Open Access. Powered by Scholars. Published by Universities.®

Data Science Commons

Open Access. Powered by Scholars. Published by Universities.®

Air Force Institute of Technology

Discipline
Keyword
Publication Year
Publication
Publication Type

Articles 31 - 57 of 57

Full-Text Articles in Data Science

Telemetry Data Mining For Unmanned Aircraft Systems, Li Yu Mar 2022

Telemetry Data Mining For Unmanned Aircraft Systems, Li Yu

Theses and Dissertations

With ever more data becoming available to the US Air Force, it is vital to develop effective methods to leverage this strategic asset. Machine learning (ML) techniques present a means of meeting this challenge, as these tools have demonstrated successful use in commercial applications. For this research, three ML methods were applied to a unmanned aircraft system (UAS) telemetry dataset with the aim of extracting useful insight related to phases of flight. It was shown that ML provides an advantage in exploratory data analysis and as well as classification of phases. Neural network models demonstrated the best performance with over …


Leveraging Machine Learning For Large Scale Analysis Of Publicly-Available Data For Gnss Interference Events, David K. Stamper Mar 2022

Leveraging Machine Learning For Large Scale Analysis Of Publicly-Available Data For Gnss Interference Events, David K. Stamper

Theses and Dissertations

This research documents architecture and implementation of an enhanced interference detection and classification analysis system, using both a database and storage solution utilizing machine learning algorithms to detect changes in Carrier-to-Noise strength over multiple GNSS sites. The system uses publicly-available government supported receivers to detect interference, and built using FOSS packaged as a programming library through Python. Two algorithms are discussed in terms of enhancing interference detection using both non-machine learning and machine learning approaches. Two algorithms are also discussed which are used for classification of events. In addition, an approach to Large Scale data analytics is demonstrated via a …


Generalized Robust Feature Selection, Bradford L. Lott Mar 2022

Generalized Robust Feature Selection, Bradford L. Lott

Theses and Dissertations

Feature selection may be summarized as identifying salient features to a given response. Understanding which features affect the response enables, in the future, only collecting consequential data; hence, the feature selection algorithm may lead to saving effort spent collecting data, storage resources, as well as computational resources for making predictions. We propose a generalized approach to select the salient features of data sets. Our approach may also be applied to unsupervised datasets to understand which data streams provide unique information. We contend our approach identifies salient features robust to the sub-sequent predictive model applied. The proposed algorithm considers all provided …


Constructing Prediction Intervals With Neural Networks: An Empirical Evaluation Of Bootstrapping And Conformal Inference Methods, Alexander N. Contarino Mar 2022

Constructing Prediction Intervals With Neural Networks: An Empirical Evaluation Of Bootstrapping And Conformal Inference Methods, Alexander N. Contarino

Theses and Dissertations

Artificial neural networks (ANNs) are popular tools for accomplishing many machine learning tasks, including predicting continuous outcomes. However, the general lack of confidence measures provided with ANN predictions limit their applicability, especially in military settings where accuracy is paramount. Supplementing point predictions with prediction intervals (PIs) is common for other learning algorithms, but the complex structure and training of ANNs renders constructing PIs difficult. This work provides the network design choices and inferential methods for creating better performing PIs with ANNs to enable their adaptation for military use. A two-step experiment is executed across 11 datasets, including an imaged-based dataset. …


Development Of Advanced Machine Learning Models For Analysis Of Plutonium Surrogate Optical Emission Spectra, Ashwin P. Rao, Phillip R. Jenkins, John D. Auxier Ii, Michael B. Shattan, Anil K. Patnaik Jan 2022

Development Of Advanced Machine Learning Models For Analysis Of Plutonium Surrogate Optical Emission Spectra, Ashwin P. Rao, Phillip R. Jenkins, John D. Auxier Ii, Michael B. Shattan, Anil K. Patnaik

Faculty Publications

This work investigates and applies machine learning paradigms seldom seen in analytical spectroscopy for quantification of gallium in cerium matrices via processing of laser-plasma spectra. Ensemble regressions, support vector machine regressions, Gaussian kernel regressions, and artificial neural network techniques are trained and tested on cerium-gallium pellet spectra. A thorough hyperparameter optimization experiment is conducted initially to determine the best design features for each model. The optimized models are evaluated for sensitivity and precision using the limit of detection (LoD) and root mean-squared error of prediction (RMSEP) metrics, respectively. Gaussian kernel regression yields the superlative predictive model with an RMSEP of …


Per-Pixel Cloud Cover Classification Of Multispectral Landsat-8 Data, Salome E. Carrasco, Torrey J. Wagner, Brent T. Langhals Jun 2021

Per-Pixel Cloud Cover Classification Of Multispectral Landsat-8 Data, Salome E. Carrasco, Torrey J. Wagner, Brent T. Langhals

Faculty Publications

Random forest and neural network algorithms are applied to identify cloud cover using 10 of the wavelength bands available in Landsat 8 imagery. The methods classify each pixel into 4 different classes: clear, cloud shadow, light cloud, or cloud. The first method is based on a fully connected neural network with ten input neurons, two hidden layers of 8 and 10 neurons respectively, and a single-neuron output for each class. This type of model is considered with and without L2 regularization applied to the kernel weighting. The final model type is a random forest classifier created from an ensemble of …


The Effects Of Individual Differences, Non‐Stationarity, And The Importance Of Data Partitioning Decisions For Training And Testing Of Eeg Cross‐Participant Models, Alexander J. Kamrud, Brett J. Borghetti, Christine M. Schubert Kabban May 2021

The Effects Of Individual Differences, Non‐Stationarity, And The Importance Of Data Partitioning Decisions For Training And Testing Of Eeg Cross‐Participant Models, Alexander J. Kamrud, Brett J. Borghetti, Christine M. Schubert Kabban

Faculty Publications

EEG-based deep learning models have trended toward models that are designed to perform classification on any individual (cross-participant models). However, because EEG varies across participants due to non-stationarity and individual differences, certain guidelines must be followed for partitioning data into training, validation, and testing sets, in order for cross-participant models to avoid overestimation of model accuracy. Despite this necessity, the majority of EEG-based cross-participant models have not adopted such guidelines. Furthermore, some data repositories may unwittingly contribute to the problem by providing partitioned test and non-test datasets for reasons such as competition support. In this study, we demonstrate how improper …


Contract Information Extraction Using Machine Learning, Zachary E. Butcher Mar 2021

Contract Information Extraction Using Machine Learning, Zachary E. Butcher

Theses and Dissertations

The Air Force Sustainment Center assisted by the Data Analytics Resource Team and the Defense Logistics Agency collected four million contracts onto one of the Air Force Research Laboratory’s high power computers. This thesis focuses on the effort to determine if parts are available through those contracts. Some information is extracted using machine learning in combination with natural language processing. Where machine learning methods are unsuccessful or inappropriate, text mining techniques, such as pattern recognition and rules, are used. Upon completion, the information is combined into a Gantt chart for quick evaluation. Only 21% of the contracts have their information …


Multi-Objective Database Queries In Combined Knapsack And Set Covering Problem Domains, Sean A. Mochocki, Gary B. Lamont, Robert C. Leishman, Kyle J. Kauffman Mar 2021

Multi-Objective Database Queries In Combined Knapsack And Set Covering Problem Domains, Sean A. Mochocki, Gary B. Lamont, Robert C. Leishman, Kyle J. Kauffman

Faculty Publications

Database queries are one of the most important functions of a relational database. Users are interested in viewing a variety of data representations, and this may vary based on database purpose and the nature of the stored data. The Air Force Institute of Technology has approximately 100 data logs which will be converted to the standardized Scorpion Data Model format. A relational database is designed to house this data and its associated sensor and non-sensor metadata. Deterministic polynomial-time queries were used to test the performance of this schema against two other schemas, with databases of 100 and 1000 logs of …


Algorithm Selection Framework: A Holistic Approach To The Algorithm Selection Problem, Marc W. Chalé Mar 2020

Algorithm Selection Framework: A Holistic Approach To The Algorithm Selection Problem, Marc W. Chalé

Theses and Dissertations

A holistic approach to the algorithm selection problem is presented. The “algorithm selection framework" uses a combination of user input and meta-data to streamline the algorithm selection for any data analysis task. The framework removes the conjecture of the common trial and error strategy and generates a preference ranked list of recommended analysis techniques. The framework is performed on nine analysis problems. Each of the recommended analysis techniques are implemented on the corresponding data sets. Algorithm performance is assessed using the primary metric of recall and the secondary metric of run time. In six of the problems, the recall of …


An Analysis Of Learning Curve Theory & Diminishing Rates Of Learning, Dakotah W. Hogan Mar 2020

An Analysis Of Learning Curve Theory & Diminishing Rates Of Learning, Dakotah W. Hogan

Theses and Dissertations

Traditional learning curve theory assumes a constant learning rate regardless of the number of units produced; however, a collection of theoretical and empirical evidence indicates that learning rates decrease as more units are produced in some cases. These diminishing learning rates cause traditional learning curves to underestimate required resources, potentially resulting in cost overruns. A diminishing learning rate model, Boones Learning Curve (2018), was recently developed to model this phenomenon. This research confirmed that Boones Learning Curve is more accurate in modeling observed learning curves using production data of 169 Department of Defense end-items. However, further empirical analysis revealed deficiencies …


Photovoltaic System Optimization For An Austere Location Using Time Series Data, Torrey J. Wagner, Eric Lang, Warren Assink, Douglas S. Dudis Jun 2018

Photovoltaic System Optimization For An Austere Location Using Time Series Data, Torrey J. Wagner, Eric Lang, Warren Assink, Douglas S. Dudis

Faculty Publications

In this work we test experimental photovoltaic, storage and generator technologies and investigate their potential to meet austere location energy needs. After defining the energy requirements and insolation of a 1,100-person base, we develop a microgrid model and simulation. Cost optimizations were then performed using hourly time-series data to explore the cost and performance trade-space of a PV-battery-generator system. The work highlights the cost of resiliency and the dependencies of optimum system component sizes on duration and the fully burdened cost of fuel.


Effects Of Data Replication On Data Exfiltration In Mobile Ad Hoc Networks Utilizing Reactive Protocols, Corey T. Willinger Mar 2015

Effects Of Data Replication On Data Exfiltration In Mobile Ad Hoc Networks Utilizing Reactive Protocols, Corey T. Willinger

Theses and Dissertations

A swarm of autonomous UAVs can provide a significant amount of ISR data where current UAV assets may not be feasible or practical. As such, the availability of the data the resides in the swarm is a topic that will benefit from further investigation. This thesis examines the impact of le replication and swarm characteristics such as node mobility, swarm size, and churn rate on data availability utilizing reactive protocols. This document examines the most prominent factors affecting the networking of nodes in a MANET. Factors include network routing protocols and peer-to-peer le protocols. It compares and contrasts several open …


Modeling Cyber Situational Awareness Through Data Fusion, Evan L. Raulerson Mar 2013

Modeling Cyber Situational Awareness Through Data Fusion, Evan L. Raulerson

Theses and Dissertations

Cyber attacks are compromising networks faster than administrators can respond. Network defenders are unable to become oriented with these attacks, determine the potential impacts, and assess the damages in a timely manner. Since the observations of network sensors are normally disjointed, analysis of the data is overwhelming and time is not spent efficiently. Automation in defending cyber networks requires a level of reasoning for adequate response. Current automated systems are mostly limited to scripted responses. Better defense tools are required. This research develops a framework that aggregates data from heterogeneous network sensors. The collected data is correlated into a single …


Methods And System For Equalizing Data, Jaiganesh Balakrishnan, Richard K. Martin, C. Richard Johnson Jr. Dec 2007

Methods And System For Equalizing Data, Jaiganesh Balakrishnan, Richard K. Martin, C. Richard Johnson Jr.

AFIT Patents

A method for equalizing data and systems utilizing the method. The method of this invention for equalizing (by shortening the channel response) data includes minimizing a function of the data and a number of equalizer characteristic parameters, where the function utilizes auto-correlation data corresponding to equalized data. Updated equalizer characteristic parameters are then obtained from the minimization and an initial set of equalizer characteristic parameters. Finally, the received data is processed utilizing the equalizer defined by the minimization. The method of this invention can be implemented in an equalizer and the equalizer of this invention may be included in a …


Surveying Cost Growth, Michael A. Greiner, Vince Sipple, Edward D. White Jan 2004

Surveying Cost Growth, Michael A. Greiner, Vince Sipple, Edward D. White

Faculty Publications

Cost growth that weapon systems incur throughout their acquisition life cycle concerns those who work in the acquisition environment. One way to reduce the amount of unexpected cost growth is to develop better cost estimates. In attaining better cost estimates though, it is often helpful to understand and account for potential cost drivers. Several cost studies, some of which specifically focus on the aircraft industry, have been performed documenting and investigating these growth factors. Overviews of these various cost growth studies are presented as other tools for the cost estimators and program managers.


Autoassociative-Heteroassociative Neural Network, Claudia V. Kropas-Hughes, Steven K. Rogers, Mark E. Oxley, Matthew Kabrisky Jun 2002

Autoassociative-Heteroassociative Neural Network, Claudia V. Kropas-Hughes, Steven K. Rogers, Mark E. Oxley, Matthew Kabrisky

AFIT Patents

An efficient neural network computing technique capable of synthesizing two sets of output signal data from a single input signal data set. The method and device of the invention involves a unique integration of autoassociative and heteroassociative neural network mappings, the autoassociative neural network mapping enabling a quality metric for assessing the generalization or prediction accuracy of the heteroassociative neural network mapping.


The Affect Of Varying Arousal Methods Upon Vigilance And Error Detection In An Automated Command And Control Environment, Brent T. Langhals Mar 2001

The Affect Of Varying Arousal Methods Upon Vigilance And Error Detection In An Automated Command And Control Environment, Brent T. Langhals

Theses and Dissertations

This study focused on improving vigilance performance through developing methods to arouse subjects to the possibility of errors in a data manipulation information warfare attack. The study suggests that by continuously applying arousal stimuli, subjects would retain initially high vigilance levels thereby avoiding the vigilance decrement phenomenon and improving error detection. The research focused on which methods were the most effective as well the impact of age upon the arousability of the subjects. Further the implications of vigilance and vigilance decrement for correct detections as well as productivity were explored. The study used a simulation experiment to provide a vigilance …


Extracting Mission Semantics From Unmanned Aerial Vehicle Telemetry And Flight Plans, Walter T. Berridge Mar 2000

Extracting Mission Semantics From Unmanned Aerial Vehicle Telemetry And Flight Plans, Walter T. Berridge

Theses and Dissertations

With the acceptance of Unmanned Aerial Vehicles (UAVs) as a primary platform within the Department of Defense (DOD) for gathering intelligence data, the amount of video information being recorded, analyzed, and archived continues to grow. Mechanisms for quickly locating and retrieving video segments of interest amongst the many hours of recorded video are required to accommodate the rapid turnaround expected in today's wartime planning environments. This research demonstrates that text-based data accompanying UAV video yields sufficient information to identify and create data items that can be indexed to provide for rapid identification and retrieval of video segments of interest. Four …


Evolving Compact Decision Rule Sets, Robert E. Marmelstein Jun 1999

Evolving Compact Decision Rule Sets, Robert E. Marmelstein

Theses and Dissertations

While data mining technology holds the promise of automatically extracting useful patterns (such as decision rules) from data, this potential has yet to be realized. One of the major technical impediments is that the current generation of data mining tools produce decision rule sets that are very accurate, but extremely complex and difficult to interpret. As a result, there is a clear need for methods that yield decision rule sets that are both accurate and compact. The development of the Genetic Rule and Classifier Construction Environment (GRaCCE) is proposed as an alternative to existing decision rule induction (DRI) algorithms. GRaCCE …


An Examination Of Multi-Tier Designs For Legacy Data Access, Michael L. Acker Dec 1997

An Examination Of Multi-Tier Designs For Legacy Data Access, Michael L. Acker

Theses and Dissertations

This work examines the application of Java and the Common Object Request Broker Architecture (CORBA) to support access to remote databases via the Internet. The research applies these software technologies to assist an Air Force distance learning provider in improving the capabilities of its World Wide Web-based correspondence system. An analysis of the distance learning provider's operation revealed a strong dependency on a non-collocated legacy relational database. This dependency limits the distance learning provider's future web-based capabilities. A recommendation to improve operation by data replication is proposed, and the implementation details are provided for two alternative test systems that support …


A Comparison Of Loose And Tight Gps/Ins Integration Using Real Ins And Gps Data, Warren H. Nuibe Dec 1995

A Comparison Of Loose And Tight Gps/Ins Integration Using Real Ins And Gps Data, Warren H. Nuibe

Theses and Dissertations

An extended Kalman filter (EKE) is used to combine the information obtained from a Global Positioning System (GPS) receiver and an Inertial Navigation System (INS) to provide a navigation solution. This research compares the results of a tightly-coupled GPS/INS integrated system with a loosely-coupled integrated system, using real world data. A fair comparison is accomplished by using the same sets of data, and keeping the integration structures as close as possible. Both integrations are feedforward and have the same error states in the navigation Kalman filters. Differences between the two, such as navigation solutions and tuning values, are shown in …


A Neural Network Approach To The Prediction And Confidence Assignation Of Nonlinear Time Series Classifications, Erin S. Heim Dec 1995

A Neural Network Approach To The Prediction And Confidence Assignation Of Nonlinear Time Series Classifications, Erin S. Heim

Theses and Dissertations

This thesis uses multiple layer perceptrons (MLP) neural networks and Kohonen clustering networks to predict and assign confidence to nonlinear time series classifications. The nonlinear time series used for analysis is the Standard and Poor's 100 (S&P 100) index. The target prediction is classification of the daily index change. Financial indicators were evaluated to determine the most useful combination of features for input into the networks. After evaluation it was determined that net changes in the index over time and three short-term indicators result in better accuracy. A back-propagation trained MLP neural network was then trained with these features to …


Acquiring Consistent Knowledge For Bayesian Forests, Darwyn O. Banks Mar 1995

Acquiring Consistent Knowledge For Bayesian Forests, Darwyn O. Banks

Theses and Dissertations

This thesis develops a methodology and a tool for knowledge acquisition with the new probabilistic knowledge representation-the Bayesian Forest. It establishes the structure of the Knowledge Acquisition and Maintenance module of the Probabilities. Expert Systems, Knowledge and Inference (PESKI) architecture. The tool, MACK, is designed to be used directly by the domain expert(s) rather than by knowledge engineer(s), and thus supports automated knowledge acquisition. This research determines and implements the constraints necessary to ensure the consistency of Bayesian Forest knowledge bases as data is both acquired and subsequently maintained. The impact to the PESKI architecture of time-dependent information and default …


The Application Of A Readiness-Based Sparing Model To Foreign Military Sales, Karen M. Klinger Jun 1994

The Application Of A Readiness-Based Sparing Model To Foreign Military Sales, Karen M. Klinger

Theses and Dissertations

Current Foreign Military Sales FMS models provide stock levels that result in a very low system availability or a funding requirement that exceeds the overall budget. The purpose of this research was to determine if an inventory model exists that can be used in FMS reparable sparing to provide a more efficient and economical inventory purchase. The Aircraft Sustainability Model ASM is such a model, providing the most aircraft availability possible from a given inventory investment by computing the optimal number of spare parts to buy for each item. FMS data was obtained from two sources - the International Data …


Data Reduction With Least Squares Differential Correction Using Equinoctial Elements, Michael S. Wasson Dec 1992

Data Reduction With Least Squares Differential Correction Using Equinoctial Elements, Michael S. Wasson

Theses and Dissertations

This study investigates earth satellite orbit estimation on a track of range, azimuth, and elevation data from a single tracking station. The estimation routine is a least squares batch filter based solely on two-body orbital motion. Using equinoctial elements for the reference orbit avoids the numerical difficulties of the classical elements at eccentricities near zero and inclinations near zero or 90 degrees. Orbits for Mir, DMSP, Explorer, Cosmos, and GPS are investigated. The goal of this study is to reduce orbit information from observations (range, azimuth, and elevation) to an element set and a covariance matrix without considering perturbation effects. …


The Application Of Kriging For Controlled Minimization Of Large Data Sets, Christopher Brodkin Dec 1991

The Application Of Kriging For Controlled Minimization Of Large Data Sets, Christopher Brodkin

Theses and Dissertations

Frequently, the quantity of data available is much greater than that which can be manipulated in an efficient and timely manner. This can cause several problems. The first, and probably most critical, problem is the excessive on-line storage needs of these huge data sets. Secondly, in the computer animation field, huge data sets may require excessive computational time for generation of each frame of a computer animation. Thirdly, computer screens have a limited resolution and need too much computational time removing excessive detail from images generated with a higher resolution than can be displayed. Lastly, too much time is required …