Open Access. Powered by Scholars. Published by Universities.®
- Keyword
-
- Design of experiments (2)
- AFSIM (1)
- Artificial intelligence (1)
- Course of action (1)
- Functional data analysis (1)
-
- Functional linear model (1)
- Image classification (1)
- Imputation (1)
- Linear regression (1)
- Machine learning (1)
- Multivariate data analysis (1)
- Neural networks (1)
- Pitot probe (1)
- SARL (1)
- Semiparametric model (1)
- Simulation (1)
- Spatiotemporal kriging (1)
- Split-plot design (1)
- Statistics (1)
- Wargaming (1)
- Wavelets (1)
- Wind tunnels (1)
Articles 1 - 5 of 5
Full-Text Articles in Entire DC Network
Characterizing Convolutional Neural Network Early-Learning And Accelerating Non-Adaptive, First-Order Methods With Localized Lagrangian Restricted Memory Level Bundling, Benjamin O. Morris
Characterizing Convolutional Neural Network Early-Learning And Accelerating Non-Adaptive, First-Order Methods With Localized Lagrangian Restricted Memory Level Bundling, Benjamin O. Morris
Theses and Dissertations
This dissertation studies the underlying optimization problem encountered during the early-learning stages of convolutional neural networks and introduces a training algorithm competitive with existing state-of-the-art methods. First, a Design of Experiments method is introduced to systematically measure empirical second-order Lipschitz upper bound and region size estimates for local regions of convolutional neural network loss surfaces experienced during the early-learning stages. This method demonstrates that architecture choices can significantly impact the local loss surfaces traversed during training. Next, a Design of Experiments method is used to study the effects convolutional neural network architecture hyperparameters have on different optimization routines' abilities to …
Wavelet Methods For Very-Short Term Forecasting Of Functional Time Series, Jared K. Nystrom
Wavelet Methods For Very-Short Term Forecasting Of Functional Time Series, Jared K. Nystrom
Theses and Dissertations
Space launch operations at Kennedy Space Center and Cape Canaveral Space Force Station (KSC/CCSFS) are complicated by unique requirements for near-real time determination of risk from lightning. Lightning forecast weather sensor networks produce data that are noisy, high volume, and high frequency time series for which traditional forecasting methods are often ill-suited. Current approaches result in significant residual uncertainties and consequentially may result in forecasting operational policies that are excessively conservative or inefficient. This work proposes a new methodology of wavelet-enabled semiparametric modeling to develop accurate and timely forecasts robust against chaotic functional data. Wavelets methods are first used to …
Statistically Defensible Wind Tunnel Models, Timothy A. Roche
Statistically Defensible Wind Tunnel Models, Timothy A. Roche
Theses and Dissertations
Wind tunnels are used to test scale-model air frames in order to collect aerodynamic data. The Subsonic Aerodynamic Research Laboratory (SARL) Wind Tunnel is a low speed wind tunnel located at Wright-Patterson Air Force Base. The SARL Wind Tunnel team approached AFIT for assistance in creating statistically defensible models for the conditions inside the wind tunnel. During a wind tunnel test, pressure sensors cannot be placed at the test model. Instead, pressure is measured by a pitot probe permanently mounted in the corner of the test chamber. The pressure at the model location is predicted from the measurements taken by …
Analysis Of Functional Responses In Experimental Design, Matthew E. Scherer
Analysis Of Functional Responses In Experimental Design, Matthew E. Scherer
Theses and Dissertations
The growth of sensor streamed data in recent years increases the demand for an analytical technique to properly address data measured continuously. The design and analysis of experiments (DOE) of U.S. Air Force assets are based off of sensor streamed data. Functional data analysis (FDA) is an approach of analyzing data existing over a continuum. This research aids in filling the intersection of FDA and DOE by examining a case study of an experimental design with a functional response in addition to insight on software capabilities in FDA. The case study considers a functional linear model of a whole-plot from …
The Wargaming Commodity Course Of Action Automated Analysis Method, William T. Deberry
The Wargaming Commodity Course Of Action Automated Analysis Method, William T. Deberry
Theses and Dissertations
This research presents the Wargaming Commodity Course of Action Automated Analysis Method (WCCAAM), a novel approach to assist wargame commanders in developing and analyzing courses of action (COAs) through semi-automation of the Military Decision Making Process (MDMP). MDMP is a seven-step iterative method that commanders and mission partners follow to build an operational course of action to achieve strategic objectives. MDMP requires time, resources, and coordination – all competing items the commander weighs to make the optimal decision. WCCAAM receives the MDMP's Mission Analysis phase as input, converts the wargame into a directed graph, processes a multi-commodity flow algorithm on …