs for assessing the heating, ventilation, and air-conditioning (HVAC) control loop performance. Both CQFs are able to detect whether the control loop is able to maintain the setpoint and identify the loop's ability to handle disturbances. In addition, the reversal behaviors are assessed as well. The first CQF (CQF-Harris) is proposed based on the normalized Harris index using the recursive least squares method. This recursive least squares method is selected because of its computational efficiency compared with the maximum likelihood estimation method. The second CQF (CQF-EWMA) is based on the exponentially weighted moving average of the error ratio. The assessment scale of excellent, good, fair, bad, and failed, which indicates the quality of the HVAC control loops, is established as well. The sensitivity analysis for both CQFs is also conducted, and it provides insights on choosing the appropriate parameters to compute such CQFs. Such parameters include the sampling frequency, the length of the moving window, and the variance of the unmeasured disturbance. The field evaluations and tests of the proposed CQFs for simulated control loops and real control loops can be found in the companion paper with the title 'Development of Control Quality Factor for HVAC Control Loop Performance Assessment - III: Field Testing and Results (ASHRAE RP-1587).'

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Joint constraints on galaxy bias and σ{sub 8} through the N-pdf of the galaxy number density

We present a full description of the N-probability density function of the galaxy number density fluctuations. This N-pdf is given in terms, on the one hand, of the cold dark matter correlations and, on the other hand, of the galaxy bias parameter. The method relies on the assumption commonly adopted that the dark matter density fluctuations follow a local non-linear transformation of the initial energy density perturbations. The N-pdf of the galaxy number density fluctuations allows for an optimal estimation of the bias parameter (e.g., via maximum-likelihood estimation, or Bayesian inference if there exists any a priori information on the bias parameter), and of those parameters defining the dark matter correlations, in particular its amplitude (σ{sub 8}). It also provides the proper framework to perform model selection between two competitive hypotheses. The parameters estimation capabilities of the N-pdf are proved by SDSS-like simulations (both, ideal log-normal simulations and mocks obtained from Las Damas simulations), showing that our estimator is unbiased. We apply our formalism to the 7th release of the SDSS main sample (for a volume-limited subset with absolute magnitudes M{sub r} ≤ −20). We obtain b-circumflex = 1.193 ± 0.074 and σ-bar{sub 8} = 0.862 ± 0.080, for galaxy number density fluctuations in cells of the size of 30h{sup −1}Mpc. Different model selection criteria show that galaxy biasing is clearly favoured.

79 ASTRONOMY AND ASTROPHYSICS↗

Stopping criteria for ending autonomous, single detector radiological source searches

While the localization of radiological sources has traditionally been handled with statistical algorithms, such a task can be augmented with advanced machine learning methodologies. The combination of deep and reinforcement learning has provided learning-based navigation to autonomous, single-detector, mobile systems. However, these approaches lacked the capacity to terminate a surveying/search task without outside influence of an operator or perfect knowledge of source location (defeating the purpose of such a system). Two stopping criteria are investigated in this work for a machine learning navigated system: one based upon Bayesian and maximum likelihood estimation (MLE) strategies commonly used in source localization, and a second providing the navigational machine learning network with a “stop search” action. A convolutional neural network was trained via reinforcement learning in a 10 m × 10 m simulated environment to navigate a randomly placed detector-agent to a randomly placed source of varied strength (stopping with perfect knowledge during training). The network agent could move in one of four directions (up, down, left, right) after taking a 1 s count measurement at the current location. During testing, the stopping criteria for this navigational algorithm was based upon a Bayesian likelihood estimation technique of source presence, updating this likelihood after each step, and terminating once the confidence of the source being in a single location exceeded 0.9. A second network was trained and tested with similar architecture as the previous but which contained a fifth action: for self-stopping. The accuracy and speed of localization with set detector and source initializations were compared over 50 trials of MLE-Bayesian approach and 1000 trials of the CNN with self-stopping. The statistical stopping condition yielded a median localization error of ~1.41 m and median localization speed of 12 steps. The machine learning stopping condition yielded a median localization error of 0 m and median localization speed of 17 steps. This work demonstrated two stopping criteria available to a machine learning guided, source localization system.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗

Kumaraswamy distribution: different methods of estimation

This paper addresses different methods of estimation of the unknown parameters of a two-parameter Kumaraswamy distribution from a frequentist point of view. We briefly describe ten different frequentist approaches, namely, maximum likelihood estimators, moments estimators, L-moments estimators, percentile based estimators, least squares estimators, weighted least squares estimators, maximum product of spacings estimators, Cramér–von-Mises estimators, Anderson–Darling estimators and right tailed Anderson–Darling estimators. Monte Carlo simulations and two real data applications are performed to compare the performances of the estimators for both small and large samples.

97 MATHEMATICS AND COMPUTING↗

Fast Gaussian Process Estimation for Large-Scale In Situ Inference using Convolutional Neural Networks

Exascale computing will bring with it significant I/O limitations. One foreseeable consequence of such restrictions is that the user can save only a small fraction of complex simulation data to disk for subsequent analysis. An alternative is to fit statistical models to data in situ, that is, inside the simulation as it runs. This option requires extremely fast statistical estimation to avoid slowing down the simulation. Gaussian processes (GPs) have state-of-the-art predictive performance for modeling spatial data. However, standard estimation methods for GPs scale quite poorly to large data sets as parameter estimation requires inverting a covariance matrix to the size of the data set. In the presented work, we use a convolutional neural network (CNN) to predict the GP parameters for a spatial data set, from a simulation or otherwise, rather than optimize the parameters directly. Here, our presented case study models spatial data from E3SM, the Department of Energy’s Exascale climate model. The CNN is trained on synthetic data simulated from GP models with known parameters and then applied to data from the climate simulation. In the presented examples, the neural network scheme produces parameter estimates that compare well with standard methods such as maximum likelihood estimation in predictive performance but is obtained four orders of magnitude faster.

big data↗

Maximum Likelihood Source Localization Using Elpasolite as a Dual Gamma Neutron Directional Detector

The problem of accurately detecting extremely low levels of nuclear radiation is rapidly increasing in importance in nuclear counter-proliferation, verification, and environmental and waste management. Because the 239Pu gamma signature may be weak, for instance, even when compared to the natural terrestrial background, coincidence counting with the 239Pu neutron signature may improve overall 239Pu detection sensitivity. However, systems with sufficient multiple-particle detectors require demonstration that the increased sensitivity be sufficiently high to overcome added cost and weight. We report the results of measurements and calculations to determine sensitivity that can be gained in detecting low levels of nuclear radiation from use of a relatively new detector technology based on elpasolite crystals. We have performed investigations exploring cerium (Ce3+)-doped elpasolites Cs2LiYCl6:Ce3+0.5% (CLYC) and Cs2LiLa(Br6)90%(Cl6)10%:Ce3+0.5% (CLLBC:Ce). These materials can provide energy resolution (r(E) = 2.35σ(E)/E) as good as 2.9% at 662 keV (FWHM). The crystals show an excellent neutron and gamma radiation response. The goals of the investigation were to set up the neutron/gamma pulse shape discrimination electronics for elpasolite detectors; perform limited static source benchmarking, testing, and evaluation to validate system performance; and explore application of a maximum likelihood algorithm for source location. Data were measured and processed through a maximum likelihood estimation algorithm, providing a direction to the radioactive source for each individual position. The estimated directions were good representations for the actual directions to the radioactive source. This paper summarizes the maximum likelihood results for our elpasolite system.

maximum likelihood, radiation detectors, neutron d↗

Scalable computations for nonstationary Gaussian processes

Nonstationary Gaussian process models can capture complex spatially varying dependence structures in spatial datasets. However, the large number of observations in modern datasets makes fitting such models computationally intractable with conventional dense linear algebra. In addition, derivative-free or even first-order optimization methods can be very slow to converge when estimating many spatially varying parameters. In this paper, we present a computational framework which couples an algebraic block diagonal plus low-rank covariance matrix approximation with stochastic trace estimation to facilitate the efficient use of second-order solvers for maximum likelihood estimation of Gaussian process models with many parameters. We demonstrate the effectiveness of these methods by simultaneously fitting 192 parameters in the popular nonstationary model of Paciorek and Schervish using 107,600 sea surface temperature anomaly measurements.

97 MATHEMATICS AND COMPUTING↗

A Model for Measurements of Lognormally Distributed Environmental Contaminants

This paper proposes a more nearly reasonable model for the actual measurement distribution, called here the “Davis Mixed Model” (DMM). The DMM is derived by multiplying the probability density function of unobservable actual concentrations (assumed LN) by the conditional density of measurements given the concentrations (assumed heteroscedastic normal), and then integrating to obtain the marginal distribution of the observable measurements. The DMM is complicated and analytically intractable; its probability density function (PDF) is itself an integral, for example, and closed-form expressions for percentiles, let alone estimators, do not exist. The DMM can be fit to data via Maximum Likelihood Estimation (MLE), however, and a fitted model can be used to generate data for evaluating the actual performance of candidate UTL or other estimation procedures. The Industrial Hygiene application motivating this work involves surface sampling surveys for removable beryllium (Be) contamination, with data from Inductively Coupled Plasma – Atomic Emission Spectroscopy (ICP-AES) analyses. Similar issues will arise quite generally with censored environmental data for other contaminants and analytical methods. The conclusions presented in this paper focus on the regions of the DMM parameter space arising in surveying numerous Department of Energy (DOE) facilities associated with the Nevada Test Site (NTS).

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Biases on cosmological parameter estimators from galaxy cluster number counts

Sunyaev-Zel'dovich (SZ) surveys are promising probes of cosmology — in particular for Dark Energy (DE) —, given their ability to find distant clusters and provide estimates for their mass. However, current SZ catalogs contain tens to hundreds of objects and maximum likelihood estimators may present biases for such sample sizes. In this work we study estimators from cluster abundance for some cosmological parameters, in particular the DE equation of state parameter w{sub 0}, the amplitude of density fluctuations σ{sub 8}, and the Dark Matter density parameter Ω{sub c}. We begin by deriving an unbinned likelihood for cluster number counts, showing that it is equivalent to the one commonly used in the literature. We use the Monte Carlo approach to determine the presence of bias using this likelihood and study its behavior with both the area and depth of the survey, and the number of cosmological parameters fitted. Our fiducial models are based on the South Pole Telescope (SPT) SZ survey. Assuming perfect knowledge of mass and redshift some estimators have non-negligible biases. For example, the bias of σ{sub 8} corresponds to about 40% of its statistical error bar when fitted together with Ω{sub c} and w{sub 0}. Including a SZ mass-observable relation decreases the relevance of the bias, for the typical sizes of current SZ surveys. Considering a joint likelihood for cluster abundance and the so-called ''distance priors'', we obtain that the biases are negligible compared to the statistical errors. However, we show that the biases from SZ estimators do not go away with increasing sample sizes and they may become the dominant source of error for an all sky survey at the SPT sensitivity. Finally, we compute the confidence regions for the cosmological parameters using Fisher matrix and profile likelihood approaches, showing that they are compatible with the Monte Carlo ones. The results of this work validate the use of the current maximum likelihood methods for present SZ surveys, but highlight the need for further studies for upcoming experiments. To perform the analyses of this work, we developed fast, accurate, and adaptable codes for cluster counts in the framework of the Numerical Cosmology Library.

79 ASTRONOMY AND ASTROPHYSICS↗

Likelihood Maximization and Moment Matching in Low SNR Gaussian Mixture Models

We derive an asymptotic expansion for the log-likelihood of Gaussian mixture models (GMMs) with equal covariance matrices in the low signal-to-noise regime. The expansion reveals an intimate connection between two types of algorithms for parameter estimation: the method of moments and likelihood optimizing algorithms such as Expectation-Maximization (EM). We show that likelihood optimization in the low SNR regime reduces to a sequence of least squares optimization problems that match the moments of the estimate to the ground truth moments one by one. This connection is a stepping stone towards the analysis of EM and maximum likelihood estimation in a wide range of models. A motivating application for the study of low SNR mixture models is cryo-electron microscopy data, which can be modeled as a GMM with algebraic constraints imposed on the mixture centers. We discuss the application of our expansion to algebraically constrained GMMs, among other example models of interest. © 2022 The Authors. Communications on Pure and Applied Mathematics published by Wiley Periodicals LLC.

97 MATHEMATICS AND COMPUTING↗

Expectation-maximization algorithms for learning a finite mixture of univariate survival time distributions from partially specified class values

Heterogeneity exists on a data set when samples from di erent classes are merged into the data set. Finite mixture models can be used to represent a survival time distribution on heterogeneous patient group by the proportions of each class and by the survival time distribution within each class as well. The heterogeneous data set cannot be explicitly decomposed to homogeneous subgroups unless all the samples are precisely labeled by their origin classes; such impossibility of decomposition is a barrier to overcome for estimating nite mixture models. The expectation-maximization (EM) algorithm has been used to obtain maximum likelihood estimates of nite mixture models by soft-decomposition of heterogeneous samples without labels for a subset or the entire set of data. In medical surveillance databases we can find partially labeled data, that is, while not completely unlabeled there is only imprecise information about class values. In this study we propose new EM algorithms that take advantages of using such partial labels, and thus incorporate more information than traditional EM algorithms. We particularly propose four variants of the EM algorithm named EM-OCML, EM-PCML, EM-HCML and EM-CPCML, each of which assumes a specific mechanism of missing class values. We conducted a simulation study on exponential survival trees with five classes and showed that the advantages of incorporating substantial amount of partially labeled data can be highly signi cant. We also showed model selection based on AIC values fairly works to select the best proposed algorithm on each specific data set. A case study on a real-world data set of gastric cancer provided by Surveillance, Epidemiology and End Results (SEER) program showed a superiority of EM-CPCML to not only the other proposed EM algorithms but also conventional supervised, unsupervised and semi-supervised learning algorithms.

97 MATHEMATICS AND COMPUTING↗

A ROBUST DETERMINATION OF THE SIZE OF QUASAR ACCRETION DISKS USING GRAVITATIONAL MICROLENSING

Using microlensing measurements for a sample of 27 image pairs of 19 lensed quasars we determine a maximum likelihood estimate for the accretion disk size of an average quasar of r{sub s} = 4.0{sup +2.4}{sub -3.1} lt-day at rest frame ({lambda}) = 1736 Angstrom-Sign for microlenses with a mean mass of (M) = 0.3 M{sub Sun }. This value, in good agreement with previous results from smaller samples, is roughly a factor of five greater than the predictions of the standard thin disk model. The individual size estimates for the 19 quasars in our sample are also in excellent agreement with the results of the joint maximum likelihood analysis.

79 ASTRONOMY AND ASTROPHYSICS↗