Open Access. Powered by Scholars. Published by Universities.®
- Publication Type
Articles 1 - 18 of 18
Full-Text Articles in Entire DC Network
Effectiveness Of A Timing Side Channel For Deriving Neural Network Depth, Matthew P. Weeks
Effectiveness Of A Timing Side Channel For Deriving Neural Network Depth, Matthew P. Weeks
Theses and Dissertations
From facial recognition on cell phones to vehicle traffic modeling for city planning, integrating ML models can be an expensive investment in resources. Protecting that investment is difficult, as information about the model and how it was built can be leaked through multiple channels, such as timing and memory access. In this thesis, one method of extracting data through a timing side-channel is examined across multiple hardware and software configurations to determine its reliability for general use. While attempting to determine the layer count of a target model solely from its inference time, the research determined that it is not …
Emotion Classification Of Indonesian Tweets Using Bidirectional Lstm, Aaron K. Glenn, Phillip M. Lacasse, Bruce A. Cox
Emotion Classification Of Indonesian Tweets Using Bidirectional Lstm, Aaron K. Glenn, Phillip M. Lacasse, Bruce A. Cox
Faculty Publications
Emotion classification can be a powerful tool to derive narratives from social media data. Traditional machine learning models that perform emotion classification on Indonesian Twitter data exist but rely on closed-source features. Recurrent neural networks can meet or exceed the performance of state-of-the-art traditional machine learning techniques using exclusively open-source data and models. Specifically, these results show that recurrent neural network variants can produce more than an 8% gain in accuracy in comparison with logistic regression and SVM techniques and a 15% gain over random forest when using FastText embeddings. This research found a statistical significance in the performance of …
Machine Learning Prediction Of Dod Personal Property Shipment Costs, Tiffany Tucker [*], Torrey J. Wagner, Paul Auclair, Brent T. Langhals
Machine Learning Prediction Of Dod Personal Property Shipment Costs, Tiffany Tucker [*], Torrey J. Wagner, Paul Auclair, Brent T. Langhals
Faculty Publications
U.S. Department of Defense (DoD) personal property moves account for 15% of all domestic and international moves - accurate prediction of their cost could draw attention to outlier shipments and improve budget planning. In this work 136,140 shipments between 13 personal property shipment hubs from April 2022 through March 2023 with a total cost of $1.6B were analyzed. Shipment cost was predicted using recursive feature elimination on linear regression and XGBoost algorithms, as well as through neural network hyperparameter sweeps. Modeling was repeated after removing 28 features related to shipment hub location and branch of service to examine their influence …
Learning Robust Radio Frequency Fingerprints Using Deep Convolutional Neural Networks, Jose A. Gutierrez Del Arroyo
Learning Robust Radio Frequency Fingerprints Using Deep Convolutional Neural Networks, Jose A. Gutierrez Del Arroyo
Theses and Dissertations
Radio Frequency Fingerprinting (RFF) techniques, which attribute uniquely identifiable signal distortions to emitters via Machine Learning (ML) classifiers, are limited by fingerprint variability under different operational conditions. First, this work studied the effect of frequency channel for typical RFF techniques. Performance characterization using the multi-class Matthews Correlation Coefficient (MCC) revealed that using frequency channels other than those used to train the models leads to deterioration in MCC to under 0.05 (random guess), indicating that single-channel models are inadequate for realistic operation. Second, this work presented a novel way of studying fingerprint variability through Fingerprint Extraction through Distortion Reconstruction (FEDR), a …
Bayesian Convolutional Neural Network With Prediction Smoothing And Adversarial Class Thresholds, Noah M. Miller
Bayesian Convolutional Neural Network With Prediction Smoothing And Adversarial Class Thresholds, Noah M. Miller
Theses and Dissertations
Using convolutional neural networks (CNNs) for image classification for each frame in a video is a very common technique. Unfortunately, CNNs are very brittle and have a tendency to be over confident in their predictions. This can lead to what we will refer to as “flickering,” which is when the predictions between frames jump back and forth between classes. In this paper, new methods are proposed to combat these shortcomings. This paper utilizes a Bayesian CNN which allows for a distribution of outputs on each data point instead of just a point estimate. These distributions are then smoothed over multiple …
Double Cone Flow Field Reconstruction Between Mach 4 And 12 Using Machine Learning Techniques, Trevor A. Toros
Double Cone Flow Field Reconstruction Between Mach 4 And 12 Using Machine Learning Techniques, Trevor A. Toros
Theses and Dissertations
No abstract provided.
Generalized Robust Feature Selection, Bradford L. Lott
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 …
Machine Learning Land Cover And Land Use Classification Of 4-Band Satellite Imagery, Lorelei Turner [*], Torrey J. Wagner, Paul Auclair, Brent T. Langhals
Machine Learning Land Cover And Land Use Classification Of 4-Band Satellite Imagery, Lorelei Turner [*], Torrey J. Wagner, Paul Auclair, Brent T. Langhals
Faculty Publications
Land-cover and land-use classification generates categories of terrestrial features, such as water or trees, which can be used to track how land is used. This work applies classical, ensemble and neural network machine learning algorithms to a multispectral remote sensing dataset containing 405,000 28x28 pixel image patches in 4 electromagnetic frequency bands. For each algorithm, model metrics and prediction execution time were evaluated, resulting in two families of models; fast and precise. The prediction time for an 81,000-patch group of predictions wasmodels, and >5s for the precise models, and there was not a significant change in prediction time when a …
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 …
Year-Independent Prediction Of Food Insecurity Using Classical & Neural Network Machine Learning Methods, Caleb Christiansen, Torrey J. Wagner, Brent Langhals
Year-Independent Prediction Of Food Insecurity Using Classical & Neural Network Machine Learning Methods, Caleb Christiansen, Torrey J. Wagner, Brent Langhals
Faculty Publications
Current food crisis predictions are developed by the Famine Early Warning System Network, but they fail to classify the majority of food crisis outbreaks with model metrics of recall (0.23), precision (0.42), and f1 (0.30). In this work, using a World Bank dataset, classical and neural network (NN) machine learning algorithms were developed to predict food crises in 21 countries. The best classical logistic regression algorithm achieved a high level of significance (p < 0.001) and precision (0.75) but was deficient in recall (0.20) and f1 (0.32). Of particular interest, the classical algorithm indicated that the vegetation index and the food price index were both positively correlated with food crises. A novel method for performing an iterative multidimensional hyperparameter search is presented, which resulted in significantly improved performance when applied to this dataset. Four iterations were conducted, which resulted in excellent 0.96 for metrics of precision, recall, and f1. Due to this strong performance, the food crisis year was removed from the dataset to prevent immediate extrapolation when used on future data, and the modeling process was repeated. The best “no year” model metrics remained strong, achieving ≥0.92 for recall, precision, and f1 while meeting a 10% f1 overfitting threshold on the test (0.84) and holdout (0.83) datasets. The year-agnostic neural network model represents a novel approach to classify food crises and outperforms current food crisis prediction efforts.
Unsupervised Clustering Of Rf-Fingerprinting Features Derived From Deep Learning Based Recognition Models, Christian T. Potts
Unsupervised Clustering Of Rf-Fingerprinting Features Derived From Deep Learning Based Recognition Models, Christian T. Potts
Theses and Dissertations
RF-Fingerprinting is focus of machine learning research which aims to characterize wireless communication devices based on their physical hardware characteristics. It is a promising avenue for improving wireless communication security in the PHY layer. The bulk of research presented to date in this field is focused on the development of features and classifiers using both traditional supervised machine learning models as well as deep learning. This research aims to expand on existing RF-Fingerprinting work by approaching the problem through the lens of an unsupervised clustering problem. To that end this research proposes a deep learning model and training methodology to …
Performance Of Various Low-Level Decoder For Surface Codes In The Presence Of Measurement Error, Claire E. Badger
Performance Of Various Low-Level Decoder For Surface Codes In The Presence Of Measurement Error, Claire E. Badger
Theses and Dissertations
Quantum error correction is a research specialty within the area of quantum computing that constructs quantum circuits that correct for errors. Decoding is the process of using measurements from an error correcting code, known as error syndrome, to decide corrective operations to perform on the circuit. High-level decoding is the process of using the error syndrome to perform corrective logical operations, while low-level decoding uses the error syndrome to correct individual data qubits. Research on machine learning-based decoders is increasingly popular, but has not been thoroughly researched for low-level decoders. The type of error correcting code used is called surface …
Low-Information Radiation Imaging Using Rotating Scatter Mask Systems And Neural Network Algorithms, Robert J. Olesen
Low-Information Radiation Imaging Using Rotating Scatter Mask Systems And Neural Network Algorithms, Robert J. Olesen
Theses and Dissertations
While recent studies have demonstrated the directional capabilities of the single-detector rotating scatter mask (RSM) system for discrete, dual-particle environments, there has been little progress towards adapting it as a true imaging device. In this research, two algorithms were developed and tested using an RSM mask design previously optimized for directional detection and simulated 137Cs signals from a variety of source distributions. The first, maximum-likelihood expectation-maximization (ML-EM), was shown to generate noisy images, with relatively low accuracy (145% average relative error) and signal-to-noise ratio (0.27) for most source distributions simulated. The second, a novel regenerative neural network (ReGeNN), performed exceptionally …
Learning Set Representations For Lwir In-Scene Atmospheric Compensation, Nicholas M. Westing [*], Kevin C. Gross, Brett J. Borghetti, Jacob A. Martin, Joseph Meola
Learning Set Representations For Lwir In-Scene Atmospheric Compensation, Nicholas M. Westing [*], Kevin C. Gross, Brett J. Borghetti, Jacob A. Martin, Joseph Meola
Faculty Publications
Atmospheric compensation of long-wave infrared (LWIR) hyperspectral imagery is investigated in this article using set representations learned by a neural network. This approach relies on synthetic at-sensor radiance data derived from collected radiosondes and a diverse database of measured emissivity spectra sampled at a range of surface temperatures. The network loss function relies on LWIR radiative transfer equations to update model parameters. Atmospheric predictions are made on a set of diverse pixels extracted from the scene, without knowledge of blackbody pixels or pixel temperatures. The network architecture utilizes permutation-invariant layers to predict a set representation, similar to the work performed …
Event-Based Visual-Inertial Odometry Using Smart Features, Zachary P. Friedel
Event-Based Visual-Inertial Odometry Using Smart Features, Zachary P. Friedel
Theses and Dissertations
Event-based cameras are a novel type of visual sensor that operate under a unique paradigm, providing asynchronous data on the log-level changes in light intensity for individual pixels. This hardware-level approach to change detection allows these cameras to achieve ultra-wide dynamic range and high temporal resolution. Furthermore, the advent of convolutional neural networks (CNNs) has led to state-of-the-art navigation solutions that now rival or even surpass human engineered algorithms. The advantages offered by event cameras and CNNs make them excellent tools for visual odometry (VO). This document presents the implementation of a CNN trained to detect and describe features within …
Ground Weather Radar Signal Characterization Through Application Of Convolutional Neural Networks, Stephen M. Lee
Ground Weather Radar Signal Characterization Through Application Of Convolutional Neural Networks, Stephen M. Lee
Theses and Dissertations
The 45th Weather Squadron supports the space launch efforts out of the Kennedy Space Center and Cape Canaveral Air Force Station for the Department of Defense, NASA, and commercial customers through weather assessments. Their assessment of the Lightning Launch Commit Criteria (LLCC) for avoidance of natural and rocket triggered lightning to launch vehicles is critical in approving space shuttle and rocket launches. The LLCC includes standards for cloud formations, which requires proper cloud identification and characterization methods. Accurate reflectivity measurements for ground weather radar are important to meet the LLCC for rocket triggered lightning. Current linear interpolation methods for ground …
Improving Optimization Of Convolutional Neural Networks Through Parameter Fine-Tuning, Nicholas C. Becherer, John M. Pecarina, Scott L. Nykl, Kenneth M. Hopkinson
Improving Optimization Of Convolutional Neural Networks Through Parameter Fine-Tuning, Nicholas C. Becherer, John M. Pecarina, Scott L. Nykl, Kenneth M. Hopkinson
Faculty Publications
In recent years, convolutional neural networks have achieved state-of-the-art performance in a number of computer vision problems such as image classification. Prior research has shown that a transfer learning technique known as parameter fine-tuning wherein a network is pre-trained on a different dataset can boost the performance of these networks. However, the topic of identifying the best source dataset and learning strategy for a given target domain is largely unexplored. Thus, this research presents and evaluates various transfer learning methods for fine-grained image classification as well as the effect on ensemble networks. The results clearly demonstrate the effectiveness of parameter …
Synaptic Annealing: Anisotropic Simulated Annealing And Its Application To Neural Network Synaptic Weight Selection, Justin R. Fletcher
Synaptic Annealing: Anisotropic Simulated Annealing And Its Application To Neural Network Synaptic Weight Selection, Justin R. Fletcher
Theses and Dissertations
Machine learning algorithms have become a ubiquitous, indispensable part of modern life. Neural networks are one of the most successful classes of machine learning algorithms, and have been applied to solve problems previously considered to be the exclusive domain of human intellect. Several methods for selecting neural network configurations exist. The most common such method is error back-propagation. Backpropagation often produces neural networks that perform well, but do not achieve an optimal solution. This research explores the effectiveness of an alternative feed-forward neural network weight selection procedure called synaptic annealing. Synaptic annealing is the application of the simulated annealing algorithm …