DATAWorks Sessions Archive

We created the Open Architectures Tradeoff (OAT) tool, a simple, computational game engine for rapidly exploring hypotheses about mission effectiveness in Battle Management Command and Control (BMC2). Each run of an OAT game simulates a military mission in contested airspace. Game objects represent U.S., adversary, and allied assets, each of which moves through the simulated airspace. Each U.S. asset has a Command and Control (C2) package the controls its actions—currently, neural networks form the basis of each U.S. asset’s C2 package. The weights of the neural network are randomized at the beginning of each game and are updated over the course of the game as the U.S. asset learns which of its actions lead to rewards, e.g., intercepting an adversary. Weights are updated via a Decentralized Partially Observable Markov Decision Process (Dec-POMDP) altered to accommodate a Reinforcement Learning paradigm. OAT allows a user to winnow down the trade space that should be considered when setting up more expensive and time-consuming campaign models. OAT could be used to weed out bad ideas for “fast failure”, thus avoiding waste of campaign modeling resources. Questions can be explored via OAT such as: Which combination of system capabilities is likely to be more or less effective in a particular military mission? For example, in an early analysis, OAT was used to test the hypothesis that increases in U.S. assets’ sensor range always lead to increases in mission effectiveness, quantified as the percent of adversaries intercepted. We ran over 2500 OAT games, each time varying the sensor range of U.S. assets and the density of adversary assets. Results show that increasing sensor range did lead to an increase in military effectiveness—but only up to a certain point. Once the sensor range surpassed approximately 10-15% of the simulated airspace size, no further gains were made in the percent of adversaries intercepted. Thus, campaign modelers should hesitate to devote resources to exploring sensor range in isolation. More recent OAT analyses are exploring more complex hypotheses regarding the trade space between sensor range and communications range.

Goodness-of-fit (GOF) testing is used in many applications, including statistical hypothesis testing to determine if a set of data come from a hypothesized distribution. In addition, combined probability tests are extensively used in meta-analysis to combine results from several independent tests to asses an overall null hypothesis. This paper summarizes a study conducted to determine which GOF and/or combined probability test(s) can be used to determine if a set of data with relative small sample size comes from the standard uniform distribution, U(0,1). The power against different alternative hypothesis of several GOF tests and combined probability methods were examined. The GOF methods included: Anderson-Darling, Chi-Square, Kolmogorov-Smirnov, Cramér-Von Mises, Neyman-Barton, Dudewicz-van der Meulen, Sherman, Quesenberry-Miller, Frosini, and Hegazy-Green; while thecombined probability test methods included: Fisher’s Combined Probability Test, Mean Z, Mean P, Maximum P, Minimum P, Logit P, and Sum Z. While no one method was determined to provide the best power in all situations, several useful methods to support model validation were identified.

Technology advances are accelerating at a rapid pace, with the potential to enable greater capability and power to the Warfighter. However, if human capabilities and limitations are not central to concepts, requirements, design, and development then new/upgraded weapons and systems will be difficult to train, operate, and maintain, may not result in the skills, job, grade, and manpower mix as projected, and may result in serious human error, injury or Soldier loss. The Army Human Systems Integration (HSI) program seeks to overcome these challenges by ensuring appropriate consideration and integration of seven technical domains: Human Factors Engineering (e.g., usability), Manpower, Personnel, Training, Safety and Occupational Health, Habitability, Force Protection and Survivability. The tradeoffs, constraints, and limitations occurring among and between these technical domains allows HSI to execute a coordinated, systematic process for putting the warfighter at the center of the design process – equipping the warfighter rather than manning equipment. To that end, the Army HSI Headquarters, currently as a directorate within the Army Headquarters Deputy Chief of Staff (DCS), G-1 develops strategies and ensures human systems factors are early key drivers in concepts, strategy, and requirements, and are fully integrated throughout system design, development, testing and evaluation, and sustainment The need to consider HSI factors early in the development cycle is critical. Too often, man-machine interface issues are not addressed until late in the development cycle (i.e. production and deployment phase) after the configuration of a particular weapon or system has been set. What results is a degraded combat capability, suboptimal system and system-of-systems integration, increased training and sustainment requirements, or fielded systems not in use. Acquisition test data are also good sources to glean HSI return on investment (ROI) metrics. Defense acquisition reports such as test and evaluation operational assessments identifies HSI factors as root causes when Army programs experience increase cost, schedule overruns, or low performance. This is identifiable by the number and type of systems that require follow-on test and evaluation (FOT&E), over reliance on field service representatives (FSRs), costly and time consuming engineering change requests (ECRs), or failures in achieving reliability, availability, and maintainability (RAM) key performance parameters (KPPs) and key system attributes (KSAs). In this presentation, we will present these data and submit several return on investment (ROI) metrics, closely aligned to the defense acquisition process, to emphasize and illustrate the value of HSI. Optimizing Warfighter-System performance and reducing human errors, minimizing risk of Soldier loss or injury, and reducing personnel and materiel life cycle costs produces data that are inextricably linked to early, iterative, and measurable HSI processes within the defense acquisition system.

The Air Force (AF) Human Systems Integration (HSI) program is led by the 711th Human Performance Wing’s Human Systems Integration Directorate (711 HPW/HP). 711 HPW HP provides direct support to system program offices and AF Major Commands (MAJCOMs) across the acquisition lifecycle from requirements development to fielding and sustainment in addition to providing home office support. With an ever-increasing demand signal for support, HSI practitioners within 711 HPW/HP assess HSI domain areas for human-centered risks and strive to ensure systems are designed and developed to safely, effectively, and affordably integrate with human capabilities and limitations. In addition to system program offices and MAJCOMs, 711 HPW/HP provides HSI support to AF Centers (e.g., AF Sustainment Center, AF Test Center), the AF Medical Service, and special cases as needed. The AF Global Strike Command (AFGSC) is the largest MAJCOM with several Programs of Record (POR), such as the B-1, B-2, and B-52 bombers, Intercontinental Ballistic Missiles (ICBM), Ground-Based Strategic Deterrent (GBSD), Airborne Launch Control System (ALCS), and other support programs/vehicles like the UH-1N. Mr. Anthony Thomas (711 HPW/HP), the AFGSC HSI representative, will discuss how 711 HPW/HP supports these programs at the MAJCOM headquarters level and in the system program offices.

Materials scientists, computer scientists and statisticians at LANL have teamed up to investigate how to make near real time decisions during fast-paced experiments. For instance, a materials scientist at a beamline typically has a short window in which to perform a number of experiments, after which they analyze the experimental data, determine interesting new experiments and repeat. In typical circumstances, that cycle could take a year. The goal of this research and development project is to accelerate that cycle so that interesting leads are followed during the short window for experiments, rather than in years to come. We detail some of our UQ work in materials science, including emulation, sensitivity analysis, and solving inverse problems, with an eye toward real-time decision making in experimental settings.

Model validation is a process for determining how accurate a model is when compared to a true value. The methodology uses uncertainty analysis in order to assess the discrepancy between a measured and predicted value. In the literature, there have been several area metrics introduced to handle these type of discrepancies. These area metrics were applied to problems that include aleatory uncertainty, epistemic uncertainty, and mixed uncertainty. However, these methodologies lack the ability to fully characterize the true dierences between the experimental and prediction data when mixed un- certainty exists in the measurements and/or in the predictions. This work will introduce a new area metric validation approach which aims to com- pensate for the shortcomings in current techniques. The approach will be described in detail and comparisons between existing metrics will be shown. To demonstrate its applicability the new area metric will be applied to a stagnation point calibration probe’s surface predictions for a low-enthalpy conditions. For this application, testing was preformed in the Hypersonic Materials Environmental Test System (HYMETS) facility located at NASA Langley Research Center.

Computational fluid dynamics is now considered to be an indispensable tool for the design and development of scramjet engine components. Unfortunately, the quantification of uncertainties is rarely addressed with anything other than sensitivity studies, so the degree of confidence associated with the numerical results remains exclusively with the subject matter expert that generated them. This practice must be replaced with a formal uncertainty quantification process for computational fluid dynamics to play an expanded role in the system design, development, and flight certification process. Given the limitations of current hypersonic ground test facilities, this expanded role is believed to be a requirement by some in the hypersonics community if scramjet engines are to be given serious consideration as a viable propulsion system. The present effort describes a simple, relatively low cost, nonintrusive approach to uncertainty quantification that includes the basic ingredients required to handle both aleatoric (random) and epistemic (lack of knowledge) sources of uncertainty. The nonintrusive nature of the approach allows the computational fluid dynamicist to perform the uncertainty quantification with the flow solver treated as a “black box”. Moreover, a large fraction of the process can be automated, allowing the uncertainty assessment to be readily adapted into the engineering design and development workflow. In the present work, the approach is applied to a model scramjet isolator problem where the desire is to validate turbulence closure models in the presence of uncertainty. In this context, the relevant uncertainty sources are determined and accounted for to allow the analyst to delineate turbulence model-form errors from other sources of uncertainty associated with the simulation of the facility flow.

No set of experimental data is perfect and researchers are aware that data from experimental studies invariably contain some margin of error. This is particularly true of studies with human subjects since human behavior is vulnerable to a range of intrinsic and extrinsic influences beyond the variables being manipulated in a controlled experimental setting. Potential sources of error may lead to wide variations in the interpretation of results and the formulation of subsequent implications. This talk will discuss specific sources of error and bias in the design of experiments and present systematic ways to overcome these effects. First, some of the basic errors in general experimental design will be discussed, including human errors, systematic errors and random errors. Second, we will explore specific types of experimental error that appear in human subjects research. Lastly, we will discuss the role of bias in experiments with human subjects. Bias is a type of systematic error that is introduced into the sampling or testing phase and encourages one outcome over another. Often, bias is the result of the intentional or unintentional influence that an experimenter may exert on the outcomes of a study. We will discuss some common sources of bias in research with human subjects, including biases in sampling, selection, response, performance execution, and measurement. The talk will conclude with a discussion of how errors and bias influence the validity of human subjects research and will explore some strategies for controlling these errors and biases

An overview of Deep Reinforcement Learning and it’s recent successes in creating high performing agents. Covering it’s application in “easy” environments up to massively complex multi-agent strategic environments. Will analyze the behaviors learned, discuss research challenges, and imagine future possibilities.

Collaborative autonomous sensor networks have recently been used in many applications including inspection, law enforcement, search and rescue, and national security. They offer scalable, low cost solutions which are robust to the loss of multiple sensors in hostile or dangerous environments. While often comprised of less capable sensors, the performance of a large network can approach the performance of far more capable and expensive platforms if nodes are effectively coordinating their sensing actions and data processing. This talk will summarize work to date at LLNL on distributed signal processing and decentralized optimization algorithms for collaborative autonomous sensor networks, focusing on ADMM-based solutions for detection/estimation problems and sequential greedy optimization solutions which maximize submodular functions, e.g. mutual information.

Cyber system defenders face the challenging task of continually protecting critical assets and information from a variety of malicious attackers. Defenders typically function within resource constraints, while attackers operate at relatively low costs. As a result, design and development of resilient cyber systems that support mission goals under attack, while accounting for the dynamics between attackers and defenders, is an important research problem. This talk will highlight decision support modeling challenges under uncertainty within non-cooperative cybersecurity settings. Multiple attacker-defender game formulations under uncertainty are discussed with steps for further research.

Software reliability models enable several quantitative predictions such as the number of faults remaining, failure rate, and reliability (probability of failure free operation for a specified period of time in a specified environment). This talk will describe recent efforts in collaboration with NASA, including (1) the development of an automated script for the SFRAT (Software Failure and Reliability Assessment Tool) to streamline application of software reliability methods to ongoing programs, (2) application to a NASA program, (3) lessons learned, (4) and future directions for model and tool development to support the practical needs of the software reliability and security assessment frameworks.

The U.S. Army Research Laboratory’s (ARL) Essential Research Program (ERP) on Artificial Intelligence & Machine Learning (AI & ML) seeks to research, develop and employ a suite of AI-inspired and ML techniques and systems to assist teams of soldiers and autonomous agents in dynamic, uncertain, complex operational conditions. Systems will be robust, scalable, and capable of learning and acting with varying levels of autonomy, to become integral components of networked sensors, knowledge bases, autonomous agents, and human teams. Three specific research gaps will be examined: (i) Learning in Complex Data Environments, (ii) Resource-constrained AI Processing at the Point-of-Need and (iii) Generalizable & Predictable AI. The talk will highlight ARL’s internal research efforts over the next 3-5 years that are connected, cumulative and converging to produce tactically-sensible AI-enabled capabilities for decision making at the tactical edge, specifically addressing topics in: (1) adversarial distributed machine learning, (2) robust inference & machine learning over heterogeneous sources, (3) adversarial reasoning integrating learned information, (4) adaptive online learning and (5) resource-constrained adaptive computing. The talk will also highlight collaborative research opportunities in AI & ML via ARL’s Army AI Innovation Institute (A2I2) which will harness the distributed research enterprise via the ARL Open Campus & Regional Campus initiatives.

“The wise use of one’s resources will keep one from poverty.” This is the definition of the proverbial saying “waste not, want not” according to www.dictionary.com. Indeed, one of the most common resources analysts encounter is text in free-form. This text might come from survey comments, feedback, websites, transcriptions of interviews, videos, etcetera. Notably, researchers have used wisely the information conveyed in text for many years. However, in many instances, the qualitative methods employed require numerous hours of reading, training, coding, and validating, among others. As technology continues to evolve, simple access to text data is blooming. For example, analysts conducting online studies can have thousands of text entries from participants’ comments. Even without recent advances in technology analysts have had access to text in books, letters, and other archival data for centuries. One important challenge, however, is figuring out how to make sense of text data without investing a large number of resources, time, and the effort involved in qualitative methodology or “old-school” quantitative approaches (such as reading a collection of 200 books and counting the occurrence of important terms in the text). This challenge has been solved in the information retrieval field –a branch of computer science—with the implementation of a technique called latent semantic analysis (LSA; Manning, Raghavan, & Schütze, 2008) and a closely related technique called topic analysis (TA; SAS Institute Inc., 2018). Undoubtedly, other quantitative methods for text analysis, such as latent Dirichlet analysis (Blei, Ng, & Jordan, 2003), are also apt for the task of unveiling knowledge from text data, but we restrict the discussion in this presentation to LSA and TA because these exclusively deal with the underlying structure of the text rather than identifying clusters. In this presentation, we aim to make quantitative text analysis –specifically LSA and TA– accessible to researchers and analysts from a variety of disciplines. We do this by leveraging understanding of a popular multivariate technique: principal components analysis (PCA). We start by describing LSA and TA by drawing comparisons and equivalencies to PCA. We make these comparisons in an intuitive, user-friendly manner and then through a technical description of mathematical statements, which rely on the singular value decomposition of a document-term matrix. Moreover, we explain the implementation of LSA and TA using statistical software to enable simple application of these techniques. Finally, we show a practical application of LSA and TA with empirical data of aircraft incidents.

Sensors that record sequences of measurements are now embedded in many products from wearable exercise watches to chemical and semiconductor manufacturing equipment. There is information in the shapes of the sensor stream curves that is highly predictive of a variety of outcomes such as the likelihood of a product failure event or batch yield. Despite this data now being common and readily available, it is often being used either inefficiently or not at all due to lack of knowledge and tools for how to properly leverage it. In this presentation, we will propose fitting splines to sensor streams and extracting features called functional principal component scores that offer a highly efficient low dimensional compression of the signal data. Then, we use these features as inputs into machine learning models like neural networks and LASSO regression models. Once one sees sensor data in this light, answering a wide variety of applied questions becomes a straightforward two stage process of data cleanup/functional feature extraction followed by modeling using those features as inputs.

Abstract: In applications where modeling and simulation runs are quick and cheap, space filling designs will give the tester all the information they need to make decisions about their system. In some applications however, this luxury does not exist, and each M&S run can be time consuming and expensive. In these scenarios, a sequential test approach provides an efficient solution where an initial screening is conducted, followed by an augmentation to fit specified models of interest. Until this point, no dedicated screening designs for UQ applications in resource constrained situations existed. Due to the Army’s frequent exposure to this type of situation, the need sparked a collaboration between Picatinny’s Statistical Methods and Analysis group and Professor V. Roshan Joseph of Georgie Tech, where a new type of UQ screening design was created. This paper provides a brief introduction to the design, its intended use, and a case study in which this new methodology was applied.

Whether it was in a Design of Experiments course or through your own work, you’ve no doubt seen and become well acquainted with the standard experimental design. You know the features: they’re “orthogonal” (no messy correlations to deal with), their correlation matrices are nice pretty diagonals, and they can only happen with run sizes of 4, 8, 12, 16, and so on. Well what if I told you that there existed optimal designs that defied convention. What if I told you that, yes, you can run an optimal design with, say, 5 factors in 9 runs. Or 10. Or even 11 runs! Join me as I show you a strange new world of optimal designs that are the best at what they do, even though they might not look very nice.

During the past several years, the DoD has begun applying rapid prototyping and fielding authorities—granted by Congress in the FY2016-FY2018 National Defense Authorization Acts (NDAA)—to many acquisition programs. Other programs have implemented an agile acquisition strategy where incremental capability is delivered in iterative cycles. As a result, Operational Test Agencies (OTA) have had to adjust their test processes to accommodate shorter test timelines and periodic delivery of capability to the warfighter. In this session, representatives from the Service OTAs will brief examples where they have implemented new practices and processes for conducting Operational Test on acquisition programs categorized as agile, DevOps, and/or Section 804 rapid-acquisition efforts. During the final 30 minutes of the session, a panel of OTA representatives will field questions from the audience concerning the challenges and opportunities related to test design, data collection, and analysis, that rapid-acquisition programs present.

NASA is investigating the dose-response relationship between quiet sonic boom exposure and community noise perceptions. This relationship is the key to possible future regulations that would replace the ban on commercial supersonic flights with a noise limit. We have built several Bayesian statistical models using pilot community study data. Using goodness of fit measures, we downselected to a subset of models which are the most appropriate for the data. From this subset of models we demonstrate how to calculate sample size requirements for a simplified example without any missing data. We also suggest how to modify the sample size calculation to account for missing data.

Swarms of small, fast speedboats can challenge even the most capable modern warships, especially when they operate in or near crowded shipping lanes. As part of the Navy’s operational testing of new ships and systems, at-sea live-fire tests against remote-controlled targets allow us to test our capability against these threats. To ensure operational realism, these events are minimally scripted and allow the crew to respond in accordance with their training. This is a trade-off against designed experiments, which ensure statistically optimal sampling of data from across the factor space, but introduce many artificialities. A recent test provided data on the effectiveness of naval gunnery. However, standard binomial (hit/miss) analyses fell short, as the number of misses was much larger than the number of hits. This prevented us from fitting more than a few factors and resulted in error bars so large as to be almost useless. In short, binomial analysis taught us nothing we did not already know. Recasting gunfire data from binomial (hit/miss) to continuous (time-to-kill) allowed us to draw statistical conclusions with tactical implications from these free-play, live-fire surface gunnery events. Using a censored-data analysis approach enabled us to make this switch and avoid the shortcomings of other statistical methods. Ultimately, our analysis provided the Navy with suggestions for improvements to its tactics and the employment of its weapons.

Recent data from Boeing’s Statistical Summary of Commercial Jet Airplane Accidents shows that Loss of Control – In Flight (LOC-I) is the leading cause of fatalities in commercial aviation accidents worldwide. The Commercial Aviation Safety Team (CAST), a joint government and industry effort tasked with reducing the rate of fatal accidents, requested that the National Aeronautics and Space Administration (NASA) conduct research on virtual day-visual meteorological conditions displays, such as synthetic vision, in order to combat LOC-I. NASA recently concluded a series of experiments using commercial pilots from various backgrounds to evaluate synthetic vision displays. This presentation will focus on the two most recent experiments: one conducted with the Navy’s Disorientation Research Device and one completed at NASA Langley Research Center that utilized the Microsoft HoloLens to display synthetic vision. Statistical analysis was done on aircraft performance data, pilot inputs, and a range of subjective questionnaires to assess the efficacy of the displays.

This presentation will use a notional armament system case-study to illustrate the use of M&S DOE, surrogate modeling, sensitivity analysis, multi-objective optimization and model calibration during early lifecycle development and design activities in the context of a new armament system. In addition to focusing on the statistician’s, data scientist’s, or analyst’s role and the key statistical techniques in engineering DoD systems, this presentation will also emphasize the non-statistical / engineering domain-specific aspects in a multidisciplinary design and development process which make uses of these statistical approaches at the subcomponent and subsystem-level as well as the end-to-end system modeling. A statistical engineering methodology which emphasizes the use of ‘virtual’ DOE-based model emulators developed at the subsystem-level and integrated using a systems-engineering architecture framework can yield a more tractable engineering problem compared to traditional ‘design-build-test-fix’ cycles or direct simulation of computationally expensive models. This supports a more informed prototype design for physical experimentation while providing a greater variety of materiel solutions thereby reducing development and testing cycles and time to field complex systems.

Combining physical measurements with computational models is key to many investigations involving validation and uncertainty quantification (UQ). This talk surveys some of the many approaches taken for validation and UQ, with large-scale computational models. Experience with such applications suggests classifications of different types of problems with common features (e.g. data size, amount of empiricism in the model, computational demands, availability of data, extent of extrapolation required, etc.). More recently, social and social-technical systems are being considered for similar analyses, bringing new challenges to this area. This talk will approaches for such problems and will highlight what might be new research directions for application and methodological development in UQ.

This case study highlights activities performed during the front-end process of a software development effort undertaken by the Fire Support Command and Control Program Office. This program office provides the U.S. Army, Joint and coalition commanders with the capability to plan, execute and deliver both lethal and non-lethal fires. Recently, the program office has undertaken modernization of its primary field artillery command and control system that has been in use for over 30 years. The focus of this case study is on the user-centered design process and activities taken prior to and immediately following contract award. A modified waterfall model comprised of three cyclic, yet overlapping phases (observation, visualization, and evaluation) provided structure for the iterative, user-centered design process. Gathering and analyzing data collected during focus groups, observational studies, and workflow process mapping, enabled the design team to identify 1) design patterns across the role/duty, unit and echelon matrix (a hierarchical organization structure), 2) opportunities to automate manual processes, 3) opportunities to increase efficiencies for fire mission processing, 4) bottlenecks and workarounds to be eliminated through design of the modernized system, 5) shortcuts that can be leveraged in design, 6) relevant and irrelevant content for each user population for streamlining access to functionality, 7) a usability baseline for later comparison (e.g., the number of steps and time taken to perform a task as captured in workflows for comparison to the same task in the modernized system), and provided the basis for creating visualizations using wireframes. Heuristic evaluations were conducted early to obtain initial feedback from users. In the next few months, usability studies will enable users to provide feedback based on actual interaction with the newly designed software. Included in this case study are descriptions of the methods used to collect user-centered design data, how results were visualized/documented for use by the development team, and lessons learned from applying user-centered design techniques during software development of a military field artillery command and control system.

Due to the immense potential use cases, configurations, and threat behaviors, thorough and efficient cyber testing is a significant challenge for the defense community. In this presentation, Phadke will present case studies where STO was successfully deployed for cyber testing, resulting in higher assurance, reduced schedule, and reduced testing cost. Phadke will also discuss importance first focusing on the engineering and science analysis, and only after that is complete, implementing statistical methods.

Advances in high performance computing have enabled detailed simulations of real-world physical processes, and these simulations produce large datasets. Even as detailed as they are, these simulations are only approximations of imperfect mathematical models, and furthermore, their outputs depend on inputs that are themselves uncertain. The main goal of a validation and uncertainty quantication methodology is to determine the uncertainty, that is, the relationship between the true value of a quantity of interest and its prediction by the simulation. The value of the computational results is limited unless the uncertainty can be quantied or bounded. Bayesian calibration is a common method for estimating model parameters and quantifying their associated uncertainties; however, calibration becomes more complicated when the data arise from dierent types of experiments. On an example from material science we will employ two types of data and demonstrate how one can obtain a set of material strength models that agree with both data sources.

Communication is critical both for analysts and decision-makers who rely on analysis to inform their choices. The Service Academies are responsible for educating men and women who may serve in both roles over the course of their careers. Analysts must be able to summarize their results concisely and communicate them to the decision-maker in a way that is relevant and actionable. Decision-makers understand that analytical results may carry with them uncertainty and be able to incorporate this uncertainty properly when evaluating different options. This panel explores the role of the US Service Academies in preparing their students for both roles. Featuring representatives from the US Air Force Academy, the US Naval Academy, and the US Military Academy, this panel will cover how future US Officers are taught to use and communicate with data. Topics include developing and motivating numerical literacy, understandings of uncertainty, how analysts should frame uncertainty to decision-makers, and how decision-makers should understand information presented with uncertainty. Panelists will discuss what they think the academies do well and areas that are ripe for improvement.

This talk focuses on a means to characterize the variability in multivariate data. This characterization, given in terms of both probability distributions and closed sets, is instrumental in assessing and improving the robustness/reliability properties of system designs. To this end, we propose the Sliced-Normal (SN) class of distributions. The versatility of SNs enables modeling complex multivariate dependencies with minimal modeling effort. A polynomial mapping is defined which injects the physical space into a higher dimensional (so-called) feature space on which a suitable normal distribution is defined. Optimization-based strategies for the estimation of SNs from data in both physical and feature space are proposed. The formulations in physical space yield non-convex optimization programs whose solutions often outperform the solutions in feature space. However, the formulations in feature space yield either an analytical solution or a convex program thereby facilitating their application to problems in high dimensions. The superlevel sets of a SN density have a closed semi-algebraic form making them amenable to rigorous uncertainty quantification methods. Furthermore, we propose a chance-constrained optimization framework for identifying and eliminating the effects of outliers in the prescription of such regions. These strategies can be used to mitigate the conservatism intrinsic to many methods in system identification, fault detection, robustness/reliability analysis, and robust design caused by assuming parameter independence and by including outliers in the dataset.

Uncertainty quantification (UQ) has emerged as the science of quantitative characterization and reduction of uncertainties in simulation and testing. Stretching across applied mathematics, statistics, and engineering, UQ is a multidisciplinary field with broad applications. A popular UQ method to analyze the effects of input variability and uncertainty on the system responses is generalized Polynomial Chaos Expansion (gPCE). This method was developed using applied mathematics and does not require knowledge of a simulation’s physics. Thus, gPCE may be used across disparate industries and is applicable to both individual component and system level simulations. The gPCE method can encounter problems when any of the input configurations fail to produce valid simulation results. gPCE requires that results be collected on a sparse grid Design of Experiment (DOE), which is generated based on probability distributions of the input variables. A failure to run the simulation at any one input configuration can result in a large decrease in the accuracy of a gPCE. In practice, simulation data sets with missing values are common because simulations regularly yield invalid results due to physical restrictions or numerical instability. We propose a statistical approach to mitigating the cost of missing values. This approach yields accurate UQ results if simulation failure makes gPCE methods unreliable. The proposed approach addresses this missing data problem by introducing an iterative machine learning algorithm. This methodology allows gPCE modelling to handle missing values in the sparse grid DOE. The study will demonstrate the convergence characteristics of the methodology to reach steady state values for the missing points using a series of simulations and numerical results. Remarks about the convergence rate and the advantages and feasibility of the proposed methodology will be provided. Several examples are used to demonstrate the proposed framework and its utility including a secondary air system example from the jet engine industry and several non-linear test functions. This is based on joint work with Dr. Mark Andrews at SmartUQ.

One of the primary roles of navy surface combatants is defending high-value units against attack by anti-ship cruise missiles (ASCMs). They accomplish this either by launching their interceptor missiles and shooting the ASCMs down with rapid-firing guns (hard kill), or through the use of deceptive jamming, decoys, or other non-kinetic means (soft kill) to defeat the threat. The wide range of hostile ASCM capabilities and the different properties of friendly defenses, combined with the short time-scale for defeating these ASCMs, makes this a difficult problem to study. IDA recently completed a study focusing on the extent to which friendly forces were vulnerable to massed ASCM attacks, and possible avenues for improvement. To do this we created a pair of complementary models with the combined flexibility to explore a wide range of questions. The first model employed a set of closed-form equations, and the second a time-dependent Monte Carlo simulation. This presentation discusses the thought processes behind the models and their relative strengths and weaknesses.

The concept of sequential testing has many disparate meanings. Often, for statisticians it takes on a purely mathematical context while possibly meaning multiple disconnected test events for some practitioners. Here we present a pedagogical approach to creating test designs involving constrained factors using JMP software. Recent experiences testing one of the U.S. military’s fast jet life support systems (LSS) serves as a case study and back drop to support the presentation. The case study discusses several lessons learned during LSS testing, applicable to all practitioners of scientific test and analysis techniques (STAT) and design of experiments (DOE). We conduct a short analysis to specifically determine a test region with a set of factors pertinent to modeling human breathing and the use of breathing machines as part of the laboratory setup. A comparison of several government and industry laboratory test points and regions with governing documentation is made, along with the our proposal for determining a necessary and sufficient test region for tests involving human breathing as a factor.

Steven Thorsen, Sarah Burke &

A challenging aspect of a system reliability assessment is integrating multiple sources of information, including component, subsystem, and full-system data, previous test data, or subject matter expert opinion. A powerful feature of Bayesian analyses is the ability to combine these multiple sources of data and variability in an informed way to perform statistical inference. This feature is particularly valuable in assessing system reliability where testing is limited and only a small number (or no failures at all) are observed. The F-35 is DoD’s largest program; approximately one-third of the operations and sustainment cost is attributed to the cost of spare parts and the removal, replacement, and repair of components. The failure rate of those components is the driving parameter for a significant portion of the sustainment cost, and yet for many of these components, poor estimates of the failure rate exist. For many programs, the contractor produces estimates of component failure rates, based on engineering analysis and legacy systems with similar parts. While these are useful, the actual removal rates can provide a more accurate estimate of the removal and replacement rates the program anticipates to experience in future years. In this presentation, we show how we applied a Bayesian analysis to combine the engineering reliability estimates with the actual failure data to overcome the problems of cases where few data exist. Our technique is broadly applicable to any program where multiple sources of reliability information need be combined for the best estimation of component failure rates and ultimately sustainment costs.

Sensors abound in aerospace testing and while many scientists look at the data from a physics perspective, the comparative statistics information is what drives decisions. A multi-company project was comparing launch data from the 1980’s to a current set of data that included 30 sensors. Each sensor was designed to gather 3000 data points during the 3-second launch event. The data included temperature, acceleration, and pressure information. This talk will compare the data analysis methods developed for this project as well as the use of the new Functional Data Analysis tool within JMP for its ability to discern in-family launch performances.

Using Combinatorial Test Design methods to select software test scenarios has repeatedly delivered large efficiency and thoroughness gains – which begs the questions: • Why are these proven methods not used everywhere? • Why do some efforts to promote adoption of new approaches stagnate? • What steps can leaders take to introduce successfully introduce and spread new test design methods? For more than a decade, Justin Hunter has helped large global organizations across six continents adopt new test design techniques at scale. Working in some environments, he has felt like Sisyphus, forever condemned to roll a boulder uphill only to watch it roll back down again. In other situations, things clicked; teams smoothly adopted new tools and techniques, and impressive results were quickly achieved. In this presentation, Justin will discuss several common challenges faced by large organizations, explain why adopting test design tools is more challenging than adopting other types of development and testing tools, and share actionable recommendations to consider when you roll out new test design approaches.

Existing characterization of Europa’s environment is enabled by the Europa Clipper mission’s successful predecessors: Pioneer, Voyager, Galileo, and most recently, Juno. These missions reveal high intensity energetic particle fluxes at Europa’s orbit, requiring a multidimensional design challenge to ensure mission success (i.e. meeting Level 1 science requirements). Risk averse JPL Design Principles and the Europa Environment Requirement Document (ERD) dictate practices and policy, which if masterfully followed, are designed to protect Clipper from failure or degradation due to radiation. However, even if workmanship is flawless and no waivers are assessed, modeling errors, shielding uncertainty, and natural variation in the Europa environment are cause for residual concern. While failure and part degradation are of paramount concern, the occurrence of temporary outages, causing loss or degradation of science observations, is also a critical mission risk, left largely unmanaged by documents like the ERD. The referenced risk is monitored and assessed through a Project Systems Engineering-led mission robustness effort, which attempts balance the risk of science data loss with potential design cost and increased mission complexity required to mitigate such risk. The Science Sensitivity Model (SSM) was developed to assess mission and science robustness, with its primary goal being to ensure a high probability of achieving Level 1 (L1) science objectives by informing the design of a robust spacecraft, instruments, and mission design. This discussion will provide an overview of the problem, the model, and solution strategies. Subsequent presentations discuss the experimental design used to understand the problem space and the graphics and visualization used to reveal important conclusions.

Using the right visualizations as part of broad system and statistical Monte Carlo analysis supports interpretation of key drivers and relationships between variables, provides context about the full system, and communicates to non-statistician stakeholders. An experimental design was used to understand the relationships between instrument and spacecraft fault rate and recovery time in relation to the probability of achieving Europa Clipper science objectives during the Europa Clipper tour. Given spacecraft and instrument outages, requirement achievement checks were performed to determine the probability of meeting scientific objectives. Visualizations of the experimental design output enabled analysis of the full parameter set. Correlation between individual instruments and specific scientific objectives is not straight forward; some scientific objectives require a single instrument to be on at certain times and during varying conditions across the trajectory, while other science objectives require multiple instruments to function concurrently. By examining the input conditions that meet scientific objectives with the highest probability, and comparing those to trials with the lowest probability of meeting scientific objectives, key relationships could be visualized, enabling valuable mission and engineering design insights. Key system drivers of scientific success were identified, such as fault rate tolerance and recovery time required for each instrument and the spacecraft. Key steps, methodologies, difficulties and result-highlights are presented, along with a discussion of next steps and options for refinement and future analysis.

The Europa Clipper Science Sensitivity Model (SSM) can be thought of as a graph in which the nodes are mission requirements at ten levels in a hierarchy, and edges represent how requirements at one level of the hierarchy depend on those at lower levels. At the top of the hierarchy, there are ten nodes representing ten, Level 1 science requirements for the mission. At the bottom of the hierarchy, there are 100 or so nodes representing instrument-specific science requirements. In between, nodes represent intermediate science requirements with complex interdependencies. Meeting, or failing to meet, bottom-level requirements depends on the frequency of faults and the lengths of recovery times on the nine Europa Clipper instruments and the spacecraft. Our task was to design and analyze the results of a Monte Carlo experiment to estimate the probabilities of meeting the Level 1 science requirements based on parameters of the distributions of time between failures and of recovery times. We simulated an ensemble of synthetic missions in which failures and recoveries were random realizations from those distributions. The pass-fail status of the bottom-level instrument-specific requirements were propagated up the graph for each of the synthetic missions. Aggregating over the collection of synthetic missions produced estimates of the pass-fail probabilities for the Level 1 requirements. We constructed a definitive screening design and supplemented it with additional space-filing runs, using JMP 14 software. Finally, we used the vectors of failure and recovery parameters as predictors, and the pass-fail probabilities of the high-level requirements as responses, and built statistical models to predict the latter from the former. In this talk, we will describe the design considerations and review the fitted models and their implications for mission success.

A crucial step in the design and development of a fight vehicle, such as NASA’s Space Launch System (SLS), is understanding its vibration behavior while in fight. Vehicle designers rely on low-cost finite element analysis (FEA) to predict the vibration behavior of the vehicle. During ground and flight tests, sensors are strategically placed at predefined locations that contribute the most vibration information under the assumption that FEA is accurate, producing points to validate the FEA models. This collaborative work focused on developing optimal sensor placement algorithms to validate FEA models against test data, and to characterize the vehicles vibration characteristics.

Abstract: Data scientists spend approximately 80% of their time preparing, cleaning, and feature engineering data sets. In this talk I will share use cases that show why this is important and why we need to do it. I will also describe the Earth System Grid Federation (ESGF) which is an open source effort providing a robust, distributed data and computation platform, enabling world wide access to Peta/Exa-scale scientific data. ESGF will help reduce the amount of effort needed for climate data preprocessing by integrating the necessary analysis and data sharing tools.

In support of the Navy’s effort to obtain improved outcomes through data-driven decision making, The Center for Naval Analyses’ Data Science Program (CNA/DSP) supports the performance-to plan(P2P) forum, which is co chaired by the Vice Chief of Naval Operations and the Assistant Secretary of the Navy (RD&A). The P2P forum provides senior Navy leadership forward looking performance forecasts, which are foundational to articulating Navy progress toward readiness and capability goals. While providing analytical support for this forum, CNA/DSP leveraged machine learning techniques, including Random Forests and Artificial Neural Networks, to develop improved estimates of future maintenance durations for the Navy. When maintenance durations exceed their estimated timelines, these delays can affect training, manning, and deployments in support of operational commanders. Currently, the Navy creates maintenance estimates during numerous timeframes including the program objective memorandum (POM) process, the Presidential Budget (PB), and at contract award leading to evolving estimates over time. The limited historical accuracy for these estimates, especially with the POM and PB estimates, have persisted over the last decade. These errors have led to a gap between planned funding and actual costs in addition to changes in the assets available for operational commanders each year. The CNA/DSP prediction model reduces the average error in forecasted maintenance duration days from 128 days to 31 days for POM estimates. Improvements in duration accuracy for the PB and contract award time frames were also achieved using similar ML processes. The data curation for these models involved numerous data sources of varying quality and required significant feature engineering to provide usable model inputs that could allow for forecasts over the Future Years Defense Program (FYDP) in order to support improved resource allocation and scheduling in support of the optimized fleet response training plan (OFRTP).

The application opportunities for behavioral analytics in the cybersecurity space are based upon simple realities. 1. The great majority of breaches across all cybersecurity venues is due to human choices and human error. 2. With communication and information technologies making for rapid availability of data, as well as behavioral strategies of bad actors getting cleverer, there is need for expanded perspectives in cybersecurity prevention. 3. Internally-focused paradigms must now be explored that place endogenous protection from security threats as an important focus and integral dimension of cybersecurity prevention. The development of cybersecurity monitoring metrics and tools as well as the creation of intrusion prevention standards and policies should always include an understanding of the underlying drivers of human behavior. As temptation follows available paths, cyber-attacks follow technology, business models, and behavioral habits. The human element will always be the most significant part in the anatomy of any final decision. Choice options – from input, to judgement, to prediction, to action – need to be better understood for their relevance to cybersecurity work. Behavioral Performance Indexes harness data about aggregate human participation in an active system, helping to capture some of the detail and nuances of this critically important dimension of cybersecurity.

Is there ever a place for the third dimension in visualizing data? Is the use of 3D inherently bad, or can a 3D visualization be used as an effective tool to communicate results? In this talk, I will show you how you can create beautiful 2D and 3D maps and visualizations in R using the rayshader package. Additionally, I will talk about the value of 3D plotting and how good aesthetic choices can more clearly communicate results to stakeholders. Rayshader is a free and open source package for transforming geospatial data into engaging visualizations using a simple, scriptable workflow. It provides utilities to interactively map, plot, and 3D print data from within R. It was nominated by Hadley Wickham to be one of 2018’s Data Visualizations of the Year for the online magazine Quartz.

With nearly continuous recording of sensor values now common, a new type of data called “functional data” has emerged. Rather than the individual readings being modeled, the shape of the stream of data over time is being modeled. As an example, one might model many historical vibration-over-time streams of a machine at start-up to identify functional data shapes associated with the onset of system failure. Functional Principal Components (FPC) analysis is a new and increasingly popular method for reducing the dimensionality of functional data so that only a few FPCs are needed to closely approximate any of a set of unique data streams. When combined with Design of Experiments (DoE) methods the response to be modeled in as fewest tests as possible is now the shape of a stream of data over time. Example analyses will be shown where the form of the curve is modeled as the function of several input variables allowing one to determine the input settings associated with shapes indicative of good or poor system performance. This allows the analyst to predict the shape of the curve as a function of the input variables.

The Algorithmic Warfare Cross Functional Team (AWCFT or Project Maven) organizes DoD stakeholders to enhance intelligence support to the warfighter through the use of automation and artificial intelligence. The AWCFT’s objective is to turn the enormous volume of data available to DoD into actionable intelligence and insights at speed. This requires consolidating and adapting existing algorithm-based technologies as well as overseeing the development of new solutions. This brief will describe some of the methodological challenges in test and evaluation that the Maven team is working through to facilitate speedy and agile acquisition of reliable and effective AI / ML capabilities.

Systems with autonomy are beginning to permeate civilian, industrial, and military sectors. Though these technologies have the potential to revolutionize our world, they also bring a host of new challenges in evaluating whether these tools are safe, effective, and reliable. The Institute for Defense Analyses is developing methodologies to enable testing systems that can, to some extent, think for themselves. In this talk, we share how we think about this problem and how this framing can help you develop a test strategy for your own domain.

The Satellite Affordability in LEO (SAL) model identifies the cheapest constellation capable of providing a desired level of performance within certain constraints. SAL achieves this using a combination of analytical models, statistical emulators, and geometric relationships. SAL is flexible and modular, allowing users to customize certain components while retaining default behavior in other cases. This is desirable if users wish to consider an alternative cost formulation or different types of payload. Uses for SAL include examining cost tradeoffs with respect to factors like constellation size and desired performance level, evaluating the sensitivity of constellation costs to different assumptions about cost behavior, and providing a first-pass look at what proliferated smallsats might be capable of. At this point, SAL is limited to Walker constellations with sun-synchronous, polar orbits.

The Icing Research Tunnel (IRT) at NASA Glenn Research Center follows the recommended practice for icing tunnel calibration outlined in SAE’s ARP5905 document. The calibration team has followed the schedule of a full calibration every five years with a check calibration done every six months following. The liquid water content of the IRT has maintained stability within in the specifications presented to customers that the variation is within +/- 10% of the calibrated, target measurement. With recent measurements and instrumentation errors, a more thorough assessment of error source was desired. By constructing statistical process control charts, the ability to determine how the instrument varies in the short term, mid term, and long term was gained. The control charts offer a view of instrument error, facility error, or installation changes. It was discovered that there was a shift from target to mean baseline thus leading to the study of the overall capability indices of the liquid water content measuring instrument to perform within specifications defined in the IRT. This presentation describes data processing procedures for the Multi-Element Sensor in the IRT, including collision efficiency corrections, canonical correlation analysis, Chauvenet’s Criterion for rejection of data, distribution check of data, and mean, median and mode for construction of control charts. Further data is presented to describe the repeatability of the IRT with the Multi-Element Sensor and the ability to maintain a stable process for the defined calibration schedule.

This briefing discusses the practical advantages of using the probabilistic programming language (PPL) Stan to answer statistical questions, especially those related to the quantification of uncertainty. Stan is a relatively new statistical tool that allows users to specify probability models and reason about the processes that generate the data they encounter. Stan has quickly become a popular language for writing statistical models because it allows one to specify rich (or sparse) Bayesian models using high level language. Further, Stan is fast, memory efficient, and robust. Stan requires users be explicit about the model they wish to evaluate, which makes the process of statistical modeling more transparent to users and decision makers. This is valuable because it forces practitioners to consider assumptions at the beginning of the model building procedure, rather than at the end (or not at all). In this sense, Stan is the opposite of a “black box” modeling approach. This approach may be tedious and labor intensive at first, but the pay-offs are large. For example, once a model is set-up inferential tasks all essentially automatic, as changing the model does not change the how one analyzes the data. This is a generic approach to inference. To illustrate these points, we use Stan to study a ballistic miss distance problem. In ballistic missile testing, the p-content circular error probable (CEP) in the circle that contains p percent of future shots fired, on average. Statistically, CEP is a bivariate prediction region, constrained by the model to be circular. In Frequentist statistics, the determination of CEP is highly dependent on the model fit, and a different calculation of CEP must be produced for each plausible model. However, with Stan, we can approach the CEP calculation invariant of the model we use to fit the data. We show how to use Stan to calculate CEP and uncertainty intervals for the parameters using summary statistics. Statistical practitioners can access Stan from several programming languages, including R and Python.

Hardly a week goes by without news of a cybersecurity breach or an attack by cyber adversaries against a nation’s infrastructure. These incidents have wide-ranging effects, including reputational damage and lawsuits against corporations with poor data handling practices. Further, these attacks do not require the direction, support, or funding of technologically advanced nations; instead, significant damage can be – and has been – done with small teams, limited budgets, modest hardware, and open source software. Due to the significance of these threats, it is critical to analyze past events to predict trends and emerging threats. In this document, we present an implementation of a cybersecurity taxonomy and a methodology to characterize and analyze all stages of a cyberattack. The chosen taxonomy, MITRE ATT&CK™, allows for detailed definitions of aggressor actions which can be communicated, referenced, and shared uniformly throughout the cybersecurity community. We translate several open source cyberattack descriptions into the analysis framework, thereby constructing cyberattack data sets. These data sets (supplemented with notional defensive actions) illustrate example Red Team activities. The data collection procedure, when used during penetration testing and Red Teaming, provides valuable insights about the security posture of an organization, as well as the strengths and shortcomings of the network defenders. Further, these records can support past trends and future outlooks of the changing defensive capabilities of organizations. From these data, we are able to gather statistics on the timing of actions, detection rates, and cyberattack tool usage. Through analysis, we are able to identify trends in the results and compare the findings to prior events, different organizations, and various adversaries.

The average and standard deviation of, say, strength or dimensional test data are basic engineering math, simple to calculate. What those resulting values actually mean, however, may not be simple, and can be surprisingly different from what a researcher wants to calculate and communicate. Mistakes can lead to overlarge estimates of spread, structures that are over- or under-designed and other challenges to understanding or communicating what your data is really telling you. This talk will discuss some common errors and missed opportunities seen in engineering and scientific analyses along with mitigations that can be applied through smart and efficient test planning and analysis. It will cover when – and when not – to report a simple mean of a dataset based on the way the data was taken; why ignoring this often either hides or overstates risk; and a standard method for planning tests and analyses to avoid this problem. And it will cover what investigators can correctly (or incorrectly) say about means and standard deviations of data, including how and why to describe uncertainty and assumptions depending on what a value will be used for. The presentation is geared toward the engineer, scientist or project manager charged with test planning, data analysis or understanding findings from tests and other analyses. Attenders’ basic understanding of quantitative data analysis is recommended; more-experienced participants will grasp correspondingly more nuance from the pitch. Some knowledge of statistics is helpful, but not required. Participants will be challenged to think about an average as not just “the average”, but a valuable number that can and must relate to the engineering problem to be solved, and must be firmly based in the data. Attenders will leave the talk with a more sophisticated understanding of this basic, ubiquitous but surprisingly nuanced statistic and greater appreciation of its power as an engineering tool.

Causality in an engineered system pertains to how a system output changes due to a controlled change or intervention on the system or system environment. Engineered systems designs reflect a causal theory regarding how a system will work, and predicting the reliability of such systems typically requires knowledge of this underlying causal structure. The aim of this work is to introduce causal modeling tools that inform reliability predictions based on biased data sources and illustrate how these tools can inform data integration in practice. We present a novel application of the popular structural causal modeling framework to reliability estimation in an engineering application, illustrating how this framework can inform whether reliability is estimable and how to estimate reliability using data integration given a set of assumptions about the subject matter and data generating mechanism. When data are insufficient for estimation, sensitivity studies based on problem-specific knowledge can inform how much reliability estimates can change due to biases in the data and what information should be collected next to provide the most additional information. We apply the approach to a pedagogical example related to a real, but proprietary, engineering application, considering how two types of biases in data can influence a reliability calculation.

The identification of the Stuxnet worm in 2010 provided a highly publicized example of a cyber attack on an industrial control system. This raised public awareness about the possibility of similar attacks against other industrial targets – including critical infrastructure. Here, we use hypergames to analyze how adversarial perturbations can be used to manipulate a system using optimal control. Hypergames form an extension of game theory that models strategic interactions between players with significantly different perceptions of the game(s) they are playing. Previous work on hypergames has been limited to simple interactions, where a small set of discrete choices are available to each player. However, we apply hypergames to larger systems with continuous variables. Our results highlight that manipulating constraints can be a more effective attacker strategy than directly manipulating objective function parameters. Moreover, the attacker need not influence the underlying system to carry out a successful attack – it may be sufficient to deceive the defender controlling the system. Finally, we identify several characteristics that will make our analysis amenable to higher-dimensional control systems.

Military grade energetics are, by design, required to operate under extreme conditions. As such, warheads in a munition must demonstrate a high level of structural integrity in order to ensure safe and reliable operation by the Warfighter. In this example which involved an artillery munition, a systematic analytics-driven approach was executed which synthesized physical test data results with probabilistic analysis, non-destructive evaluation, modeling and simulation, and comprehensive risk analysis tools in order to determine the probability of a catastrophic event. Once the severity, probability of detection, occurrence, were synthesized, a model was built to determine the risk of a catastrophic event during firing which then accounts for defect growth occurring as a result of rough-handling. This comprehensive analysis provided a defensible, credible, and dynamic snapshot of risk while allowing for a transparent assessment of contribution to risk of the various inputs through sensitivity analyses. This paper will illustrate intersection of product safety, reliability, systems-safety policy, and analytics, and highlight the impact of a holistic multidisciplinary approach. The benefits of this rigorous assessment included quantifying risk to the user, supporting effective decision-making, improving resultant safety and reliability of the munition, and supporting triage and prioritization of future Non-Destructive Evaluation (NDE) screening efforts by identifying at-risk subpopulations.

Target detection in hyperspectral images (HSI) has broad value in defense applications, and neural networks have recently begun to be applied for this problem. A common criticism of neural networks is they give a point estimate with no uncertainty quantification (UQ). In defense applications, UQ is imperative because the cost of a false positive or negative is high. Users desire high confidence in either “target” or “not target” predictions, and if high confidence cannot be achieved, more inspection is warranted. One possible solution is Bayesian neural networks (BNN). Compared to traditional neural networks which are constructed by choosing a loss function, BNN take a probabilistic approach and place a likelihood function on the data and prior distributions for all parameters (weights and biases), which in turn implies a loss function. Training results in posterior predictive distributions, from which prediction intervals can be computed, rather than only point estimates. Heatmaps show where and how much uncertainty there is at any location and give insight into the physical area being imaged as well as possible improvements to the model. Using pytorch and pyro software, we test BNN on a simulated HSI scene produced using the Rochester Institute of Technology (RIT) Digital Imaging and Remote Sensing Image Generation (DIRSIG) model. The scene geometry used is also developed by RIT and is a detailed representation of a suburban neighborhood near Rochester, NY, named “MegaScene.” Target panels were inserted for this effort, using paint reflectance and bi-directional reflectance distribution function (BRDF) data acquired from the Nonconventional Exploitation Factors Database System (NEFDS). The target panels range in size from large to subpixel, with some targets only partially visible. Multiple renderings of this scene are created under different times of day and with different atmospheric conditions to assess model generalization. We explore the uncertainty heatmap for different times and environments on MegaScene as well as individual target predictive distributions to gain insight into the power of BNN.

Abstract. While fault testing a system with many factors each appearing at some number of levels, it may not be possible to test all combinations of factor levels. Most faults are caused by interactions of only a few factors, so testing interactions up to size t will often find all faults in the system without executing an exhaustive test suite. Call an assignment of levels to t of the factors a t-way interaction. A covering array is a collection of tests that ensures that every t-way interaction is covered by at least one test in the test suite. Locating arrays extend covering arrays with the additional feature that they not only indicate the presence of faults but locate the faulty interactions when there are no more than d faults in the system. If an array is (d, t)-locating, for every pair of sets of t-way interactions of size d, the interactions do not appear in exactly the same tests. This ensures that the faulty interactions can be differentiated from non-faulty interactions by the results of some test in which interactions from one set or the other but not both are tested. When the property holds for t-way interaction sets of size up to d, the notation (d, t ¯ ) is used. In addition to fault location, locating arrays have also been used to identify significant effects in screening experiments. Locating arrays are fairly new and few techniques have been explored for their construction. Most of the available work is limited to finding only one fault (d = 1). Known general methods require a covering array of strength t + d and produce many more tests than are needed. In this talk, we present Partitioned Search with Column Resampling (PSCR), a computational search algorithm to verify if an array is (d, t ¯ )-locating by partitioning the search space to decrease the number of comparisons. If a candidate array is not locating, random resampling is performed until a locating array is constructed or an iteration limit is reached. Algorithmic parameters determine which factor columns to resample and when to add additional tests to the candidate array. We use a 5 × 5 × 3 × 2 × 2 full factorial design to analyze the performance of the algorithmic parameters and provide guidance on how to tune parameters to prioritize speed, accuracy, or a combination of both. Last, we compare our results to the number of tests in locating arrays constructed for the factors and levels of real-world systems produced by other methods.

Scientific computing has undergone extraordinary growth in sophistication in recent years, enabling the simulation of a wide range of complex multiphysics and multiscale phenomena. Along with this increase in computational capability is the growing recognition that uncertainty quantification (UQ) must go hand-in-hand with numerical simulation in order to generate meaningful and reliable predictions for engineering applications. If not rigorously considered, uncertainties due to manufacturing defects, material variability, modeling assumptions, etc. can cause a substantial disconnect between simulation and reality. Packaging these complex computational models within an UQ framework, however, can be a significant challenge due to the need to repeatedly evaluate the model when even a single evaluation is time-consuming. This talk discusses efforts at NASA Langley Research Center (LaRC) to enable rapid UQ for problems with expensive computational models. Under the High Performance Computing Incubator (HPCI) program at LaRC, several open-source software libraries are being developed and released to provide access to general-purpose, state-of-the-art UQ algorithms. The common denominator of these methods is that they all expose parallelism among the model evaluations needed for UQ and, as such, are implemented to leverage HPC resources when available to achieve tremendous computational speedup. While the methods and software presented are broadly applicable, they will be demonstrated in the context of applications that have particular interest at NASA, including structural health management and trajectory simulation.

For chemical/biological testing, test chambers are sometimes designed with a vapor or aerosol homogeneity requirement. For example, a test community may require that the difference in concentration between any two test locations in a chamber be no greater than 20 percent. To validate the chamber, testers must demonstrate that such a requirement is met with a specified amount of certainty, such as 80 percent. With a validated chamber, multiple systems can be simultaneously tested at different test locations with the assurance that each system is exposed to nearly the same concentration. In some cases, however, homogeneity requirements are difficult to achieve. This presentation demonstrates a valid Bayesian method for testing probability of detection as a function of concentration in a chamber that fails to meet a homogeneity requirement. The demonstrated method of analysis is based on recent experience with an actual test chamber. Multiple systems are tested simultaneously at different locations in the chamber. Because systems tested in the chamber are exposed to different concentrations depending on these locations, the differences must be quantified to the greatest extent possible. To this end, data from the failed validation efforts are used to specify informative prior distributions for probability-of-detection modeling. Because these priors quantify and incorporate uncertainty in model parameters, they ensure that the final probability-of-detection model constitutes a valid comparison of the performance of the different systems.

Dr. Parker is Team Lead for Advanced Measurement Systems at the National Aeronautics and Space Administration’s Langley Research Center in Hampton, Virginia. He serves an Agency-wide statistical expert across all of NASA’s mission directorates of Exploration, Aeronautics, and Science to infuse statistical thinking, engineering, and methods including statistical design of experiments, response surface methodology, and measurement system characterization. His expertise is in collaboratively integrating research objectives, measurement sciences, test design, and statistical methods to produce actionable knowledge for aerospace research and development. He holds a B.S. in Mechanical Engineering, a M.S. in Applied Physics and Computer Science and a M.S. and Ph.D. in Statistics from Virginia Tech. Dr. Parker is a senior member of the American Institute for Aeronautics and Astronautics, American Society for Quality, and the American Statistical Association. Dr. Parker currently Chairs the American Society for Quality’s Publication Management Board and previously served as Editor-in-Chief of the journal Quality Engineering.

David Chu serves as President of the Institute for Defense Analyses. IDA is a non-profit corporation operating in the public interest. Its three federally funded research and development centers provide objective analyses of national security issues and related national challenges, particularly those requiring extraordinary scientific and technical expertise. As president, Dr. Chu directs the activities of more than 1,000 scientists and technologists. Together, they conduct and support research requested by federal agencies involved in advancing national security and advising on science and technology issues. Dr. Chu served in the Department of Defense as Under Secretary of Defense for Personnel and Readiness from 2001-2009, and earlier as Assistant Secretary of Defense and Director for Program Analysis and Evaluation from 1981-1993. From 1978-1981 he was the Assistant Director of the Congressional Budget Office for National Security and International Affairs. Dr. Chu served in the U. S. Army from 1968-1970. He was an economist with the RAND Corporation from 1970-1978, director of RAND’s Washington Office from 1994-1998, and vice president for its Army Research Division from 1998-2001. He earned a bachelor of arts in economics and mathematics, and his doctorate in economics, from Yale University. Dr. Chu is a member of the Defense Science Board and a Fellow of the National Academy of Public Administration. He is a recipient of the Department of Defense Medal for Distinguished Public Service with Gold Palm, the Department of Veterans Affairs Meritorious Service Award, the Department of the Army Distinguished Civilian Service Award, the Department of the Navy Distinguished Public Service Award, and the National Academy of Public Administration’s National Public Service Award.

Over the past 45 years, Mike’s primary focus has been on the management of research and development, focusing on making the results more useful in meeting the needs of the user community. Since 1984, he has specialized in communications, data and processing systems, including projects in NASA, the US Air Force, the FAA and the Census Bureau. Before that, he worked on Major System Acquisition Programs, in the Department of Defense including Marine Corps combat vehicles and US Navy submarines. Currently, Mike manages a comprehensive program to provide NASA’s Earth Science research efforts with the information technologies it will need in the 2020-2035 time-frame to characterize, model and understand the Earth. This Program addresses the full range of data lifecycle from generating data using instruments and models, through the management of the data and including the ways in which information technology can help to exploit the data. Of particular interest today are the ways in which NASA can measure and understand transient and transitional phenomena and the impact of climate change. The AIST Program focuses the application of applied math and statistics, artificial intelligence, case-based reasoning, machine learning and automation to improve our ability to use observational data and model output in understanding Earth’s physical processes and natural phenomena. Training and odd skills: Application of cloud computing US Government Computer Security US Navy Nuclear Propulsion operations and maintenance on two submarines

Wendy Martinez has been serving as the Director of the Mathematical Statistics Research Center at the Bureau of Labor Statistics (BLS) for six years. Prior to this, she served in several research positions throughout the Department of Defense. She held the position of Science and Technology Program Officer at the Office of Naval Research, where she established a research portfolio comprised of academia and industry performers developing data science products for the future Navy and Marine Corps. Her areas of interest include computational statistics, exploratory data analysis, and text data mining. She is the lead author of three books on MATLAB and statistics. Dr. Martinez was elected as a Fellow of the American Statistical Association (ASA) in 2006 and is an elected member of the International Statistical Institute. She was honored by the American Statistical Association when she received the ASA Founders Award at the JSM 2017 conference. Wendy is also proud and grateful to have been elected as the 2020 ASA President.

Dr. Laura Freeman is an Assistant Director of the Operational Evaluation Division at the Institute for Defense Analyses. In that position, she established and developed an interdisciplinary analytical team of statisticians, psychologists, and engineers to advance scientific approaches to DoD test and evaluation. Her focus areas include test design, statistical data analysis, modeling and simulation validation, human-system interactions, reliability analysis, software testing, and cybersecurity testing. Dr. Freeman currently leads a research task for the Chief Management Officer (CMO) aiming to reform DoD testing. She guides an interdisciplinary team in recommending changes and developing best practices. Reform initiatives include incorporating mission context early in the acquisition lifecycle, integrating all test activities, and improving data management processes. During 2018, Dr. Freeman served as that acting Senior Technical Advisor for Director Operational Test and Evaluation (DOT&E). As the Senior Technical Advisor, Dr. Freeman provided leadership, advice, and counsel to all personnel on technical aspects of testing military systems. She served as a liaison with Service technical advisors, General Officers, and members of the Senior Executive Service on key technical issues. She reviewed test strategies, plans, and reports from all systems on DOT&E oversight. During her tenure at IDA, Dr. Freeman has designed tests and conducted statistical analyses for programs of national importance including weapon systems, missile defense, undersea warfare systems, command and control systems, and most recently the F-35. She prioritizes supporting the analytical community in the DoD workforce. She developed and taught numerous courses on advanced test design and statistical analysis, including two new Defense Acquisition University (DAU) Courses on statistical methods. She is a founding organizer of DATAWorks (Defense and Aerospace Test and Analysis Workshop), a workshop designed to share new methods, provide training, and share best practices between NASA, the DoD, and National Labs. Dr. Freeman is the recipient of the 2017 IDA Goodpaster Award for Excellence in Research and the 2013 International Test and Evaluation Association (ITEA) Junior Achiever Award. She is a member of the American Statistical Association, the American Society for Quality, the International Statistical Engineering Association, and ITEA. She serves on the editorial boards for Quality Engineering, Quality Reliability Engineering International, and the ITEA Journal. Her areas of statistical expertise include designed experiments, reliability analysis, and industrial statistics. Prior to joining IDA in 2010, Dr. Freeman worked at SAIC providing statistical guidance to the Director, Operational Test and Evaluation. She also consulted with NASA on various projects. In 2008, Dr. Freeman established the Laboratory for Interdisciplinary Statistical Analyses at Virginia Tech and Served as its inaugural Director. Dr. Freeman has a B.S. in Aerospace Engineering, a M.S. in Statistics and a Ph.D. in Statistics, all from Virginia Tech. Her Ph.D. research was on design and analysis of experiments for reliability data.

Charles Clancy is the Bradley Professor of Electrical and Computer Engineering at Virginia Tech where he serves as the Executive Director of the Hume Center for National Security and Technology. Clancy leads a range of strategic programs at Virginia Tech related to security, including the Commonwealth Cyber Initiative. Prior to joining VT in 2010, Clancy was an engineering leader in the National Security Agency, leading research programs in digital communications and signal processing. He received his PhD from the University of Maryland, MS from University of Illinois, and BS from the Rose-Hulman Institute of Technology. He is co-author to over 200 peer-reviewed academic publications, six books, over twenty patents, and co-founder to five venture-backed startup companies.

Mr. Timothy S. Dare is the Deputy Director for Developmental Test, Evaluation and Prototyping (DD(DTEP)). As the DD(DTEP), he serves as the principal advisor on developmental test and evaluation (DT&E) to the Secretary of Defense, Under Secretary of Defense for Research and Engineering, and Director of Defense Research and Engineering for Advanced Capabilities. Mr. Dare is responsible for DT&E policy and guidance in support of the acquisition of major Department of Defense (DoD) systems, and providing advocacy, oversight, and guidance to the DT&E acquisition workforce. He informs policy and advances leading edge technologies through the development of advanced technology concepts, and developmental and operational prototypes. By working closely with interagency partners, academia, industry and governmental labs, he identifies, develops and demonstrates multi-domain technologies and concepts that address high-priority DoD, multi-Service, and Combatant Command warfighting needs. Prior to his appointment in December 2018, Mr. Dare was a Senior Program Manager for program management and capture at Lockheed Martin (LM) Space. In this role he was responsible for the capture and execution phases of multiple Intercontinental Ballistic Missile programs for Minuteman III, including a new airborne Nuclear Command and Control (NC2) development program. His major responsibilities included establishing program working environments at multiple locations, policies, processes, staffing, budget and technical baselines. Mr. Dare has extensive T&E and prototyping experience. As the Engineering Program Manager for the $1.8B Integrated Space C2 programs for NORAD/NORTHCOM systems at Cheyenne Mountain, Mr. Dare was the Integration and Test lead focusing on planning, executing, and evaluating the integration and test phases (developmental and operational T&E) for Missile Warning and Space Situational Awareness (SSA) systems. Mr. Dare has also been the Engineering Lead/Integration and Test lead on other systems such as the Hubble Space Telescope; international border control systems; artificial intelligence (AI) development systems (knowledge-based reasoning); Service-based networking systems for the UK Ministry of Defence; Army C2 systems; Space Fence C2; and foreign intelligence, surveillance, and reconnaissance systems. As part of the Department’s strategic defense portfolio, Mr. Dare led the development of advanced prototypes in SSA C2 (Space Fence), Information Assurance (Single Sign-on), AI systems, and was the sponsoring program manager for NC2 capability development. Mr. Dare is a graduate of Purdue University and is a member of both the Association for Computing Machinery and Program Management Institute. He has been recognized by the U.S. Air Force for his contributions supporting NORAD/NORTHCOM’s strategic defense missions, and the National Aeronautics and Space Administration for his contributions to the original Hubble Space Telescope program. Mr. Dare holds a U.S. Patent for Single Sign-on architectures.

Jared Freeman, Ph.D., is Chief Scientist of Aptima and Chair of the Human Systems Division of the National Defense Industry Association. His research and publications address measurement, assessment, and enhancement of human learning, cognition, and performance in technologically complex military environments.

This course will cover the basics of the Bayesian approach to practical and coherent statistical inference. Particular attention will be paid to computational aspects, including MCMC. Examples/practical hands-on exercises will the run gamut from toy illustration to real-world data analysis from all areas of science, with R implementations/coaching provided. The course closely follows P.D. Hoff’s “A First Course in Bayesian Statistical Methods”—Springer 2009. Some examples are borrowed from two other texts which are nice references to have. J. Albert’s’ “Bayesian Computation with R”— Springer 2nd ed. 2009; and “A. Gelman, J.B. Carlin, H.S. Stern, D. Dunson, A. Vehtari and D.B. Rubin’ s “Bayesian Data Analysis”—3rd ed. 2013.

Categorical data is abundant in the 21st century, and its analysis is vital to advance research across many domains. Thus, data-analytic techniques that are tailored for categorical data are an essential part of the practitioner’s toolset. The purpose of this short course is to help attendees develop and sharpen their abilities with these tools. Topics covered in this short course will include logistic regression, ordinal regression, and classification, and methods to assess predictive accuracy of these approaches will be discussed. Data will be analyzed using the R software package, and course content loosely follow Alan Agresti’s excellent textbook An Introduction to Categorical Data Analysis, Third Edition.

In this one-day workshop, we will explore five techniques that are commonly used to model human behavior: principal component analysis, factor analysis, cluster analysis, mixture modeling, and multidimensional scaling. Brief discussions of the theory of each method will be provided, along with some examples showing how the techniques work and how the results are interpreted in practice. Accompanying R-code will be provided so attendees are able to implement these methods on their own.

Statistical Methods for Modeling and Simulation Verification and Validation is a 1-day tutorial in applied statistical methods for the planning, designing and analysis of simulation experiments and live test events for the purposes of verifying and validating models and simulations. The course covers the fundamentals of verification and validation of models and simulations, as it is currently practiced and as is suggested for future applications. The first session is largely an introduction to modeling and simulation concepts, verification and validation policies, along with a basic introduction to data visualization and statistical methods. Session 2 covers the essentials of experiment design for simulations and live test events. The final session focuses on analysis techniques appropriate for designed experiments and tests, as well as observational data, for the express purpose of simulation validation. We look forward to your participation in this course.

We increasingly rely on mathematical and statistical models to predict phenomena ranging from nuclear power plant design to profits made in financial markets. When assessing the feasibility of these predictions, it is critical to quantify uncertainties associated with the models, inputs to the models, and data used to calibrate the models. The synthesis of statistical and mathematical techniques, which can be used to quantify input and response uncertainties for simulation codes that can take hours to days to run, comprises the evolving field of uncertainty quantification. The use of data, to improve the predictive accuracy of models, is central to uncertainty quantification so we will begin by providing an overview of how Bayesian techniques can be used to construct distributions for model inputs. We will subsequently describe the computational issues associated with propagating these distributions through complex models to construct prediction intervals for statistical quantities of interest such as expected profits or maximal reactor temperatures. Finally, we will describe the use of sensitivity analysis to isolate critical model inputs and surrogate model construction for simulation codes that are too complex for direct statistical analysis. All topics will be motivated by examples arising in engineering, biology, and economics.

Overview/Course Outcomes- Well-designed experiments are a powerful tool for developing and validating cause and effect relationships when evaluating and improving product and process performance and for operational testing of complex systems. Designed experiments are the only efficient way to verify the impact of changes in product or process factors on actual performance. The course outcomes are: • Ability to plan and execute experiments • Ability to collect data and analyze and interpret these data to provide the knowledge required for business success • Knowledge of a wide range of modern experimental tools that enable practitioners to customize their experiment to meet practical resource constraints The topics covered during the course are: • Fundamentals of DOX – randomization, replication, and blocking. • Planning for a designed experiment – type and size of design, factor selection, levels and ranges, response measurement, sample sizes. • Graphical and statistical approaches to DOX analysis. • Blocking to eliminate the impact of nuisance factors on experimental results. • Factorial experiments and interactions. • Fractional factorials – efficient and effective use of experimental resources. • Optimal designs • Response surface methods • A demonstration illustrating and comparing the effectiveness of different experimental design strategies. This course is focused on helping you and your organization make the most effective utilization of DOX. Software usage is fully integrated into the course Who Should Attend- The course is suitable for participants from an engineering or technical background. Participants will need some previous experience and background in statistical methods. Reference Materials- The course is based on the textbook Design and Analysis of Experiments, 9th Edition, by Douglas C. Montgomery. JMP Software will be discussed and illustrated.

Combinatorial methods have attracted attention as a means of providing strong assurance at reduced cost, but when are these methods practical and cost-effective? This tutorial includes two sections on the basis and application of combinatorial test methods: The first section explains the background, process, and tools available for combinatorial testing, with illustrations from industry experience with the method. The focus is on practical applications, including an industrial example of testing to meet FAA-required standards for life-critical software for commercial aviation. Other example applications include modeling and simulation, mobile devices, network configuration, and testing for a NASA spacecraft. The discussion will also include examples of measured resource and cost reduction in case studies from a variety of application domains. The second part explains combinatorial testing-based techniques for effective security testing of software components and large-scale software systems. It will develop quality assurance and effective re-verification for security testing of web applications and testing of operating systems. It will further address how combinatorial testing can be applied to ensure proper error-handling of network security protocols and provide the theoretical guarantees for detecting Trojans injected in cryptographic hardware. Procedures and techniques, as well as workarounds will be presented and captured as guidelines for a broader audience.

This tutorial is an abbreviated version of a 36-hour short course recently provided by UVA to a class composed of engineers working at the Defense Intelligence Agency. The tutorial provides a definition for cyber attack resilience that is an extension of earlier definitions of system resilience that were not focused on cyber attacks. Based upon research results derived by the University of Virginia over an eight year period through DoD/Army/AF/Industry funding , the tutorial will illuminate the following topics: 1) A Resilence Design Requirements methodology and the need for supporting analysis tools, 2) a System Architecture approach for achieving resilience, 3) Example resilience design patterns and example prototype implementations, 4) Experimental results regarding resilience-related roles and readiness of system operators, and 5) Test and Evaluation Issues. The tutorial will be presented by UVA Munster Professor Barry Horowitz.

In recent years, the programming language Python with its supporting ecosystem has established itself as a significant capability to support the activities of the typical data scientist. Recently, version 1.0 of the programming language Julia has been released; from a software engineering perspective, it can be viewed as a modern alternative. This tutorial presents both Python and Julia from both a user and developer point of view. From a user’s point of view, the basic syntax of each, along with fundamental prerequisite knowledge presented. From a developers point of view the underlying infrastructure of the programming language / interpreter / compiler is discussed.

In the test community, we frequently use statistics to extract meaning from data. These inferences may be drawn with respect to topics ranging from system performance to human factors. In this mini-tutorial, we will begin by discussing the use of descriptive and inferential statistics, before exploring the basics of interval estimation and hypothesis testing. We will introduce common statistical techniques and when to apply them, and conclude with a brief discussion of how to present your statistical findings graphically for maximum impact.

Analyses are “reproducible” if the same methods applied to the same data produce identical results when run again by another researcher (or you in the future). Reproducible analyses are transparent and easy for reviewers to verify, as results and figures can be traced directly to the data and methods that produced them. There are also direct benefits to the researcher. Real-world analysis workflows inevitably require changes to incorporate new or additional data, or to address feedback from collaborators, reviewers, or sponsors. These changes are easier to make when reproducible research best practices have been considered from the start. Poor reproducibility habits result in analyses that are difficult or impossible to review, are prone to compounded mistakes, and are inefficient to re-run in the future. They can lead to duplication of effort or even loss of accumulated knowledge when a researcher leaves your organization. With larger and more complex datasets, along with more complex analysis techniques, reproducibility is more important than ever. Although reproducibility is critical, it is often not prioritized either due to a lack of time or an incomplete understanding of end-to-end opportunities to improve reproducibility. This tutorial will discuss the benefits of reproducible research and will demonstrate ways that analysts can introduce reproducible research practices during each phase of the analysis workflow: preparing for an analysis, performing the analysis, and presenting results. A motivating example will be carried throughout to demonstrate specific techniques, useful tools, and other tips and tricks where appropriate. The discussion of specific techniques and tools is non-exhaustive; we focus on things that are accessible and immediately useful for someone new to reproducible research. The methods will focus mainly on work performed using R, but the general concepts underlying reproducible research techniques can be implemented in other analysis environments, such as JMP and Excel, and are briefly discussed. By implementing the approaches and concepts discussed during this tutorial, analysts in defense and aerospace will be equipped to produce more credible and defensible analyses of T&E data.

The DoD uses psychological measurement to aid in decision-making about a variety of issues including the mental health of military personnel before and after combat, and the quality of human-systems interactions. To develop quality survey instruments (scales) and interpret the data obtained from these instruments appropriately, analysts and decision-makers must understand the factors that affect the reliability and validity of psychological measurement. This tutorial covers the basics of scale development and validation and discusses current efforts by IDA, DOT&E, ATEC, and JITC to develop validated scales for use in operational test and evaluation.