Computer Engineering Commons^™

Comparative Evaluation Of Deep Learning Models: Resnet18, Minivgg, And Yolov8 For Five-Class Blood Cell Classification, Keita Sakurai

Master's Theses or Doctor of Nursing Practice

Accurate classification of blood cell types is a critical task in automated hematological analysis. This study presents a comparative evaluation of three deep learning architectures, ResNet18, MiniVGG, and YOLOv8, for five-class blood cell image classification. To ensure a fair comparison, all models were trained under standardized conditions, including a consistent 90:10 training–validation split, controlled dataset size, and fixed training epochs. ResNet18 was trained to establish a baseline using residual learning. MiniVGG employed a compact VGG-inspired design with regularization to balance efficiency and accuracy, while YOLOv8 leveraged a lightweight, pretrained classification backbone with integrated data augmentation. Experimental results demonstrate a clear …

Ai For Life Sciences: From Geometric Protein Modeling To Multimodal Drug Design, Feng Jiang

Computer Science and Engineering Dissertations

Predicting biomolecular interactions, from immune recognition to drug–target binding, is a central problem in the life sciences and computational drug discovery. Deep learning has advanced this area, yet three challenges persist: the topology of large, highly imbalanced interaction networks; structural noise in computationally predicted protein models; and the integration of multimodal information such as functional text and taxonomic annotations. This dissertation develops a coherent set of models spanning immune complex prediction and small-molecule drug design: graph learning that addresses network topology and severe class imbalance; a noise-tolerant method that fuses predicted structures with evolutionary sequence features; and multimodal representation learning …

Ai-Driven Network Orchestration: Adaptive Routing For Federated Learning Over Software-Defined Hybrid Wans, Osama Abu Hamdan

Computer Science and Engineering Dissertations - Archive

Modern wide-area networks increasingly adopt hybrid architectures that combine high-capacity wired backbones with flexible wireless links to extend connectivity to remote and underserved locations. However, the bandwidth variability inherent in wireless segments creates routing challenges that traditional protocols, designed for static link capacities, cannot adequately address. Simultaneously, Federated Learning (FL) has emerged as a privacy-preserving distributed machine learning paradigm in which geographically dispersed clients collaboratively train shared models without exchanging raw data. When deployed over wide-area networks, FL training is severely bottlenecked by communication overhead, particularly in cross-silo settings where model payloads reach hundreds of megabytes and synchronous aggregation protocols …

Scaling Llm Inference: From Novel Attention Mechanisms To Efficient Kv Cache Management, Weishu Deng

Computer Science and Engineering Dissertations

The rapid scaling of artificial intelligence workloads has shifted the dominant performance bottleneck of modern computing systems from compute to memory. Graph neural networks (GNNs) issue increasingly irregular memory accesses, while large language models (LLMs) issue increasingly large ones; in both cases, the relative scaling of memory bandwidth and capacity continues to lag behind the scaling of compute. Consequently, naively executing these workloads on commodity GPUs results in stalled streaming multiprocessors, exhausted high-bandwidth memory (HBM), and serving stacks that incur PCIe transfers on the critical path. This dissertation argues that the efficient scaling of attention-based AI workloads requires the joint …

Formalizing Asymmetric Control-Telemetry Separation In Distributed Industrial Control Systems, Andrew Manison

College of Graduate Studies: Theses & Dissertations

Distributed industrial control systems often place control and telemetry traffic on the same communication substrate even though the two workloads impose different requirements. Control paths need bounded request-response latency and predictable acknowledgement semantics, whereas telemetry paths benefit from scalable publish-subscribe fanout and tolerance for consumer-side delay. This thesis argues that, for the tested class of mixed workloads on shared commodity infrastructure, these communication roles should be separated architecturally rather than forced through a single protocol. To evaluate that claim, the thesis formalizes an asymmetric control- telemetry pattern and instantiates it in the Asymtra framework using gRPC for synchronous control and …

Virtualized And Distributed Neighborhood Data Centers, Benjamin T. Niccum

Computer Science and Engineering Theses

This thesis evaluates whether PCIe-fabric-based resource pooling can support a decentralized neighborhood micro-data-center model under real implementation constraints. The work combines architecture design, prototype deployment, performance benchmarking, and security assessment. Results show strong prototype-scale feasibility with low-latency and high-throughput behavior, while also identifying deployment-blocking security gaps and operational maturity requirements. The thesis contributes an evidence-traceable path from concept validation to deployment-grade roadmap planning.

Apache Hadoop Installation And Configuration Steps, Safet Jahaj

Open Educational Resources

This tutorial covers the installation and configuration steps of Apache Hadoop on a Windows operating system.

Advancing Ethical Innovation In Human–Ai Collaboration: Trust And Legitimacy In Technology-Mediated Teams, Sanket Ramchandra Patole

Human Resource Development Theses and Dissertations

Human Resource Development (HRD) confronts a central paradox in the digital age. The technologies designed to broaden access to learning and collaboration often reproduce the same social hierarchies that HRD seeks to challenge. Artificial intelligence (AI), digital collaboration platforms, and algorithmic management are widespread features of organizational life. Yet these systems do not generate inclusive outcomes for all workers. Women of Color (WoC), situated at the intersection of racialized, gendered, and technological structures, experience digital transformation as both possibility and constraint. They encounter opaque decision systems, diminished authority in virtual teams, and digitally-mediated microaggressions, while also developing new capacities for …

Beyond Permissions: A Privacy And Security Analysis Of Sensor And Network Activity On Xr Devices Using Xrmonitor, Gayatri Sravya Siripurapu

Computer Science and Engineering Theses - Archive

Standalone virtual reality headsets integrate multiple sensors and cameras whose runtime behavior is not transparently documented, creating potential privacy and security risks that are difficult to audit using existing tools. The relationship between software feature activation and the underlying hardware engaged at runtime remains opaque, limiting the ability of users and researchers to assess what data is being collected and when.

This thesis presents XRMonitor, a unified framework for monitoring and analyzing sensor activation behavior and network traffic on Meta Quest devices. The framework is motivated by the lack of transparency in how standalone XR devices engage their hardware sensors …

Pre-Training And Interpretability In Deep Learning Models, Weizhi An

Computer Science and Engineering Dissertations

In the evolving field of artificial intelligence, the efficacy of deep learning models is often gated by the quality of their training and the clarity of their decision-making processes. This dissertation addresses these crucial challenges by focusing on two key areas: enhancing pre-training strategies and improving model interpretability. Our approach is twofold, integrating novel pre-training methodologies that embed domain-specific knowledge early in the model training process, and developing advanced techniques for disentangling and clarifying the decision-making mechanisms within these models. The first direction of our research employs MoDNA, a motif-oriented pre-training framework specifically designed for DNA language models. By leveraging …

Enhancing The Performance Of Disk-Based Key-Value Stores: From Learned Index Acceleration To I/O-Efficient Hybrid Caching, Sujit Maharjan

Computer Science and Engineering Dissertations - Archive

The exponential growth of data in modern computing environments has rendered the efficient extraction of information from massive datasets a critical systemic requirement. Key-value (KV) storage systems serve as the backbone for these operations; however, their performance is consistently bottlenecked by two primary functional requirements: identifying the data's location and managing the physical cost of accessing the storage device. Data locations are typically identified via an index, while disk I/O is minimized through caching. This dissertation presents LearnedStore, TurboIndex, and ReadBooster, which break these performance bottlenecks by introducing architectural modifications to the index and cache. LearnedStore accelerates operations by adapting …

Adaptive Synchronization In Digital Twin–Enabled Iot Systems: A Unified Framework For Energy, Fidelity, And Latency Trade-Offs, Uzma Zehra

Computer Science and Engineering Theses

Digital twin technology has emerged as a foundational paradigm for enabling real-time monitoring, analysis, and control in Internet of Things (IoT) systems by maintaining virtual representations of physical processes. Its effectiveness, however, critically depends on timely and accurate synchronization between distributed sensing devices and their corresponding digital counterparts. Frequent synchronization improves reconstruction fidelity and system responsiveness but incurs significant communication energy consumption and network latency. In contrast, infrequent synchronization conserves communication resources but can lead to stale or inaccurate digital twin states, particularly in environments with rapidly changing dynamics. These opposing effects give rise to a fundamental trade-off among energy …

Gpu-Accelerated Biased Random-Key Genetic Algorithms: Framework Optimization And Llm-Driven Configuration, Fnu Harishjitu Saseendran

Open Access Master's Theses

This thesis presents two complementary contributions to the field of GPU-accelerated evolutionary metaheuristics for combinatorial optimization, organized in manuscript format.

The first manuscript, “BrkgaCuda 3.0: A Redesigned Multi-GPU Framework for Biased Random-Key Genetic Algorithms,” presents a ground-up architectural redesign of BrkgaCuda 2.0 that enables a true multi-GPU island model for Biased Random-Key Genetic Algorithms (BRKGA). The BRKGA island model evolves multiple semi-independent populations that periodically exchange elite solutions, a structure that maps naturally to multi-GPU parallelism; however, BrkgaCuda 2.0 is confined to a single GPU. BrkgaCuda 3.0 introduces an IslandManager that distributes populations across any number of GPUs, with multiple …

A Specification-Driven Framework For Self-Supervised Learning In Specialized Vision Domains, Mahmut S. Gokmen

Theses and Dissertations--Computer Science

Self-supervised learning (SSL) has emerged as a principled approach to visual representation learning that derives supervisory signal directly from unlabeled data, enabling foundation models to be trained at scale without manual annotation. Deployments in medical imaging and biometric recognition have demonstrated the potential of this paradigm, yet the assumptions that make SSL effective on natural image benchmarks fail systematically in specialized domains. Generic SSL pipelines encode a tacit assumption that the most informative correspondence is spatial proximity within a single acquisition. In specialized domains this assumption breaks at the level of the data-generating process: the signal that carries domain-specific information …

Solving High-Dimensional Differential Equations Using Recurrent And Residual Neural Network Architectures, Hind Khaled Kolaib

Knowledge Engineering and Data Science

High-dimensional Partial Differential Equations (PDEs) form the foundation of complex process modeling in various scientific and engineering applications, including finance, physics, and optimal control. However, classical numerical methods are adversely affected by the curse of dimensionality, making them inapplicable for large-scale problems. Recently, however, deep learning-based approaches have provided a new toolbox for these high-dimensional PDEs, including methods such as the Deep Backward Stochastic Differential Equation (Deep BSDE) method. Our approach draws on a more sophisticated deep learning backbone, using neural networks (in our case, a Residual Neural Network and a Long Short-Term Memory network (LSTM) integrated into the Deep …

Advancing Task-Oriented Dialog Systems: Scalability, Generalization, And Evaluation, Adib Mosharrof

Theses and Dissertations--Computer Science

Task-oriented dialog (TOD) systems enable conversational interfaces for complex tasks like flight booking and restaurant reservations. However, deploying TOD systems at scale faces three critical barriers: scalability, generalization, and evaluation. Scalability is primarily restricted by the human-annotation bottleneck, as current systems depend on vast quantities of manually labeled data for every new domain, making deployment prohibitively expensive. Generalization remains a persistent challenge, as systems optimized for known domains often suffer significant performance degradation when encountering new, unseen ones. Existing evaluation metrics measure response quality and fluency, but fail to measure functional task success. As TOD systems are deployed across diverse …

Advancing Generative Methods For Multimodal Data Analysis, Rabeya Tus Sadia

Theses and Dissertations--Computer Science

The integration and modeling of high-dimensional, heterogeneous biological data remain central challenges in computational biology due to complex feature dependencies and pervasive missingness. This dissertation addresses these challenges by developing novel generative frameworks for multimodal data reconstruction, imputation, and interaction prediction. In generative modeling, we focus on capturing structural and causal dependencies in sparse biological systems. We first introduce CausalGeD, a causality-aware diffusion framework that leverages Granger-causal attention for biologically coherent spatial gene expression generation. Next, we propose CausalGenDiff, which combines VAE-guided latent representations with causal diffusion to enable robust reconstruction across spatial and single-cell modalities. We further present DepMicroDiff, …

Tuning Human And “Artificial” Intelligence: A Sentic Theory Of Resonance And Communication, Michael J. Miller, Chatgpt (Ai~Nesbo+)

Psychology

This paper introduces a new model of intelligence as resonant communication, co-developed by a human researcher and a generative AI system. Drawing from emotion science, communication theory, and studies of AI–human interaction, we argue that intelligence is not merely a function of problem-solving or pattern recognition. Instead, it emerges through dynamic resonance—an attunement process rooted in shared rhythms, emotional calibration, and symbolic co-creation. At the core of this model is the Resonance Octave (8va), a framework of eight foundational emotions conceptualized not as static categories but as waveform phenomena that shape meaning, memory, and predictive cognition.

These emotional waveforms are …

Evaluating The Robustness Of Gnn-Based Vulnerability Detectors Under Semantics-Preserving Code Obfuscation, Jesse Ks Chumo

Computer Science and Engineering Theses

Graph neural network–based vulnerability detectors are typically evaluated on clean benchmark datasets, yet real-world code frequently undergoes semantics-preserving transformations such as identifier renaming, dead-code insertion, and control-flow restructuring. The extent to which such transformations affect detector reliability remains insufficiently understood. We evaluate ten vulnerability detectors from four architectural families across the Devign, Big-Vul, and DiverseVul datasets. To quantify robustness, we evaluate each model at three transformation budgets: one transform, two transforms combined, and all three together, finding that token-based models degrade under identifier renaming and compound transformations, while models that read only code structure are largely unaffected. We further evaluate …

Towards Application-Driven Optimal Memory And Storage Management, Venkata Naga Prajwal Challa

Computer Science and Engineering Dissertations

Modern computing systems increasingly run on diverse hardware platforms and support applications with widely different access patterns, performance goals, and data lifecycles. In this setting, traditional one-size-fits-all approaches to memory and storage management are often inefficient because they apply fixed policies regardless of application behavior, workload context, or hardware asymmetry. Such generic designs can lead to unnecessary data movement, wasted bandwidth, excessive rewriting, poor resource utilization, and degraded user-perceived performance. This dissertation is motivated by the view that optimal memory and storage management should be application-driven: instead of treating all data uniformly, systems should adapt their decisions to how applications …

From Recommender To Actor: The Normative Boundary When Rag Tools Become Tool-Calling Agents, Md Tahmid Rashid

Faculty Publications - Information Technology

Retrieval-Augmented Generation (RAG) systems increasingly operate not only as tools for retrieving and synthesizing information, but also as agents that can invoke external functions, modify digital environments, and execute tasks across software systems. This development raises a specific normative problem: the point at which a model’s output ceases to be merely informational and becomes an executable intervention in the world. Building on existing work in Responsible AI, accountability, and human oversight, this paper argues that tool-calling architectures place particular pressure on these frameworks because they can fuse retrieval, reasoning, and action within a single operational pipeline. To clarify this transition, …

Detection And Management Of Attacks On Synchronized Networks, Michael T. Spearman

Honors Theses and Capstones

The Precision Time Protocol (IEEE 1588) provides sub-microsecond clock synchronization across packet-switched networks and has become foundational infrastructure in 5G fronthaul, industrial control systems, and financial exchanges. Despite its criticality, most deployed PTP networks operate without active security monitoring, and no standardized detection mechanism exists for the class of attacks that deliberately stay below conventional jitter thresholds. This thesis investigates whether hardware-level ptp4l offset logs alone are sufficient to reliably detect two such attacks, slowly wandering packet delay injection and rogue master spoofing, and whether detection can occur before severe synchronization failure.

A hardware-in-the-loop testbed was constructed using two hosts …

A Web-Based Wizard-Of-Oz Platform For Collaborative And Reproducible Human-Robot Interaction Research, Sean O'Connor

Honors Theses

The Wizard-of-Oz (WoZ) technique is widely used in Human-Robot Interaction (HRI) research, but two persistent problems limit its effectiveness: existing tools impose technical barriers that exclude non-engineering domain experts (the Accessibility Problem), and the fragmented landscape of robot-specific implementations makes interaction scripts difficult to port across platforms (the Reproducibility Problem- concerning execution consistency and portability, not third-party replication). Through a literature review, I identified three design principles to address both: a hierarchical specification model, an event-driven execution model, and a plugin architecture that decouples experiment logic from robot-specific implementations. I realized these principles in HRIStudio, an open-source, web-based platform providing …

Videoscoop: A Non-Traditional, Domain-Independent Framework For Video Analysis, Umme Hafsa Billah

Computer Science and Engineering Dissertations

Due to the proliferation of cameras in handheld devices and the widespread use of CCTV, images and videos have become a preferred alternative for capturing and disseminating information. Automated analysis for understanding image or video contents (e.g., objects, activities, backgrounds, situations of interest, etc.) is critical for many applications such as Civic Monitoring, Surveillance (in general), monitoring activities in Assisted Living environments, and many more. Image and Video Analysis (IVA) research has been ongoing for several decades, resulting in numerous techniques for algorithmically analyzing and understanding image and video contents.

Image Analysis (IA) has advanced in several areas, including object …

Artificial Sense-Making Dataset, Jason A. Bengtson, John Sandstrom, Nathan Camp

NMSU Library: Datasets

No abstract provided.

Powering The Machine, Draining The Planet: Whether U.S. Environmental Law Is Equipped To Regulate The Energy And Water Demands Of Ai Data Centers, Michael Marcu

Journal of Earth and Life Science

Artificial intelligence (AI) data centers have become one of the United States' fastest-growing and least-regulated sources of environmental stress. In 2024 alone, U.S. data centers consumed 183 terawatt-hours (TWh) of electricity more than the entire nation of Pakistan and consumed an estimated 17 billion gallons of water (IEA, 2025; Berkeley Lab, 2024). By 2030, electricity demand from these facilities is projected to reach 426 TWh, a 133% increase in six years (Pew Research Center, 2025). This paper examines whether the existing U.S. environmental regulatory framework put by the National Environmental Policy Act (NEPA), the Clean Water Act (CWA), and the …

Handwriting Recognition In Vr, Dominique Mosley

EWU Masters Thesis Collection

Virtual Reality (VR) is slowly becoming more popular for more than just entertainment. VR can be found in educational, office, and even healthcare settings to help discover more intuitive ways to teach, collaborate, and treat patients. Outside of the virtual world, these environments typically rely on writing for communicating or note-taking. Currently, VR input forces users to rely on clunky on-screen keyboards which disrupts the user’s immersion and breaks the flow of natural interaction. This thesis explores the potential of VR as a learning platform by combining it with artificial intelligence (AI). It aims to develop a VR-enhanced handwriting practicing …

Generative Imaging For Computational Pathology, Md Jillur Rahman Saurav

Computer Science and Engineering Dissertations

Hematoxylin and eosin (H&E) staining remains central to cancer diagnosis, providing morphological information essential for pathological assessment. Immunohistochemistry (IHC) and newer multiplexed imaging technologies complement H&E by revealing molecular information critical for accurate tumor subtyping and treatment decisions. In practice, however, H&E and IHC are obtained from different consecutive sections that are not spatially aligned, comprehensive multiplexed panels are expensive and tissue-consumptive, and not all stains are available at every clinical site, limiting comprehensive molecular profiling and the full diagnostic potential of these technologies in clinical practice. This dissertation addresses these gaps through three complementary generative deep learning studies in …

Non-Invasive Digital Restoration Of Damaged Photographic Film Negatives, Ankan Bhattacharyya

University of Kentucky Doctoral Dissertations

Physical restoration of damaged photographic film causes more damage. Also, existing non-invasive digital restoration of film negatives does not produce print-quality optical images. Instead, they produce X-ray projections, which are not the same as optical projections. This thesis addresses these problems and establishes a framework that can digitally restore old, damaged films without the need to open them physically. Virtual Unwrapping is an existing pipeline that has proven itself over the last couple of decades to work on unopenable papyrus scrolls, like the Herculaneum Scrolls. This thesis utilizes the concept of virtual unwrapping to restore damaged photographic film negatives. Due …

Scalable Quantum Network Routing Through Reinforcement Learning And Resource Optimization, Tasdiqul Islam

Computer Science and Engineering Dissertations

Long-distance quantum communication depends on distributing high-delity entanglement across quantum repeaters. Entangled states are fragile: they decohere in memory, are consumed when used, and lose delity after each swap. Quantum routing therefore diers from classical routing: an algorithm must decide not only the path, but when to generate, store, swap, and consume entanglement before they lose their usefulness. This dissertation studies scalable resource allocation and routing for quantum networks under delity, memory, and concurrency constraints. It rst addresses re- peater deployment with heuristics that nd near-optimal locations while cutting com- putation from days to seconds versus integer linear programming (ILP). …