11115 research outputs found
Sort by
Reduced Order Non-INtrusive Modeling Methodology Formulation and Application for Mission Analysis
No full textThe Department of Defense (DOD) has documented a strategic gap how exploratory analyses are accomplished to support capability development which was decomposed into areas of needed focused research. We began with an exploration of current methods for integrating different models to meet the concerns of Congress with regard to quality, accuracy, and dependability, noting that they have become too computationally prohibitive for exploring large trade spaces. In addition, current model abstraction methods have difficulty accounting for the increasing dimensionality associated with increasingly complex simulations. These observations led to the formulation of the Reduced Order Non-INtrusive (RONIN) modeling methodology, which generates predictive reduced order surrogate models, which capture more information regarding behaviors as compared to traditional methods. The RONIN modeling methodology works to create surrogate models which emulate stochastic full-order models (FOMs) by leveraging order reduction approaches, stochastic modeling methods, and regression techniques. To demonstrate the RONIN modeling methodology, a notional United States Air Force use case was defined, and a DOD standard simulation framework was used to create relevant simulation scenario which output a set of response distributions. Ultimately, the RONIN modeling method was used to create a predictive surrogate model which was able to reconstruct output distributions which are statically consistent with the original FOM on average over 99% of the time while reducing the time needed to generate a distribution of outputs from minutes down to less than a second
Global Sporadic-E Prediction and Climatology Using Deep Learning
Full text linkSporadic-E (Es) is an ionospheric phenomenon defined by strong layers of plasma which may interfere with radio wave propagation. In this work, we develop deep learning models to improve the understanding of Es, including the presence, intensity and height of the layers. We developed three separate models. The first, building off earlier work in (J. A. Ellis et al., 2024, link in AFIT Scholar, 10.1029/2023sw003669), includes only the main features from radio occultation (RO) measurements. The second adds to that time, date, location, geomagnetic and solar indices, solar winds, x-ray flux, weather and lightning. A third model excludes RO measurements but includes the rest. In training the first two models, the Es ordinary mode critical frequency (foEs), a measure of intensity, and height (hEs) parameters extracted from ionosondes were used as the “ground truth” target variables. In training the third model, estimated foEs and hEs values from the RO model were added as target variables to augment the data set and produce physically reasonable model predictions globally. We find that the second model performs well with intensity prediction tasks, but struggles with height estimations, which is likely due to the tangent point assumption made during RO signal processing and errors inherent in the ionosonde extracted virtual heights. The third model performed reasonably well considering the lack of in situ RO measurement. The third model performs the best on height predictions, which points to the height being very climatologically driven, whereas the intensity is a more complex interaction of several variables
Analyzing Stability of Estimates at Completion for Long Duration Development Efforts
Full text linkDefense program managers utilize Earned Value Management (EVM) methodologies to measure, report, and predict the cost and schedule performance of their programs. Previous research conducted by Christensen (1996) and Kim et al. (2019) has shown varied results in the stability of EVM Estimates at Completion (EACs). Stability is defined as a 10% or less deviation from the final EAC at a specified percent completion point of the program. The Christensen (1996) and Kim et al. (2019) studies also noted that program-specific factors, such as phase, can impact the accuracy of EVM metrics. This study builds upon those works by assessing EAC stability specifically for defense programs characterized by one such factor: long duration development efforts. In addition, EAC stability using Earned Schedule (ES) metrics are also assessed and statistically compared to the EVM results. The authors found stability at around 70% completion. This result is later in the project than previous research wherein the researchers did not consider project duration in their findings. Additionally, no significant differences in EAC stability were found with ES metrics in comparison to the EVM results. Therefore, program managers are cautioned to consider the length of duration when estimating cost at completion up to and slightly beyond the 70% completion point
A Survey of Sampling Methods for Hyperspectral Remote Sensing: Addressing Bias Induced by Random Sampling
Full text linkIdentified as early as 2000, the challenges involved in developing and assessing remote sensing models with small datasets remain, with one key issue persisting: the misuse of random sampling to generate training and testing data. This practice often introduces a high degree of correlation between the sets, leading to an overestimation of model generalizability. Despite the early recognition of this problem, few researchers have investigated its nuances or developed effective sampling techniques to address it. Our survey highlights that mitigation strategies to reduce this bias remain underutilized in practice, distorting the interpretation and comparison of results across the field. In this work, we introduce a set of desirable characteristics to evaluate sampling algorithms, with a primary focus on their tendency to induce correlation between training and test data, while also accounting for other relevant factors. Using these characteristics, we survey 146 articles, identify 16 unique sampling algorithms, and evaluate them. Our evaluation reveals two broad archetypes of sampling techniques that effectively mitigate correlation and are suitable for model development
Sparse Regularization Effects on Radar ATR
No full textSynthetic aperture radar images of civilian vehicles are shown to have reduced correlation in images formed using basis pursuit denoising compared to those formed using polar format algorithm. Sidelobe reduction via windowing has a negative effect on target recognition due to the inherent mainlobe spreading. Basis pursuit image does not suffer the same mainlobe spreading while reducing sidelobes, though benefits come at the cost of computations
Exploring the Translation Lookaside Buffer (TLB) for Low-Level Task Differentiation and Classification
Full text linkThe primary focus of modern Central Processing Unit (CPU) technologies is performance improvement, with security often considered a secondary concern. As a result, vulnerabilities within the system are overlooked. While significant research, both offensive and defensive, has been conducted on CPU caches, relatively little attention has been given to the Translation Lookaside Buffer (TLB) due to its perceived lack of data granularity. Prior studies have typically combined multiple Hardware Performance Counters (HPCs) or relied on timing analysis to extract meaningful insights. In contrast, this study introduces a novel methodology that leverages only TLB related HPCs for multi-task classification, without incorporating data from other microarchitectural components such as caches or branch predictors, and without relying on timing memory accesses. In the proposed methodology, an attacker uses HPCs to collect TLB based data while a victim executes tasks. We demonstrate that statistical learning models, including Random Forest (RF) and Logistic Regression (LR), achieve classification accuracy of up to 87%, surpassing the next best TLB only method (based on timing analysis) by 11%. Furthermore, neural networks, such as Artificial Neural Network (ANN) and Convolution Neural Network (CNN), achieve 88% accuracy, improving prior TLB based approaches by 12%. These findings demonstrate the potential of TLB based methodologies for task classification, victim monitoring, and future security enhancements in microarchitectural design
Factor-Graph Optimization for Robust Navigation via High-Precision GNSS/IMU Corroboration
No full textExcerpt: This paper proposes an approach for resilient navigation through a factor-graph formulation that incorporates high-quality IMU measurements into a robust GNSS factor-graph formulation. The approach effectively enables the corroboration of GNSS measurements based on their consistency with the precise relative-motion information available from a high-quality IMU
Stabilized and unstabilized sampling methods result in differential fecal 16S rRNA microbial sequencing results
Full text linkOver the past decade, studies have been conducted to increase the understanding of associations between the fecal microbiome and human health. In conjunction, researchers have investigated the effects of study design, methods, molecular processing, and sequencing techniques. However, a lack of standardization of fecal sample collection methodology has introduced heterogeneity in sequencing results. Sources of variability include sample collection methods, storage temperatures, and transport times. Here we present 16S rRNA gene amplicon sequencing results from two sample collection methods (unstabilized sterile swab and stabilized OmniGene Gut Kits) collected from the same fecal specimens. The paired samples were collected either at the research facility or the participants’ home and ground shipped to the research facility at ambient temperature. Therefore, samples were exposed to variable temperatures and transport times. We found that fecal sample collection methods resulted in taxonomic and diversity differences that showed distinct patterns between swab and OmniGene samples. Swab samples were disproportionally affected by increased transport time, but differences in taxa and diversity were driven more by sample collection method, as compared to transport time. Based on previous studies, many of the taxa that were associated with sample collection methods and transport times have clinical relevance. Collectively, this research highlights: 1) the need for further standardization of methods for fecal microbiome studies; 2) limitations of direct comparisons between different fecal sample collection methods; and 3) the importance of careful consideration of sample collection methods for future studies and meta-analyses
Adaptive Selection of Decomposed Function Information Sources for Rapid Neural Networks
No full textA new approach is proposed of adaptively selecting emulators for emulator embedded neural networks. Emulators typically take the form of physics-based low-fidelity models. If querying a low-fidelity model incurs significant computational costs, an emulator can still be constructed by training a surrogate model, such as a Gaussian process model or neural network, on data gathered from the low-fidelity model. These emulators are embedded into the neural network architecture and play an important role in boosting neural network accuracy with limited training data. However, in some practical situations finding a suitable low-fidelity emulator model is challenging. Use of an improper emulator can fail to improve learning performance, and in some cases no viable emulator is available. To address this technical gap, this study proposes using the decomposed functions of the actual physics-based model as emulators. Global sensitivity analysis is performed to compute the Sobol indices of the decomposed functions, which reveal their contribution ranks to the system response of interest. Based on the contribution ranks, the decomposed functions are embedded as emulators in an adaptive and iterative process. The proposed method is demonstrated with fundamental analytical examples. Additionally, a representative hypersonic vehicle design problem is included as a practical engineering example
