Computer Sciences Commons^™

Using Machine Learning And Deep Learning Algorithms To Improve Low Birthweight Prediction, Yang Ren

Theses and Dissertations

Low birthweight (LBW) is a major public health issue resulting in increased neonatal mortality and long-term health complications. Traditional LBW analysis methods, focusing on incidence rates and risk factors through statistical models, often struggle with complex unseen data, and thus, their effectiveness is limited in early prevention of LBW, requiring more advanced LBW prediction models. Therefore, this dissertation delves into this important research area by proposing and examining novel machine learning (ML) and deep learning (DL) algorithms, aiming to predict LBW more accurately during the early stage of pregnancy. This dissertation consists of three studies, strategically designed to build upon …

Advancing Robust Autonomous System Localization: Labeling Optimizations For Convolutional Neural Networks, Jeffrey L. Choate

Theses and Dissertations

AAR is increasingly critical as aircraft autonomy advances, particularly for the Global Strike mission of the USAF, enhancing operational range and endurance. Traditional methods relying on GPS and custom communication links are limited in GPS-denied environments. This dissertation advances a single camera method to estimate object pose across three interconnected studies. The system trains a CNN on synthetic imagery to predict bboxes for object components, Solve-PnP algorithm finds the 6DoF pose, then employs novel pseudo-labeling on real-world images. These findings are pivotal for the AAR community and contribute to robotics, computer vision, and CNN research. By enabling robust GPS-free autonomous …

Autonomous Experimentation For Accelerated Calibration Of Fused Deposition Modeling 3d Printers, Graig S. Ganitano

Theses and Dissertations

Additive Manufacturing (AM), also known as 3D printing, has emerged as a key component of Industry 4.0, enabling reduced cost, quick production, greater sustainability, and increased design complexity compared to its traditional manufacturing counterpart. Currently, Fused Deposition Modeling (FDM) technology dominates the AM market with respect to the number of 3D printers in use. However, the FDM process is sensitive to changes in system configuration, especially the feedstock material. Utilizing a new feedstock requires a time-consuming trial-and-error process to identify optimal settings for a large number of process parameters, acting as a barrier to the technology.

To enable greater accessibility …

Integrating Blockchain Technology Into The Software Development Life Cycle To Satisfy The Software Bill Of Materials Requirement For Government Software Systems, Walter T. Scott Ii

Theses and Dissertations

This thesis explores the integration of Blockchain Technology (BT) into the Software Development Life Cycle (SDLC) to satisfy the Software Bill of Materials (SBOM) requirement for government software systems. This study begins by synthesizing a standard SDLC definition from various government and industry references, which may provide the foundation for future efforts to standardize software development practices across the government software development community. This study proceeds to define working definitions for the software supply chain (SSC) and software supply chain management (SCM) before introducing and detailing the SBOM requirement as well as providing an overview of prior research regarding SBOMs …

Estimating Dis Performance Using Mininet, Ryan D. Winz

Theses and Dissertations

Real time distributed simulation is an exceptionally useful tool for training and wargaming used by the military and industry alike. This research aims to provide scenarios and structures to evaluate the effect of distributing simulations among different compute nodes. Specific scenarios involve the analysis of performance as a function of latency and the degree network protocols and reliability affect simulation performance. Various standards exist for administering geographically separated simulations. The focus of this thesis will be on the Distributed Interactive Simulation standard, a peer-to-peer open standard for simulation messages to adhere to, but lessons can be extended to other standards.

Machine Visual Perception For Autonomous Docking Maneuvers, Derek B. Worth

Theses and Dissertations

This dissertation presents a novel approach to autonomous docking using machine learning for visual perception, particularly during probe and drogue aerial refueling. Autonomous vehicles have become pervasive in both civilian and defense sectors, and their ability to interact with their surroundings and each other autonomously is critical for future operations. Traditional methods relying on signals or inertial sensors face significant limitations such as interference, jamming, and drift. This research focuses on developing a computer vision-based solution to overcome these limitations. A novel pipeline, termed relative vectoring, is introduced, which utilizes dual object detection and machine learning to estimate relative positions …

Evaluating Audio-Compression Techniques, Thien Tuong Duong

Theses and Dissertations

Perceptual quality of audio is the aural accuracy and fidelity perceived by the listener. It is how humans respond to the accuracy, intelligibility, or fidelity of aural media. High-fidelity audio represents high accuracy and minimal distortion and is the standard against which compression algorithms are assessed. We argue that maintaining high-fidelity audio is not just a technical requirement but is also about respecting the artistry of music production. The primary goal of this thesis is to compare the performance of popular audio codecs including file size and conversion speed of each codec when encoding the same file. A secondary goal …

Predictive Filtering-Based Image Inpainting, Xiaoguang Li

Theses and Dissertations

Image inpainting is an important challenge in the computer vision field. The primary goal of image inpainting is to fill in the missing parts of an image. This technique has many real-life uses including fixing old photographs and restoring ancient artworks, e.g., the degraded Dunhuang frescoes. Moreover, image inpainting is also helpful in image editing. It has the capability to eliminate unwanted objects from images while maintaining a natural and realistic appearance, e.g., removing watermarks and subtitles. Disregarding the fact that image inpainting expects the restored result to be identical to the original clean one, existing deep generative inpainting methods …

Object Classification, Detection And Tracking In Challenging Underwater Environment, Md Modasshir

Theses and Dissertations

The main contributions of this thesis is the applicability and architectural designs of deep learning algorithms in underwater imagery. In recent times, deep learning techniques for object classification and detection have achieved exceptional levels of accuracy that surpass human capabilities. However, the effectiveness of these techniques in underwater environments has not been thoroughly researched. This thesis delves into various research areas related to underwater environments, such as object classification, detection, semantic segmentation, pose regression, and semi-supervised retraining of detection models.

The first part of the thesis studies image classification and detection. Image classification is a fundamental process that involves assigning …

Automated Data-Flow Optimization For Digital Signal Processors, Madushan Thilina Abeysinghe

Theses and Dissertations

Digital signal processors (DSP), which are characterized by statically-scheduled Very-Long Instruction Word architectures and software-defined scratchpad memory, are currently the go-to processor type for low-power embedded vision systems, as exemplified by the DSP processors integrated into systems-on-chips from NVIDIA, Samsung, Qualcomm, Apple, and Texas Instruments. DSPs achieve performance by statically scheduling workloads, both in terms of data movement and instructions. We developed a method for scheduling buffer transactions across a data flow graph using data-driven performance models, yielding a 25% average reduction in execution time and a reduction of up to 85% DRAM utilization for randomly-generated data flow graphs. We …

Scene Text Detection And Recognition Via Discriminative Representation, Liang Zhao

Theses and Dissertations

Scene texts refer to arbitrary text presented in an image captured by a camera in the real world. The tasks of scene text detection and recognition from complex images play a crucial role in computer vision, with potential applications in scene understanding, information retrieval, robotics, autonomous driving, etc. Despite the notable progress made by existing deep-learning methods, achieving accurate text detection and recognition remains challenging for robust real-world applications. The challenges in scene text detection and recognition stem from: 1) diverse text shapes, fonts, colors, styles, layouts, etc.; 2) countless combinations of characters with unfixed attributes for complete detection, coupled …

On Parallelization Of Graph Algorithms, Performance Modelling And Autonomous 3d Printable Object Synthesis, Shams-Ul-Haq Syed

Theses and Dissertations

The degree of hardware level parallelism offered by today’s GPU architecture makes it ideal for problem domains with massive inherent parallelism potential, fields such as computer vision, image processing, graph theory and graph computations. We have identified three problem areas for purpose of this research dissertation, under the umbrella of performance improvement by harnessing the power of GPUs for novel applications. The first area is concerned with k-vertex connectivity in graph theory, the second area deals performance evaluation using extended roofline models for GPU parallel applications and finally the third problem area is related to synthesis 3D printable objects from …

Mapping Urban Tree Canopy Using Publicly Available Satellite Data, Rosemary Mcguinness

Theses and Dissertations

This project addresses the need for accessible, cost-effective tools for quantifying spatial and temporal changes in tree canopy cover in urban areas. Urban tree canopy provides a wide range of ecosystem services, including lowering air temperatures, reducing pollution, and mitigating stormwater runoff. Cities around the world have placed the expansion of their urban forests at the center of their sustainability goals. Consistent and timely data on urban tree canopy is essential for urban greening initiatives to succeed. Existing methods of accessing information about urban tree canopy are highly technical, costly, and labor-intensive, while the freely available source of tree canopy …

White Light Specular Reflection Data Augmentation For Polyp Detection, Jose Angel Nunez

Theses and Dissertations

Colorectal cancer is among the deadliest cancers, but fortunately, this type of cancer can be prevented. The best current method of prevention is via detecting the bad polyps in the colon in time. Furthermore, the best method we have available to detect these bad polyps is through colonoscopies. Even though a lot of lives have been saved via these methods, it is still not perfect because of human error. Integrating artificial intelligence into colonoscopy procedures is our next evolution in increasing our prevention of colorectal cancer. Polyp detectors are one of the tools brought by advancements in technology that may …

Comparative Analysis Of Transfer Learning Strategies For Polyp Detection In Colonoscopy Images Using Yolov8, Fabian Vazquez Jr.

Theses and Dissertations

The early detection of polyps during colonoscopy procedures is crucial for preventing colorectal cancer, a leading cause of cancer-related deaths globally. Traditional methods for polyp detection are often time-consuming and prone to human error. This thesis investigates the effectiveness of transfer learning, the process of taking a pre-trained model that was trained on a large dataset and adapting it to a new, but related task, requiring less data and time for training. This research compares whether the YOLOv8 model trained from scratch on a specific polyp dataset is outperformed by transfer learning methods such as utilizing a pretrained model on …

Exploratory Analysis In Rio Grande Valley Real Estate Development, Mathew C. Sosa

Theses and Dissertations

The aim of this thesis is to apply exploratory data analysis and machine learning (ML) to answer critical questions in real estate development within the Rio Grande Valley (RGV). The real estate development sector is highly dynamic and relies on multiple integrated systems and planning to bring new homes to the market. Customer Relationship Management (CRM) software and real estate listing services, such as Redfin, are utilized to manage the customer lifecycle and analyze the local market, respectively. Two of the most crucial aspects of the customer lifecycle are the first home tour and deal closing. This paper explores the …

Intrinsic Universality In Tile Automata And Related Results, Elise C. Grizzell

Theses and Dissertations

The Tile Automata (TA) model describes self-assembly systems in which monomers can build structures and transition with an adjacent monomer to change their states. This paper shows that seeded TA is a non-committal intrinsically universal model of self-assembly. We present a single universal Tile Automata system containing approximately 4600 states that can simulate (a) the output assemblies created by any other Tile Automata system Γ, (b) the dynamics involved in building Γ’s assemblies, and (c) Γ’s internal state transitions. It does so in a non-committal way: it preserves the full non-deterministic dynamics of a tile’s potential attachment or transition by …

Increasing The Robustness Of Machine Learning By Adversarial Attacks, Gourab Mukhopadhyay

Theses and Dissertations

By perturbation or physical attacks any machine can be fooled into predicting something else other than the intended output. There are training data based on which the model is trained to predict unknown things. The objective was to create noises and shades of different levels on the images and do experiments for measuring accuracy and making the model classify the traffic signs. When it comes to adding shades to the pictures, pixels were modified for three different layers of the pictures. The experiment also shows that with the shadows getting deeper, the accuracies drop significantly. Here, some changes in pixels …

Personalized Driving Using Inverse Reinforcement Learning, Rodrigo J. Gonzalez Salinas

Theses and Dissertations

This thesis introduces an autonomous driving controller designed to replicate individual driving behaviors based on a provided demonstration. The controller employs Inverse Reinforcement Learning (IRL) to formulate the reward function associated with the provided demonstration. IRL is implemented through a dual-feedback loop system. The inner loop utilizes Q-learning, a model-free reinforcement learning technique, to optimize the Hamilton-Jacobi-Bellman (HJB) equation and derive an appropriate control solution. The outer loop leverages this derived control solution to generate parameters for the reward function, which are subsequently integrated into the HJB equation. The ultimate control policy is deduced from the final reward function obtained …

A Scalable Parallel Processing Design For The Data Washing Machine: An Unsupervised Entity Resolution System, Nicholas Kofi Akortia Hagan

Theses and Dissertations

Entity Resolution (ER) has been one of the bedrocks in the creation of information systems by ensuring ambiguous entities are identified and resolved by linking. One common design approach of traditional ER systems is to run in single-threaded mode, which makes the system prone to out-of-memory error when processing larger datasets. The Data Washing Machine (DWM) as a proof-of-concept of an unsupervised cluster ER system is indifferent from this common design bottleneck. The original prototype design of the DWM requires shared memory tables and dictionaries of tokens, and its single-threaded nature makes it not scalable, hence not viable for real-world …

Design And Implementation Of Truly Random Number Generation Using Memristors For In-Memory Computing, Nick Felker

Theses and Dissertations

This paper proposes a new security module based on non-volatile memory. The module uses a memristor-based true random number generator to generate random numbers which can be used for cryptography. The module is implemented in software using a modified RISC-V instruction set architecture. The paper evaluates the performance of the module using the RISC-V simulator Gem5. The results show that the module can generate random numbers at a rate of 63 microseconds per number, which is faster than the standard C library’s random number generator. The module can also be used to scramble strings of characters and generate hashes of …

Efficient Machine Learning On Scientific Data Using Bayesian Optimization, Rui Xin

Theses and Dissertations

Deep Learning is pivotal in advancing data analysis across various scientific fields, from genomics to materials discovery. Despite its widespread use, efficiently learning from limited data and operating under resource constraints remains a significant challenge, often limiting its full potential in environments where data is scarce or resources are restricted. This dissertation explores Active Learning and Automated Machine Learning (AutoML) powered by Bayesian Optimization to enhance the efficiency of machine learning across multiple disciplines. It focuses on algorithm optimization and data management through three interconnected studies. In the first study, we investigate how data management technique - active learning helps …

Multi-Scale Deep Representation Learning In Synthetic Biology, Xiaoyi Liu

Theses and Dissertations

Synthetic biology advances and combines the expertise of engineers and biologists, bridging the gap between engineering and natural life. Synthetic biology has been generally categorized into two broad branches by developing new biological components, networks, and systems to reprogram organisms. The first branch involves using synthetic molecules to mimic natural biological functions. The second branch focuses on assembling natural biological components in novel ways, aiming to produce systems with unique, practical functions. Thus, the de novo engineering of biological modules and synthetic pathways is used in related practical bioengineering applications, such as drug-targeting strategies and microbial product manufacturing. Therefore, synthetic …

Integration Analysis Of Nssm, Winlogon, And P2p Networks: Security Impact And Mitigation Strategies, Maisa Emneina

Theses and Dissertations

In today's digital world, where everything is interconnected, new security threats are constantly emerging. This thesis explores the security risks that come from the combination of three specific technologies: Non-Sucking Service Manager (NSSM), Winlogon Helper DLLs, and Peer-to-Peer (P2P) networks. NSSM is a powerful tool for managing services, but it can also be used by attackers to find and exploit system weaknesses. Winlogon Helper DLLs are essential for the Windows logon process, making them a critical target for attacks. When you add P2P networks into the mix, which many applications use to share data, the potential for security issues increases …

Preregistration And Registration As A New Method For Transparency Of External Validation In Artificial Intelligence And Machine Learning Applications To Address Overfitting, Underspecification, And Shortcut Learning, Marilyn Elaine Gartley

Theses and Dissertations

Statistics has been defined as the study of how information should be employed to reflect on and give guidance for action in a practical situation involving uncertainty. The essence of uncertainty is that there is more than one possible outcome, and the actual outcome is unknown in advance; it is indeterminate. The goal of statistical methods is inference: namely, statistical inference—reaching conclusions about populations or deriving scientific insights from data which are collected from a representative sample of that population through providing a mathematical understanding of inference, quantifying the degree of support that data offer for assertions of knowledge, as …

Enhancing The Security In High-Speed Networks Using P4 Programmable Switches, Ali Alsabeh

Theses and Dissertations

Network security has become increasingly essential in today's networks due to the growth of various network applications, such as Machine Learning (ML) and Fifth-Generation (5G) Networks. One fundamental Internet protocol is the Domain Name System (DNS), which maps domain names to Internet Protocol (IP) addresses. Despite its importance, DNS traffic is often forwarded without being analyzed, making it a center of ever-evolving attacks. Traditionally, defense strategies are implemented on fixed-function security middleboxes that are costly, proprietary, and hard to manage. Alternatively, defenses implemented in software use general-purpose servers (e.g., the control plane of a Software-Defined Networking (SDN) network) that cannot …

Hierarchical Quantized Autoencoders: Using Hierarchical Models For Data Compression Across Multiple Domains, Armani Lorenzo Rodriguez

Theses and Dissertations

In the era of vast data processing and transmission, sending data over a channel for downstream operations is a very common occurrence. The bandwidth of this data channel acts as a limiting factor in this operation, capping the amount of data that can be sent over a time period. Therefore, in addition to pursuing advancements in networking technology, there exists a need for more efficient means of data compression. Learned compression is the application of machine learning models to the data compression problem, and in this study, we leverage the ability of neural networks to learn the underlying structure of …

An Empirical Study On Detecting And Explaining Global Structural Change In Evolving Graph Using Martingale, Tarun Teja Kairamkonda

Theses and Dissertations

There is a growing interest in practical applications involving networks of interacting entities such as sensor networks, social networks, urban traffic networks, and power grids, all of which can be represented using evolving graphs. Changes in these evolving graphs can signify shifts in the behavior of interacting entities or alterations in the patterns of their interactions. Identifying and detecting these changes is crucial for addressing potential challenges or opportunities in various domains. In this study, we propose an approach for detecting structure change in evolving graphs based on the martingale change detection framework on multiple graph features extracted over time. …

Integration Of Machine Learning In Structural Health Monitoring For Damage Identification And Response Prediction In Bridges, Naga Lakshmi Chittitalli Ravuri

Theses and Dissertations

Machine learning-based structural health monitoring (ML-SHM) plays a pivotal role in enhancing structural resilience. By recognizing potential hazards, implementing resistance measures, facilitating swift recovery, and continuously monitoring structural health, ML-SHM ensures proactive maintenance and minimizes recovery delays post-events. Leveraging machine learning algorithms and sensor data, ML-SHM enables early detection of anomalies, prediction of failures, and adaptive responses, enhancing the structure's ability to withstand and recover from adverse conditions. This integrated approach not only improves the structure's performance and adaptability but also contributes to overall safety and longevity. This thesis presents a comprehensive exploration of structural health monitoring (SHM) techniques for …

Reinforcement Learning For Robotic Tasks: Analyzing And Understanding The Learning Process Using Explainable Artificial Intelligence Methods, Brian J. Campana

Theses and Dissertations

As deep reinforcement learning (RL) models gain traction across more industries, there is a growing need for reliable agent-explanation techniques to understand these models. Researchers have developed explainable artificial intelligence (XAI) methods to help understand these 'black boxes'. While these models have been tested on many supervised learning tasks, there is a lack of examination of how these well these methods can explain hard reinforcement learning problems like robotic control. The sequential nature of learning RL policies and testing episodes create fundamentally different policies over time compared to more traditional supervised learning models. In this thesis, two important questions are …