Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “probabilistic modeling”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,027 records · Page 57

Assessing State Nuclear Weapons Proliferation: Using Bayesian Network Analysis of Social Factors

A Bayesian network (BN) model of social factors can support proliferation assessments by estimating the likelihood that a state will pursue a nuclear weapon. Social factors including political, economic, nuclear capability, security, and national identity and psychology factors may play as important a role in whether a State pursues nuclear weapons as more physical factors. This paper will show how using Bayesian reasoning on a generic case of a would-be proliferator State can be used to combine evidence that supports proliferation assessment. Theories and analysis by political scientists can be leveraged in a quantitative and transparent way to indicate proliferation risk. BN models facilitate diagnosis and inference in a probabilistic environment by using a network of nodes and acyclic directed arcs between the nodes whose connections, or absence of, indicate probabilistic relevance, or independence. We propose a BN model that would use information from both traditional safeguards and the strengthened safeguards associated with the Additional Protocol to indicate countries with a high risk of proliferating nuclear weapons. This model could be used in a variety of applications such a prioritization tool and as a component of state safeguards evaluations. This paper will discuss the benefits of BN reasoning, the development of Pacific Northwest National Laboratory’s (PNNL) BN state proliferation model and how it could be employed as an analytical tool.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Leveraging Structured Biological Knowledge for Counterfactual Inference: A Case Study of Viral Pathogenesis

Counterfactual inference is a useful tool for comparing outcomes of interventions on complex systems. It requires us to represent the system in form of a structural causal model, complete with a causal diagram, probabilistic assumptions on exogenous variables, and functional assignments. Specifying such models can be extremely difficult in practice. The process requires substantial domain expertise, and does not scale easily to large systems, multiple systems, or novel system modifications. At the same time, many application domains, such as molecular biology, are rich in structured causal knowledge that is qualitative in nature. This manuscript proposes a general approach for querying a causal knowledge graph with a causal question and converting the qualitative result into a quantitative structural causal model that can learn from data to answer the question. Here, we demonstrate the feasibility, accuracy and versatility of this approach using two case studies in systems biology. The first demonstrates the appropriateness of the underlying assumptions and the accuracy of the results. The second demonstrates the versatility of the approach by querying a knowledge base for the molecular determinants of a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-induced cytokine storm and performing counterfactual inference to predict the causal effect of medical countermeasures for severely ill COVID-19 patients.

60 APPLIED LIFE SCIENCES↗

A comprehensive modeling procedure for estimating statistical properties of forced ignition

In this work, a comprehensive modeling procedure for estimating the probability of ignition with application to high altitude relights of aircraft combustors is developed. In these configurations, an ignitor is used to introduce high-enthalpy discharge into a fuel-laden but stratified flow. Because of the inherent variabilities inflow conditions, kernel discharge process, and the chaotic turbulent flow, ignition cannot be described deterministically, but only as a probabilistic measure. The proposed modeling framework consists of three components. The turbulent flow is represented using the large eddy simulation (LES) framework. The ignition process is modeled using a manifold approach, where the initial kernel evolution is represented using a homogeneous reactor while the latter part of the evolution is represented as a competition between diffusion and chemical reactions using a flamelet-type mapping. A combined lookup table that can track the evolution of the kernel through these two distinct reaction stages is developed. The table lookup variables are solved within the LES framework. The resulting simulation tool is then embedded within an uncertainty quantification approach, where variations in the turbulent flow, as well as operating and kernel properties, are simulated using a Monte-Carlo-based sampling approach. Techniques to reduce computational cost are used to obtain a robust, numerically accurate, and physically representative model for engine relight. The method is validated using experimental data for ignition of methane/air mixtures. Due to the comprehensive nature of the modeling procedure, it is found that the simulation tool reproduces experimentally observed ignition probabilities over a wide range of operating conditions.

42 ENGINEERING↗

A DATA-DRIVEN MODEL FOR SPECTRA: FINDING DOUBLE REDSHIFTS IN THE SLOAN DIGITAL SKY SURVEY

We present a data-driven method-heteroscedastic matrix factorization, a kind of probabilistic factor analysis-for modeling or performing dimensionality reduction on observed spectra or other high-dimensional data with known but non-uniform observational uncertainties. The method uses an iterative inverse-variance-weighted least-squares minimization procedure to generate a best set of basis functions. The method is similar to principal components analysis (PCA), but with the substantial advantage that it uses measurement uncertainties in a responsible way and accounts naturally for poorly measured and missing data; it models the variance in the noise-deconvolved data space. A regularization can be applied, in the form of a smoothness prior (inspired by Gaussian processes) or a non-negative constraint, without making the method prohibitively slow. Because the method optimizes a justified scalar (related to the likelihood), the basis provides a better fit to the data in a probabilistic sense than any PCA basis. We test the method on Sloan Digital Sky Survey (SDSS) spectra, concentrating on spectra known to contain two redshift components: these are spectra of gravitational lens candidates and massive black hole binaries. We apply a hypothesis test to compare one-redshift and two-redshift models for these spectra, utilizing the data-driven model trained on a random subset of all SDSS spectra. This test confirms 129 of the 131 lens candidates in our sample and all of the known binary candidates, and turns up very few false positives.

79 ASTRONOMY AND ASTROPHYSICS↗

Probabilistic Prognosis of Non-Planar Fatigue Crack Growth

Quantifying the uncertainty in model parameters for the purpose of damage prognosis can be accomplished utilizing Bayesian inference and damage diagnosis data from sources such as non-destructive evaluation or structural health monitoring. The number of samples required to solve the Bayesian inverse problem through common sampling techniques (e.g., Markov chain Monte Carlo) renders high-fidelity finite element-based damage growth models unusable due to prohibitive computation times. However, these types of models are often the only option when attempting to model complex damage growth in real-world structures. Here, a recently developed high-fidelity crack growth model is used which, when compared to finite element-based modeling, has demonstrated reductions in computation times of three orders of magnitude through the use of surrogate models and machine learning. The model is flexible in that only the expensive computation of the crack driving forces is replaced by the surrogate models, leaving the remaining parameters accessible for uncertainty quantification. A probabilistic prognosis framework incorporating this model is developed and demonstrated for non-planar crack growth in a modified, edge-notched, aluminum tensile specimen. Predictions of remaining useful life are made over time for five updates of the damage diagnosis data, and prognostic metrics are utilized to evaluate the performance of the prognostic framework. Challenges specific to the probabilistic prognosis of non-planar fatigue crack growth are highlighted and discussed in the context of the experimental results.

Leser, Patrick E.↗

CHP-PRA Proof-of-Concept [Simulation] Sensitivity Assessment

An effort is underway to establish a Crew Health and Performance system (CHP) tradespace tool using a Probabilistic Risk Assessment (PRA) modeling and simulation system. The goal of the CHP-PRA effort is to provide a means of quantifying the integrated influence of CHP functions and capabilities on risk outcome metrics associated with health, performance, and long-term health. These metrics can then be used to establish potential risk-based trades on CHP system designed functionality and capabilities. Previously, our team demonstrated a proof-of-concept PRA approach that estimated the integrated influence of exercise countermeasures on 8 human system risks (Figure 1a) with outcomes associated with health and medical risk metrics. We reported that the change in the integrated relative health risk was small (Figure 1b) and that the small change in overall risk resulted from compounding and competing contribution levels of the individual risks. This interesting observation illustrates the emergent complexity of even straightforward representations of the human health and performance risk space and the ability of PRA models to capture this balance of global risk concerns. A key question that is not addressed in the initial analysis is “even though the global risk is relatively nominal, do any of the local risks become unacceptable?” In essence, we seek to determine what relative change in the human system risks are contributing to the relatively muted sensitivity of the proof-of-concept model combined risk assessments. Evaluations at the component risk level should elucidate if any individual risk reaches a high level that is subsequentially balanced by reductions in other areas. To further understand the relative changes in the component risks in the proof-of-concept model, and to elucidate how future refinements can be targeted, a means of establishing the contributions of the robustness of the proof-of-concept approach will be demonstrated.

astronaut health↗

Enhancements to Engineering-scale Reactor Pressure Vessel Fracture Capabilities in Grizzly

The Grizzly code is being developed to model the effect of aging in nuclear power plant systems, components, and structures. A significant part of this effort has been to develop capabilites to model the effects of embrittlement in reactor pressure vessels on their integrity. This includes both modeling of microstructure and engineering property evolution, and engineering-scale probabilistic fracture mechanics analysis. This report documents recent advances to the engineering-scale fracture mechanics capability for evaluation of RPV integrity under transient loading in Grizzly. These developments are in three areas: probabilistic fracture mechanics, general reduced order models for fracture a flaw locations, and improvments to the XFEM capability used in Grizzly for fracture mechanics analysis. The combination of these developments brings Grizzly closer to a state where it can be applied in production as a general tool for engineering-scale fracture mechanics analysis. The probabilistic fracture mechanics capabilities are included in the 1.5 testing version of Grizzly, and will be further refined in preparation for production use in the 2.0 released planned for fiscal year 2018.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Incorporating polar field data for improved solar flare prediction

In this paper, we consider incorporating data associated with the sun’s north and south polar field strengths to improve solar flare prediction performance using machine learning models. When used to supplement local data from active regions on the photospheric magnetic field of the sun, the polar field data provides global information to the predictor. While such global features have been previously proposed for predicting the next solar cycle’s intensity, in this paper we propose using them to help classify individual solar flares. We conduct experiments using HMI data employing four different machine learning algorithms that can exploit polar field information. Additionally, we propose a novel probabilistic mixture of experts model that can simply and effectively incorporate polar field data and provide on-par prediction performance with state-of-the-art solar flare prediction algorithms such as the Recurrent Neural Network (RNN). Our experimental results indicate the usefulness of the polar field data for solar flare prediction, which can improve Heidke Skill Score (HSS2) by as much as 10.1%.

79 ASTRONOMY AND ASTROPHYSICS↗

Forecasting seasonal influenza with a state-space SIR model

Seasonal influenza is a serious public health and societal problem due to its consequences resulting from absenteeism, hospitalizations, and deaths. The overall burden of influenza is captured by the Centers for Disease Control and Prevention’s influenza-like illness network, which provides invaluable information about the current incidence. This information is used to provide decision support regarding prevention and response efforts. Despite the relatively rich surveillance data and the recurrent nature of seasonal influenza, forecasting the timing and intensity of seasonal influenza in the U.S. remains challenging because the form of the disease transmission process is uncertain, the disease dynamics are only partially observed, and the public health observations are noisy. Fitting a probabilistic state-space model motivated by a deterministic mathematical model [a susceptible-infectious-recovered (SIR) model] is a promising approach for forecasting seasonal influenza while simultaneously accounting for multiple sources of uncertainty. A significant finding of this work is the importance of thoughtfully specifying the prior, as results critically depend on its specification. Our conditionally specified prior allows us to exploit known relationships between latent SIR initial conditions and parameters and functions of surveillance data. We demonstrate advantages of our approach relative to alternatives via a forecasting comparison using several forecast accuracy metrics.

60 APPLIED LIFE SCIENCES↗

Waste Acceptance and Disposal at the Nevada National Security Site

The U.S. Department of Energy (DOE), National Nuclear Security Administration Nevada Field Office (NNSA/NFO) relies on the Performance Assessment (PA) and the PA model of the Area 5 Radioactive Waste Management Site (RWMS) at the Nevada National Security Site (NNSS) to make disposal decisions on waste stream profiles submitted by waste generators. Disposal decisions are made following a formal process called the PA review process. This paper presents the PA review process, which is a part of the NNSA/NFO’s Waste Generator Documentation Approval Process. The review process includes three major steps and employs quantitative tools such as the PA-derived factors and probabilistic inventory and PA models developed in GoldSim. The PA-derived radionuclide action levels for waste characterization and reporting are included in the NNSS Waste Acceptance Criteria. Each year, NNSA/NFO reviews approximately 100 to 150 new or revised waste stream profiles. The PA review group ensures that, with the disposal of approved waste streams, the Area 5 RWMS will remain in compliance with the performance objectives of DOE O 435.1, “Radioactive Waste Management.” Examples of the PA review process are given that demonstrate how waste streams with special characteristics are evaluated and disposal decisions are made. The most rigorous profile evaluation is done with the PA model, run with the proposed profile inventory added and results compared to the PA results. Additional analysis and process-modeling are performed for those waste streams with unique features not addressed with the PA model. The results of these analyses and modeling are documented in special analysis reports, reviewed and approved by NNSA/NFO. The examples include (1) profiles with radionuclides not in the PA, (2) profiles including radionuclides with high radon-generation potential, and (3) profiles with high-heat generating radionuclides.

Yucel, Vefa↗

Managing Solar Photovoltaic Integration in the Western United States: Resource Adequacy Considerations

This study examines the impact of reserve margin-based reliability assessment, as commonly used in capacity expansion models, on planning resource-adequate power systems under high penetrations of solar photovoltaics (PV). As a generation resource, PV is operationally different from the conventional dispatchable resources for which most capacity expansion models were designed. The question this study attempts to answer is whether large amounts of PV on a system (in this case, the Western Interconnection of North America) would bias the results of conventional reserve margin-based capacity expansion modeling towards an over- or under-provisioning of resource adequacy. This analysis used NREL’s Resource Planning Model (RPM) for capacity expansion modeling and NREL’s Probabilistic Resource Adequacy Suite (PRAS) for resource adequacy assessment. RPM uses a reserve margin requirement to enforce resource adequacy. PRAS, a collection of tools for studying the resource adequacy of power systems and the adequacy contributions of individual resources on a probabilistic basis, was used to compute multiple resource adequacy metrics across a number of simulated scenarios and system representations with differing regional detail. In all cases, including high PV penetrations (up to 33% annual generation from PV, interconnection-wide), RPM was able to produce resource-adequate systems as measured by normalized expected unserved energy and loss-of-load expectation results from PRAS. The accuracy of reserve margin approaches depends heavily on the underlying assumptions informing the capacity credit assigned to variable and energy-limited resources, particularly when such resources are abundant in the modeled system. RPM’s standard methodology for estimating variable and flexible resources’ capacity contributions, which is based on the top 100 hours of net load, did not appear to systematically undervalue or overvalue variable generation relative to a more rigorous equivalent firm capacity assessment using PRAS, although both over- and under-valuations were observed in specific scenarios. In the worst cases, the top 100 hour method underestimated the equivalent firm capacity of PV by two percentage points, and overestimated the equivalent firm capacity of PV by five percentage points. Calculating capacity contributions based on the top 10 hours of net load systematically underestimated equivalent firm capacities at more modest PV penetrations, but was often a better approximation of equivalent firm capacity than the existing 100-hour approach in scenarios with higher PV penetrations.

14 SOLAR ENERGY↗

Correlated Trajectory Uncertainty for Adaptive Sequential Decision Making

One of the great challenges with decision making tasks on real world systems is the fact that data is sparse and acquiring additional data is expensive. In these cases, it is often crucial to make a model of the environment to assist in making decisions. At the same time, limited data means that learned models are erroneous, making it just as important to equip the model with good predictive uncertainties. In the context of learning sequential decision making policies, these uncertainties can prove useful for informing which data to collect for the greatest improvement in policy performance \citep{mehta2021experimental, mehta2022exploration} or informing the policy about unsure regions of state and action space to avoid during test time \citep{yu2020mopo}. Additionally, assuming that realistic samples of the environment can be drawn, an adaptable policy can be trained that attempts to make optimal decisions for any given possible instance of the environment \citep{ghosh2022offline, chen2021offline}. In this work, we examine the so-called ``probabilistic neural network'' (PNN) model that is ubiquitous in model-based reinforcement learning (MBRL) works. We argue that while PNN models may have good marginal uncertainties, they form a distribution of non-smooth transition functions. Not only are these samples unrealistic and may hamper adaptability, but we also assert that this leads to poor uncertainty estimates when predicting multiple step trajectory estimates. To address this issue, we propose a simple sampling method that can be implemented on top of pre-existing models.We evaluate our sampling technique on a number of environments, including a realistic nuclear fusion task, and find that, not only do smooth transition function samples produce more calibrated uncertainties, but they also lead to better downstream performance for an adaptive policy.

Offline Reinforcement Learning↗

Evaluation of the Benefits of ATF, FLEX, and Passive Cooling System for an Enhanced Resilient PWR Model

This report, along with report INL/EXT-19-53556, documents the activities performed by Idaho National Laboratory (INL) during the fiscal year (FY) 2019 for the DOE Light Water Reactor Sustainability (LWRS) Program, Risk-Informed System Analysis (RISA) Pathway, Enhanced Resilient Plant (ERP) Systems research. This report includes additional accident scenarios for risk-informed ATF analysis, and examines the risk impacts of FLEX and passive cooling system on nuclear power plants (NPPs) from additional perspectives. The same analysis process, risk analysis approaches, and analysis tools as in FY 2018 were used for near-term ATF cladding (i.e., Iron-Chromium-Aluminum [FeCrAl] cladding and Chromium [Cr]-coated cladding) designs under the small break loss-of-coolant accident (SBLOCA) scenarios, two types of general transient scenarios (locked rotor and turbine trip), and main steam line break (MSLB) scenarios. Benchmark calculations for ATF analysis using RELAP5-3D and MELCOR are also presented. In addition, based on the generic probabilistic risk assessment (PRA) model, the risk impacts of FLEX and the dynamic natural convection (DNC) system on NPPs are evaluated through a risk-informed decision-making process, i.e., significance determination process (SDP).

99 GENERAL AND MISCELLANEOUS↗

Improving online risk assessment with equipment prognostics and health monitoring

The current approach to evaluating the risk of nuclear power plant (NPP) operation relies on static probabilities of component failure, which are based on industry experience with the existing fleet of nominally similar light water reactors (LWRs). As the nuclear industry looks to advanced reactor designs that feature non-light water coolants (e.g., liquid metal, high temperature gas, molten salt), this operating history is not available. Many advanced reactor designs use advanced components, such as electromagnetic pumps, that have not been used in the US commercial nuclear fleet. Given the lack of rich operating experience, we cannot accurately estimate the evolving probability of failure for basic components to populate the fault trees and event trees that typically comprise probabilistic risk assessment (PRA) models. Online equipment prognostics and health management (PHM) technologies can bridge this gap to estimate the failure probabilities for components under operation. The enhanced risk monitor (ERM) incorporates equipment condition assessment into the existing PRA and risk monitor framework to provide accurate and timely estimates of operational risk.

active and passive components↗

Stochastic agent-based model for predicting turbine-scale raptor movements during updraft-subsidized directional flights

Rapid expansion of wind energy development across the world has highlighted the need to better understand turbine-caused avian mortality. The risk to golden eagles (Aquila chrysaetos) is of particular concern due to their small population size and conservation status. Golden eagles subsidize their flight in part by soaring in orographic updrafts, which can place them in conflict with wind turbines utilizing the same low-altitude wind resource. Understanding the behavior of soaring raptors in varying atmospheric conditions can therefore be relevant to predicting and mitigating their risk of collision. We present a predictive movement model that simulates individual paths of golden eagles during directional flight (such as migration) that is subsidized by orographic updraft. We modeled eagles in a 50 km by 50 km study area in Wyoming containing three wind power plants with documented golden eagle collisions with turbines. The movement model is applicable to any region where ground elevation is known at turbine scale (50 m) and wind conditions are known at facility scale (3 km). For a given set of atmospheric conditions, the model simulates movements of thousands of orographic soaring eagles to produce a density map quantifying the relative probability of eagle presence. We validated the simulated tracks with GPS telemetry data showing four directional tracks made by golden eagles transiting through the area in 2019 and 2020. For each eagle track, validation was performed using the ratio of the model-simulated eagle presence likelihood with uniform eagle presence and the presence computed using directed random-walk movements. We found that the predictive performance of the model was significantly better (likelihood ratio 1) for low-altitude movements than high-altitude movements that can involve thermal-soaring. We employed the model to produce seasonal presence maps for migrating golden eagles. We found significant turbine-level variations in eagle presence between northerly and southerly migration routes through the study area. Overall, the proposed model offers a generalizable, probabilistic, and predictive tool to assist wind energy developers, ecologists, wildlife managers, and industry consultants in estimating the potential for conflict between soaring birds and wind turbines, thereby reducing the need for site-specific data on golden eagle movements.

17 WIND ENERGY↗

A Bayesian method for using simulator data to enhance human error probabilities assigned by existing HRA methods

In the past several years, several international organizations have begun to collect data on human performance in nuclear power plant simulators. The data collected provide a valuable opportunity to improve human reliability analysis (HRA), but these improvements will not be realized without implementation of Bayesian methods. Bayesian methods are widely used to incorporate sparse data into models in many parts of probabilistic risk assessment (PRA), but Bayesian methods have not been adopted by the HRA community. In this paper, we provide a Bayesian methodology to formally use simulator data to refine the human error probabilities (HEPs) assigned by existing HRA methods. We demonstrate the methodology with a case study, wherein we use simulator data from the Halden Reactor Project to update the probability assignments from the SPAR-H method. The case study demonstrates the ability to use performance data, even sparse data, to improve existing HRA methods. Furthermore, this paper also serves as a demonstration of the value of Bayesian methods to improve the technical basis of HRA.

99 GENERAL AND MISCELLANEOUS↗

Distribution Development for Residual Inventory at the New York West Valley Site - 20400

The New York State Energy Research and Development Authority (NYSERDA) is the owner of the Western New York Nuclear Service Center (WNYNSC), a 1,351 ha site located approximately 48 km south of Buffalo, New York. In 1962, Nuclear Fuel Services, Inc. (NFS) entered into Agreements with the Atomic Energy Commission and New York State to construct the first commercial reprocessing plant of nuclear fuel in the United States. NFS, a private company, built and operated the spent fuel reprocessing plant and waste disposal facilities, processing 640 Mg of spent nuclear fuel from 1966 to 1972 under an Atomic Energy Commission license. Nuclear fuel reprocessing operations ended in 1972 and never reopened, leaving behind radioactive and chemical wastes. Operations led to contamination in a number of facilities and locations. Some of that contamination has migrated from waste disposal zones to other layers, formations, and features on and off the WNYNSC. Phase I decommissioning activities are ongoing and involve the removal of a number of areas and structures that have been associated with contamination. The purpose of this work is to outline the approach for characterizing contamination not associated with disposed wastes, contaminated structures, or specific releases. In this work, the term, residual radiological activity, is used to describe environmental contamination that exists subsequent to the completion of Phase I decommissioning activities, that is not associated with disposed wastes, contaminated structures, or specific releases. Contamination from the Site was quantified relative to data that characterize the concentrations of radionuclides that exist in background. Background concentrations are those present in the area but having no influence from Site related activities. The existence of residual radiological activity that is elevated relative to background has the potential to contribute to future risks to human health and the environment. As a consequence, the residual inventory information is used to inform the West Valley Probabilistic Performance Assessment (PPA) model to characterize potential future risks to human health and the environment. The centralized West Valley Data Management System (DMS) was the source of information for the data assembled in this analysis. The DMS is a fairly large compilation consisting of thousands of records from investigation studies, with sample dates ranging from 1990 to present. Samples from monitoring wells, boreholes, geoprobe studies, surface water, surface soils, storm water outfalls, ventilation stack filters, plant and animal tissues, and more are included in the DMS. Results are typically reported in units of activity per unit volume. For the purpose of the analyses presented here, all results were converted into consistent units of pCi per unit volume. Since 1990, data have been collected from various locations across the WNYNSC at different times with varying frequency over the course of several decades. As a consequence, a number of potential issues can arise with respect to the assembly of a dataset that is deemed adequate for the characterization of residual radiological activity. These issues were assessed and resolved to the extent possible through careful consideration of the properties of the distributions. The intent was to use data which characterize the current state of the Site. Radionuclides can be designated to one of several groups depending on their origin. In this work the groups considered were 1) Naturally Occurring Radioactive Material (NORM), 2) fallout, and 3) Other (including power plant, medical research, etc). This grouping is a useful construct with respect to the interpretation of fixed laboratory results. For example, NORM radionuclides that exist within a decay chain should have approximately equivalent distributions of concentrations if they are representative of background conditions. Insights such as these can be used as a check to identify sample results that need to be further investigated or omitted due to issues associated with reported results from fixed laboratory analyses. This type of analysis provided a foundation for the assessment of the adequacy of sample results for use in subsequent components of an assessment. The general process for the assessment of residual radiological contamination at the Site consists of a sequence of several steps. First, for each analyte, several statistical tests were performed to assess the weight of evidence against the null hypothesis that the mean of the distribution of concentrations was equal to zero. If the mean of the distribution of concentrations for a given radionuclide was not found to be greater than zero, then it was removed from consideration as a component of the residual radiological contamination. If there was significant evidence to reject the hypothesis of the mean being equal to zero, the second step was to compare the distribution of the data from the Site to that of the corresponding background. A suite of tests was used to compare the distributions of the site and background data. The results of these tests were collectively used to determine if site data are elevated relative to background. The third step was to develop distributions using a Bayesian framework to characterize the distribution of mean of the increment present above background for each of the radionuclides. The Bayesian model implemented allowed for the comparison of site-specific records to background concentrations to better approximate contamination attributed to the Site. A final screening step was employed for radionuclides that exceed background. This screening step compared 95% upper confidence limits (UCLs) from the increment distribution developed in the previous step to the risk screening levels. This approach yields a list of analytes that were determined to be elevated relative to background.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Learning from Crowds by Modeling Common Confusions

Crowdsourcing provides a practical way to obtain large amounts of labeled data at a low cost. However, the annotation quality of annotators varies considerably, which imposes new challenges in learning a high-quality model from the crowdsourced annotations. In this work, we provide a new perspective to decompose annotation noise into common noise and individual noise and differentiate the source of confusion based on instance difficulty and annotator expertise on a per-instance-annotator basis. We realize this new crowdsourcing model by an end-to-end learning solution with two types of noise adaptation layers: one is shared across annotators to capture their commonly shared confusions, and the other one is pertaining to each annotator to realize individual confusion. To recognize the source of noise in each annotation, we use an auxiliary network to choose from the two noise adaptation layers with respect to both instances and annotators. Extensive experiments on both synthesized and real-world benchmarks demonstrate the effectiveness of our proposed common noise adaptation solution.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗