Computer Sciences Commons^™

Cache-Augmented Generation In Rag Pipelines: Fast And Memory-Efficient Approach To Multi-Agent Knowledge Query Systems, Yaju Gopal Shrestha

Honors Theses

This thesis presents an implementation and evaluation of Cache-Augmented Generation (CAG) for knowledge query systems, building upon the approach introduced by Chan et al. (2024). Traditional Retrieval-Augmented Generation (RAG) systems (Lewis et al., 2020) face challenges including high latency, excessive memory usage, and complex infrastructure requirements. By implementing a cache-augmented architecture that preloads relevant knowledge and eliminates real-time retrieval, our approach significantly improves response time while reducing resource requirements. The research demonstrates the effectiveness of CAG through a comprehensive implementation for The University of Southern Mississippi's chatbot system, achieving a 49.02% improvement in response time compared to traditional RAG approaches. …

What References Are Chatgpt, Gemini, Copilot, And Perplexity Providing For Consumer Health Questions?, Scott Johnson, Catherine Johnson, Ivan Portillo

Library Presentations, Posters, and Audiovisual Materials

Background

With the growing popularity of generative artificial intelligence (AI) models such as ChatGPT, consumers may turn to these tools to easily seek health information. To our knowledge, no study has analyzed the references provided by multiple models for consumer health questions.

Objective

We aimed to analyze the references provided by ChatGPT, Gemini, Copilot, and Perplexity for consumer health questions in order to determine the most frequently appearing references.

Methods

AI generative models ChatGPT 4.0, Google Gemini, Microsoft Copilot, and Perplexity were each asked 30 consumer health questions and prompted to provide the corresponding references. The references were recorded.

The …

A Machine Learning Framework For Packet Anomaly Detection In Smartgrid Substation Networks, Sowmya Bandari

School of Computing: Dissertations, Theses, and Student Research

The increasing reliance on Smart Grid Substation Networks for efficient electricity distribution has amplified cybersecurity vulnerabilities, particularly within Supervisory Control and Data Acquisition (SCADA) systems. The IEC 60870-5-104 (IEC-104) protocol, widely adopted for communication between Remote Terminal Units (RTUs) and Human-Machine Interfaces (HMIs), lacks inherent encryption and authentication mechanisms, rendering it susceptible to sophisticated cyberattacks. Threats such as False Data Injection Attacks (FDIAs), command injection, covert attacks and replay attacks pose significant risks by manipulating grid control signals, potentially leading to undetected operational disruptions, cascading failures, or system-wide instability. Conventional signature-based Intrusion Detection Systems (IDS) often fail to identify zero-day …

Simulating Chill: Exploring The Cognitive And Therapeutic Potential Of Cold Vr Environments, Jessica Turner, Piper Hutson, James Hutson

Faculty Scholarship

This study investigates the cognitive and therapeutic potential of immersive virtual reality (VR) environments designed to simulate cold conditions. Through the engagement of participants through multisensory stimuli—including vivid visual representations of the Athabasca Glacier, auditory effects of icy winds, and corresponding haptic feedback—the research evaluates neurological and physiological responses associated with attention, emotional regulation, and stress modulation. Participants experienced virtual scenarios featuring icy winds and snow, activating specific neurological pathways involving the occipital lobe, primary visual cortex, superior colliculus, and insula, thus reinforcing sensory integration. Through predictive coding, the anterior insula and hypothalamus were engaged, prompting thermoregulatory simulations and subconscious …

1785 Parisian Salon Project: New Methods And Practical Innovations In Digital Heritage Reconstruction, James Hutson, Charles O'Brien, Wesley Wolfe, Kayla Kraff, Trent Olsen

Faculty Scholarship

The 1785 Salon Unreal Engine Reconstruction Project represents a significant advance in digital heritage and immersive art historical research by combining generative AI-based asset creation, modular user experience (UX) design, and historically informed workflows. During the Spring 2025 phase, the project achieved major milestones, including the successful development of a replicable pipeline for transforming 2D reference images into period-accurate 3D sculpture models using generative AI and digital sculpting tools. Simultaneously, a robust and adaptable Inspection System was engineered within Unreal Engine, offering granular interaction controls, bilingual (English/French) content integration, dynamic metadata display, and enhanced accessibility. These innovations collectively enabled historically …

Predicting Cardiac Resynchronization Therapy Response: Development And Validation Of A Single Photon Emission Computed Tomography-Based Nomogram, Zhongwei Jiang, Zhongqiang Zhao, Zhuo He, Qiushi Chen, Ju Bu, Chunxiang Li, Dianfu Li, Chang Cui, Weihua Zhou, Huiyuan Qin, Cheng Wang

Michigan Tech Publications

Background: Cardiac resynchronization therapy (CRT) is an effective treatment for patients with drug-refractory heart failure. However, more than thirty percent of patients do not benefit from CRT. This study aimed to develop and validate a novel model based on single photon emission computed tomography (SPECT) phase analysis features to predict CRT response. Methods: We identified 163 CRT patients who received gated resting SPECT myocardial perfusion imaging (MPI) between 2010 and 2020 at The First Affiliated Hospital of Nanjing Medical University. All variables were first processed by univariate logistic regression, and those with a P value < 0.05 were retained. The selected variables were subsequently used in the least absolute shrinkage and selection operator (LASSO) regression to construct a predictive model, which was then represented as a nomogram. Nomogram performance was assessed via receiver operating characteristic (ROC) curves, calibration curves, and decision curve analyses (DCAs). Internal validation was performed by bootstrapping with 1,000 replicates. Results: Of the 163 patients, 93 (57.1%) responded to CRT during follow-up. Responders had a wider QRS complex duration (QRSd) (164.80 vs. 154.51 ms, P=0.003), fewer premature ventricular contractions (PVCs) (1,392.98 vs. 2,283.60, P=0.003), lower prevalence of non-sustained ventricular tachycardia (NS-VT) (45.2% vs. 77.1%, P< 0.001), and better cardiac function [based on N-terminal pro-B-type natriuretic peptide (NT-proBNP), New York Heart Association (NYHA), and left ventricle (LV) parameters] compared to non-responders. Univariate logistic regression revealed 14 variables significantly associated with CRT response (all P< 0.05). The area under the ROC curve (AUC) value for the nomogram was 0.845 [95% confidence interval (CI): 0.785–0.906; sensitivity: 0.771; specificity: 0.849]. Internal validation yielded a mean AUC of 0.814 (95% CI: 0.777–0.836). The calibration curve demonstrated strong consistency between the predicted and observed outcomes. DCA revealed that the nomogram consistently provides a net benefit over the baseline, demonstrating its high practical value in clinical decision-making. A web-based dynamic nomogram (https:// jzw20000624.shinyapps.io/CRTpredictionmodel/) was developed for clinical application. Conclusions: We developed and validated a SPECT-based prediction model for predicting CRT response, which can assist clinicians in optimizing CRT candidacy preoperatively. Pacing at the latest contraction and relaxation segments, while avoiding scarred regions and optimizing preoperative status, is anticipated to improve CRT response.

Comparing Spatial Interfaces Of The Tower Of London Task, Paean Luby

Honors Theses

Executive functioning involves key mental skills like self-control and problem-solving, which are often impaired by brain injuries. The Tower of London (TOL) task is a problem set commonly used to assess planning, but both traditional and digital versions can lack consistency, and, in the case of digital versions, realism and physical engagement. Research shows that 3D tasks, such as a 3D version of the Tower of Hanoi, engage the brain in distinct and meaningful ways, likely due to increased spatial involvement. By merging immersive 3D environments with the consistency of digital tools, virtual reality (VR) has the potential to enhance …

Privacy-Aware Ai-Based Agricultural Monitoring Using Internet Of Drones, Md Benozir Hossain

Honors Theses

Artificial Intelligence (AI) has become a vital tool for agricultural farming. AI-based image processing models utilizing different machine learning (ML) algorithms and deep learning (DL) offer advanced functionalities in disease detection, yield estimation, land use, etc. This thesis examines AI-driven techniques utilizing Convolutional Neural Networks (CNN) with the addition of Federated Learning (FL) to analyze satellite and drone images for agricultural insights, especially in detecting Cotton diseases. The AI models improve agricultural farming in many ways, such as using data to make critical decisions, reducing labor costs, pest infestations, etc. Moreover, these models allow farmers to minimize yield losses by …

Effects Of Reflective Journaling And Nudges On Academic Motivation And Engagement, Lakshmi Satya Sai Veda Mahita Uppuluri

Theses and Dissertations

This study investigates the effects of reflective journaling and motivational nudges on academic motivation and engagement among college students. Grounded in Self- Determination Theory (SDT), the research examines how different interventions influence intrinsic motivation, and academic behaviors such as class attendance, participation, and preparation. The study employed a between-group experimental design with three conditions: a control group, a journaling group, and a journaling group that also received daily motivational nudges. Results showed that students in the journaling groups—particularly those who received nudges—experienced a significant increase in academic motivation. While changes in academic engagement were not significant, the effect size suggested …

Expressive And Interpretable User Engagement Prediction Using Multivariate Survival Processes, Akshay Aravamudan

Theses and Dissertations

The ability to characterize how information diffuses online is of paramount importance to stakeholders that are interested in tasks such as proposing solutions for mitigating and countering dis/misinformation, predicting user engagement of content in social media, planning marketing campaigns to roll-out products and planning dissemination of political campaign messaging among others. One such facet of learning the dynamics of information diffusion is the ability to predict user engagement or the popularity of a single piece of information as it spreads through an online medium. Existing works in this regard mainly either obfuscate user level information or utilize frameworks that are …

Exploring Educational Affordances Of Interactive Murals, Alyshia Bustos

Computer Science ETDs

Interactive murals are a new technology that blend traditional mural painting and embedded electronics. My work contrasts traditional STEAM projects by introducing youth to programming, building electronics, and interaction design principles within the context of designing and constructing interactive murals. Accordingly my dissertation addresses the following: 1) How can we integrate traditional mural practices and ubiquitous computing in an interactive mural, 2) How do interactive murals support youth in learning how to program and build electronics, and 3) How can we design learning activities that support youth as interaction designers?

To address my research questions, I first conducted technical tests …

Physics Embedded Neural Network: A Novel Data-Free Numerical Method For Solving Computational Physics Problems, Pawan Gaire

Dissertations and Doctoral Documents, University of Nebraska-Lincoln, 2023–

A novel approach for solving partial differential equations (PDEs) using neural networks for scientific computing is introduced. The proposed approach, referred to as physics-embedded neural network (PENN), features a unique architecture that incorporates the PDE and boundary conditions information directly within the final fully-connected layer of the feed-forward neural network (NN). The key aspect of PENN is the parallel numerical embedding of a differential equation associated with physical problems within the activation function of the network’s final layer. This integration leads to a new class of computational solvers competitive with classical methods like the Finite Element Method (FEM) and capable …

5g Network Slice Vulnerabilities And Exploit Chaining Through Enablers, John Breeden

Electronic Theses and Dissertations

Network slicing provides fundamental support for the enhanced features of 5G. Network slicing enablers, such as software-defined networking and network function virtualization facilitate the separation of the physical distributed infrastructures and the functions that create isolated slices. With this framework, the network slice is no longer under the control of a single entity. Multiple infrastructure providers share responsibility for the slice. The 5G architecture derives from multiple services, distributed over great distances, and managed by multiple parties. Each enabler and provider adds vulnerability to network slicing. We examine the interweaving of these enablers to identify vulnerabilities, discuss potential mitigation, and …

Radar Precursors To Severe Weather Reports In Left-Moving Supercells, Eric A. Carothers

Department of Earth and Atmospheric Sciences: Dissertations, Theses, and Student Research

While much research has examined dual-polarimetric signatures of right-moving supercells, very little has been done with left-moving supercells. Given that left-moving supercells are thought to be disproportionate producers of large hail, understanding their internal dynamics is vitally important. This study examines differences and trends in the dual-polarimetric signatures of left-moving supercells to identify precursors to severe weather reports. A dataset of left-moving supercells associated with severe weather reports was created. These storms are processed with an automated analysis algorithm that identifies and quantifies the polarimetric signatures in each storm. A method for analysis of differences and trends in their dual-polarization …

Understanding Automation From A Computer Science Perspective, Matthew Donsig

Honors Program: Senior Projects (Public)

This thesis looks into automation and analyzes its benefits and problems. It begins with an explanation of a capstone project, automating the UNL State Museum’s reservation system. Problems of automation are presented in unsuccessful attempts and some pitfalls of automation. Next this thesis turns to an examination of artificial intelligence in automation. Along with that, we look at bias in automation and how people can bias automated tools or be biased by them. Turning successful examples of automation and then automation in manufacturing shows its benefits. Automation creates new jobs or changes work as much as it eliminates positions. At …

Cyber Resiliency Framework And Mechanisms For Software Defined Tactical Networks, Anthony Castanares

Open Access Theses & Dissertations

Traditional tactical networks fail to achieve cyber resiliency for many reasons, but the most prevalent causes include flat designs, the absence of cyber detection capabilities at the lowest level, and immutable resource allocations after instantiation. These design choices allow network threats direct visibility to each device on the network, and the lack of detection allows infections to proliferate. Furthermore, tactical battlefield networks are difficult to secure because of the lack of persistent oversight by an intelligent agent that can exercise control over the network's topology or resources in real-time. However, advances in software-defined networking (SDN) provide an opportunity to address …

Generative Ai For 3d Printed Antenna Design, Jennifer Ann Chavez

Open Access Theses & Dissertations

This research explores the integration of generative artificial intelligence (AI) with a physics-informed particle swarm optimizer (PSO) to develop 3D printable microstrip patch antennas. A neural network was trained on a dataset of microstrip patch antenna geometries and their corresponding performance metrics: return loss and gain. The PSO used a fitness function prioritizing low return loss in potential antennas, eventually yielding novel antenna geometries with parasitic components. 3D printing constraints were also hard coded into the framework, thus preventing any geometries being generated that cannot be fabricated. When simulated using Ansys HFSS, the AI generated microstrip patch antennas exceeded the …

Hidden Layer Reshaping On Convolutional Neural Network, Alan Delgado

Open Access Theses & Dissertations

Artificial Intelligence (AI) technologies have become really popular in recent years. From ChatGPT to Tesla cars, many applications can benefit from these type of technologies. Automotive, healthcare, biomedical, cybersecurity, finances, and retail are some of the fields that take advantage of it. It has been seen that AI can solve complex problems, but there is still work to be done to optimize it. A deep learning neural network (DLNN) tries to simulate how a human brain operates. These DLNNs are made up of artificial neurons which are connected by weight that are modified when the network is trained. These networks …

From Text To Utility: Distance-Aware Contrastive Learning For Detection-Ready And Shareable Malware Descriptions, Ivan Alejandro Montoya Sanchez

Open Access Theses & Dissertations

The rapid rise of sophisticated malware variants poses significant challenges for cybersecurity analysts, particularly due to the scarcity of data on newly emerging threats. Due to privacy, legal, and operational constraints, malware samples are often not shareable; instead, organizations publish cyber threat intelligence (CTI) in natural language. However, these reports are typically unstructured and inconsistent, limiting their utility in machine learning (ML) models. This thesis explores whether high-fidelity, shareable threat intelligence can be automatically generated from structured malware behaviors to supplement ML models when direct access to malware samples is limited. Two central questions are addressed:(i) How can descriptions be …

Enhancing Security And Resiliency In Operational Technology Environments Through Network Slicing And Federated Learning, Brian Giovanni Rodiles Delgado

Open Access Theses & Dissertations

The growing convergence of Information Technology (IT) and Operational Technology (OT) within Industry 4.0 environments has introduced new demands on industrial network infrastructure. As cyber-physical systems become increasingly interconnected, ensuring the secure, timely, and efficient exchange of critical data is essential. This thesis explores how network slicing, a method of creating isolated virtual network segments, can be applied within OT environments to address challenges such as latency, security, and resource allocation.

The first research question addressed in this thesis is: How can OT networks take advantage of NFV and SDN technology to become cyber resilient? This study examines the operational, …

Machine Learning And Time Series Forecasting For Hydropower Predictions, Jose Reynaldo Vega

Open Access Theses & Dissertations

Recent advancements in machine learning have led to the design of many neural network architectures aimed at solving real-world problems. Each network works to make predictions by finding patterns in the provided data. One common application is time series forecasting, where a model predicts future events based on historical time series data. Time series forecasting is used in a variety of fields, one example being in predicting water releases of Hybrid Floating Photovoltaic-Hydropower (HFPVH) systems. As global population growth drives an increase in energy demand, the need for resilient and sustainable energy generation has become urgent. HFPVH systems have emerged …

Alphamissense Predictions And Clinvar Annotations: A Deep Learning Approach To Uveal Melanoma, David J. Taylor Gonzalez, Mak B. Djulbegovic, Meghan Sharma, Michael Antonietti, Colin K. Kim, Vladimir N. Uversky, Carol L. Karp, Carol L. Shields, Matthew W. Wilson

Wills Eye Hospital Papers

OBJECTIVE: Uveal melanoma (UM) poses significant diagnostic and prognostic challenges due to its variable genetic landscape. We explore the use of a novel deep learning tool to assess the functional impact of genetic mutations in UM.

DESIGN: A cross-sectional bioinformatics exploratory data analysis of genetic mutations from UM cases.

SUBJECTS: Genetic data from patients diagnosed with UM were analyzed, explicitly focusing on missense mutations sourced from the Catalogue of Somatic Mutations in Cancer (COSMIC) database.

METHODS: We identified missense mutations frequently observed in UM using the COSMIC database, assessed their potential pathogenicity using AlphaMissense, and visualized mutations using AlphaFold. Clinical …

Ai Model For Predicting Asthma Prognosis In Children, Elham Sagheb, Chung-Il Wi, Katherine S King, Bhavani Singh Agnikula Kshatriya, Euijung Ryu, Hongfang Liu, Miguel A Park, Hee Yun Seol, Shauna M Overgaard, Deepak K Sharma, Young J Juhn, Sunghwan Sohn

Faculty, Staff and Student Publications

BACKGROUND: Childhood asthma often continues into adulthood, but some children experience remission. Utilizing electronic health records (EHRs) to predict asthma prognosis can aid health care providers and patients in developing effective prioritized care plans.

OBJECTIVE: We aimed to develop artificial intelligence (AI) models using various clinical variables extracted from EHRs to predict childhood asthma prognosis (remission vs no remission) in different age groups.

METHODS: We developed AI models utilizing patients' EHRs during the first 6, 9, or 12 years of their lives to predict their asthma prognosis status at ages 6 to 9, 9 to 12, or 12 to 15 …

Exploring The Biasing Effects Of Gender On Personality Disorder Diagnoses Formulated By Artificial Intelligence, Zoe Colclough

Student Theses

Gender bias is prevalent in personality disorder assessments, and while artificial intelligence has been posited as a solution to improve diagnostic objectivity and accuracy, the potential for such technologies to propagate human gender bias in mental health contexts remains underexplored. This study investigated the influences of gender bias on the diagnostic performance of ChatGPT-4o for personality disorders using three factorial research designs, which involved experimentally manipulating patient gender in a combined sample of 360 vignettes and case studies. Vignettes were synthesized through a novel artificial intelligence-assisted methodology established for this research, and case studies were identified from the literature. Significant …

Advancing Precision And Autonomy In Agriculture And Medical Imaging Through Ai And Computer Visions, Tan-Hanh Pham

Theses and Dissertations

Deep learning has revolutionized numerous fields by enhancing precision, automation, and decision-making capabilities. This dissertation explores its applications in agriculture and medical image processing, introducing novel methodologies to improve accuracy and efficiency in these domains. These fields hold critical societal importance -- agriculture underpins global food security and sustainability, while medical imaging drives advancements in diagnostics and personalized healthcare, both benefiting significantly from data-driven innovations. In agriculture, deep learning is applied to precision spray systems through droplet analysis. Specifically, a generative model is designed to create synthetic droplet images, addressing the challenge of limited training samples, which are expensive …

Stability Analysis Of Turbulent Fluid Flow, Adam D. Schroeder

Mathematics, Statistics, and Computer Science Honors Projects

Hydrodynamic stability refers to the study of when and how laminar flows transition to turbulence. This includes investigations of the mechanisms of transition, as well as the classification of known flow configurations as either stable or unstable and the identification of critical values of flow parameters at which this bifurcation occurs. In this thesis, we introduce the mathematical theory behind continuum mechanics and fluid dynamics as well as some tools from the study of dynamical systems. We apply these concepts to the linear stability analysis of zero pressure gradient flat plate flow via numerical simulations in OpenFOAM, discussing both the …

Generalizable Skill Learning In Robotic Agents Using Transformer Models, Erik Enriquez

Theses and Dissertations

This work explores the application of Transformer models to robotic skill learning, aiming to enhance generalization across various physical tasks and environments with continuous control. Despite their success in other domains, our experiments reveal that the utility of Transformers in robotics heavily depends on pretraining strategies. Specifically, Transformers pretrained on reinforcement learning tasks generalized effectively, while those trained with task-agnostic masking strategies did not. These findings challenge assumptions about the universality of Transformer-based methods and underscore the importance of domain-aligned pretraining for developing versatile robotic agents.

Fixdrive: Automatically Repairing Autonomous Vehicle Driving Behaviour For $0.08 Per Violation, Yang Sun, Christopher M. Poskitt, Kun Wang, Jun Sun

Research Collection School Of Computing and Information Systems

Autonomous Vehicles (AVs) are advancing rapidly, with Level-4 AVs already operating in real-world conditions. Current AVs, however, still lag behind human drivers in adaptability and performance, often exhibiting overly conservative behaviours and occasionally violating traffic laws. Existing solutions, such as runtime enforcement, mitigate this by automatically repairing the AV's planned trajectory at runtime, but such approaches lack transparency and should be a measure of last resort. It would be preferable for AV repairs to generalise beyond specific incidents and to be interpretable for users. In this work, we propose FixDrive, a framework that analyses driving records from near-misses or law …

Decictor: Towards Evaluating The Robustness Of Decision-Making In Autonomous Driving Systems, Mingfei Cheng, Xiaofei Xie, Yuan Zhou, Junjie Wang, Guozhu Meng, Kairui Yang

Research Collection School Of Computing and Information Systems

Autonomous Driving System (ADS) testing is crucial in ADS development, with the current primary focus being on safety. However, the evaluation of non-safety-critical performance, particularly the ADS's ability to make optimal decisions and produce optimal paths for autonomous vehicles (AVs), is also vital to ensure the intelligence and reduce risks of AVs. Currently, there is little work dedicated to assessing the robustness of ADSs' path-planning decisions (PPDs), i.e., whether an ADS can maintain the optimal PPD after an insignificant change in the environment. The key challenges include the lack of clear oracles for assessing PPD optimality and the difficulty in …

Hello Again! Llm-Powered Personalized Agent For Long-Term Dialogue, Hao Li, Chenghao Yang, An Zhang, Yang Deng, Xiang Wang, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Open-domain dialogue systems have seen remarkable advancements with the development of large language models (LLMs). Nonetheless, most existing dialogue systems predominantly focus on brief single-session interactions, neglecting the real-world demands for long-term companionship and personalized interactions with chatbots. Crucial to addressing this real-world need are event summary and persona management, which enable reasoning for appropriate long-term dialogue responses. Recent progress in the human-like cognitive and reasoning capabilities of LLMs suggests that LLM-based agents could significantly enhance automated perception, decision-making, and problem-solving. In response to this potential, we introduce a model-agnostic framework, the Long-term Dialogue Agent (LD-Agent), which incorporates three independently …