Digital Commons Network^™

Advancing Multi-Modal Learning, Swetha Sirnam

Graduate Studies Theses and Dissertations 2026

Multimodal learning aims to develop systems that understand and reason across multiple sources of information such as images, videos, audio, and text. These capabilities are central to applications including visual search, digital assistants, robotics, and visual question answering. Despite recent progress in multimodal large language models, current approaches struggle to capture the structured complexity of real-world visual data. Fine-grained visual details are often lost during multimodal alignment, models frequently fail to capture nuanced human intent, and existing methods remain limited in understanding long-form videos with complex temporal dynamics. Furthermore, many evaluation benchmarks rely heavily on explicit visual cues, which can …

Rrto: A High-Performance Transparent Offloading System For Model Inference In Mobile Edge Computing, Zekai Sun, Xiuxian Guan, Zheng Lin, Yuhao Qing, Haoze Song, Zihan Fang, Zhe Chen, Fangming Liu, Heming Cui, Wei Ni, Jun Luo

Research outputs 2022 to 2026

Deploying Machine Learning (ML) applications on resource-constrained mobile devices remains challenging due to limited computational resources and poor platform compatibility. While Mobile Edge Computing (MEC) offers an offloading-based inference paradigm using GPU servers, existing approaches can be divided into non-transparent and transparent methods, with the former necessitating source-code modifications. Non-transparent offloading achieves high performance but requires intrusive code changes, limiting compatibility with diverse applications. Transparent offloading, in contrast, offers wide compatibility but introduces two challenges: performance degradation under MEC's low and fluctuating bandwidth, where per-operator remote procedure calls (RPCs) inflate end-to-end latency and energy consumption; and black-box logic reconstruction, which …

Dyno : Dynamic Neurosymbolic Orchestrator For Multi-Agent Systems, Ritvik Garimella, Chathurangi Shyalika, Renjith Prasad, Amit Sheth

Publications

Large Language Model (LLM)-based multi-agent systems (LaMAS) represent an emerging paradigm for tackling complex, multi-step reasoning and decision-making problems. As these systems scale, orchestration, which is the ability to coordinate, manage, and evaluate the interactions among diverse agents, becomes central to their success. While recent orchestrators such as AgentFlow have demonstrated promise in managing communication and task delegation, they remain limited in their ability to understand task semantics, coordinate heterogeneous agent types (e.g., reactive vs. cognitive), and adaptively align outputs with human-defined goals. In this position paper, we introduce the DYNO (Dynamic Neurosymbolic Orchestrator), a system developed as part of …

Geovig And Purevig: Geometry-Aware Architectures For Efficient Computer Vision, Omar Ismail

Theses and Dissertations (Comprehensive)

Deploying deep learning models for medical image analysis on mobile devices requires a balance between inference latency, memory footprint, and delineating anatomical boundaries with high accuracy. While Convolutional Neural Networks (CNNs) and mobile Vision Transformers (ViTs) offer efficiency, they often struggle to model the irregular, non-local geometric structures inherent in biological tissues without incurring prohibitive computational costs. In this thesis, we introduce GeoViG (Geometric Vision Graph), an architecture that bridges the gap between efficient grid-based processing and explicit Geometric Deep Learning. GeoViG introduces a novel transition from high-resolution pixel grids to low-resolution dynamic graphs via a SpreadEdgePool operator, a geometry-aware …

Advancing Cybersecurity Through Userland Memory Forensics: From Runtime Analysis To Security Applications, Hala Ali

Theses and Dissertations

Memory forensics has become a crucial component of digital investigations, particularly for detecting malware operating solely in system memory. As operating system vendors implemented kernel access restrictions, malware authors shifted to userland malware. However, existing memory forensics techniques have largely focused on kernel-level analysis, leaving userland runtimes insufficiently covered. This dissertation addresses this gap by expanding memory analysis capabilities across two distinct paradigms: interpreted and compiled runtimes. The first phase targets the Python runtime, developing automated recovery techniques that enable several security applications. For malware detection, these techniques extract critical forensic artifacts such as encryption keys and command-and-control configurations. For …

The Lawyer Of The Future: Ethics And Identity In The Age Of Ai, David Yosifon, Michael Santoro, Isaac Nikssarian

American University Law Review

The dawn of artificial intelligence (AI) poses a fundamental challenge to the legal profession’s practical function and social identity. We argue that this challenge should be met as an opportunity to reclaim the profession’s core ethical intentions. We approach the emergence of AI as an occasion to define and safeguard what is irreducibly human in legal work. We ground our approach in the ethical imperatives set out in the American Bar Association’s Model Rules of Professional Conduct, while extending them into an invigorated framework for the AI era.

We survey the uses and ethical challenges associated with cutting-edge deployment of …

Safe Optimal Control Framework For Cooperative Manipulation Of Objects In Human–Robot Teams, Irfan Ganie, Sarangapani Jagannathan

Electrical and Computer Engineering Faculty Research & Creative Works

This article introduces a distributed deep neural network (NN)-based adaptive control framework for cooperative object manipulation in human–robot teams with unknown agent dynamics by using three distinct multilayer NN observers (MNNOs). The first observer, termed the reference point estimator, enables each robotic agent to estimate the object's reference center using consensus-based learning, even without direct access to global reference trajectories. The second observer, referred to as the human force-to-trajectory estimator, uses human-applied forces to infer the intended position, velocity, and acceleration of the object, enabling real-time estimation of human intent. Together, these two observers allow distributed estimation of human-intended motion. …

Self-Calibrating Uav Navigation: Reinforcement Learning Approaches For Horizontal Trajectory Estimation, Shirin Nasr-Esfahani, S. Jagannathan

Electrical and Computer Engineering Faculty Research & Creative Works

Accurate unmanned aerial vehicle (UAV) trajectory estimation is essential for autonomous navigation, particularly in GPS-denied environments. Visualodometry and simultaneous localization and mapping (SLAM) approaches require precise camera intrinsic parameters, which are typically obtained through predefined or offline calibration. Instead, in this work, we propose a reinforcement learning (RL)-based self-calibration framework that estimates camera intrinsic parameters directly from monocular video sequences, without requiring prior knowledge of the camera, environment, or calibration targets. This intrinsic parameter estimation is then leveraged to achieve robust UAV trajectory estimation using only video data. We formulate the problem as a sequential decision-making task, where an RL …

Rf-Attennet: A Hybrid Attention-Enhanced Network For Mixed Signal Classification In Uav Swarm Detection, Prajoy Podder, Mohammad Atikur Rahman, Maciej Zawodniok, Sanjay Madria

Electrical and Computer Engineering Faculty Research & Creative Works

The continuous increase of UAVs, particularly in swarms, creates significant challenges for security and airspace regulation. Traditional RF fingerprinting methods struggle to detect and classify UAV swarms due to overlapping signals and interference. This study introduces RF-AttenNet, a hybrid deep learning model designed to classify mixed UAV signals by analyzing composite RF spectrograms. RF-AttenNet uses dual attention mechanisms, channel and spatial attention to focus on critical spectral features, enabling the model to effectively separate and identify overlapping UAV signals. We have developed custom composite UAV datasets that simulate real-world swarm interference, incorporating both single and mixed UAV classes. RF-AttenNet achieves …

Fixed-Time Consensus Tracking For Nonlinear Multi-Agent Systems Under Aperiodically Intermittent Control, Lei Xue, Tong Wu, Wenwen Jia, Donald C. Wunsch

Electrical and Computer Engineering Faculty Research & Creative Works

This article explores the problem of fixed-time consensus tracking (FT-CT) for nonlinear multi-agent systems utilizing the a periodically intermittent control (AIC) strategy. In contrast to existing control algorithms, the proposed algorithm utilizes the AIC strategy instead of the conventional continuous-time control strategy, effectively reducing the consumption of communication resources. Moreover, the problem of intermittent FT-CT is well handled by proposing the average control rate of the AIC strategy. Two theorems based on the cases of directed and undirected graphs are proposed, respectively. Finally, the validity of these results is confirmed through numerical simulations on a general nonlinear system and a …

Nash Equilibrium Strategies For Multicluster Pursuit–Evasion Game With Disturbances: A Prescribed-Time Convergence Approach, Lei Xue, Xian Yu, Yongbao Wu, Jian Liu, Changyin Sun, D. C. Wunsch

Electrical and Computer Engineering Faculty Research & Creative Works

This article investigates the problem of prescribed-time Nash equilibrium (NE) seeking for a multicluster pursuit–evasion game (PEG) subject to external disturbances. To mitigate the impact of disturbances and reach the NE within a user-defined prescribed time, a prescribed-time disturbance observer (PTDO) is devised to estimate and compensate for them. Based on this observation, a novel control algorithm is developed, which facilitates collaboration among multiple pursuers to capture multiple evaders within the prescribed time. It is theoretically demonstrated that the designed algorithm ensures prescribed-time convergence to the NE of the multicluster PEG with disturbances. Finally, numerical simulations are conducted to verify …

Ai-Driven Real-Time Detection Of Zero-Day Browser Exploits Using Webassembly-Based Instrumentation, Temitope Damilola Elijah

College of Graduate Studies: Theses & Dissertations

The rapid evolution of web browsers into fully fledged application execution environments has significantly expanded their attack surface, making them prime targets for sophisticated zero-day exploits that evade traditional signature-based security mechanisms. To address this challenge, this research proposes an AI-driven framework for real-time detection and analysis of zero-day exploits in web browsers by integrating browser-level telemetry monitoring, unsupervised anomaly detection, and large language model–based threat interpretation. The framework introduces a lightweight WebAssembly telemetry agent embedded within the browser runtime to capture low-level execution behaviors, including WASM module instantiation, memory growth patterns, network interactions, and runtime API activity. These telemetry …

Swimming In Uncertainty: Filling Data Gaps And Providing An Educational Platform For Beach Water Quality At Tybee Island, Georgia, Lukas Roberson

College of Graduate Studies: Theses & Dissertations

@font-face {font-family:"Cambria Math"; panose-1:2 4 5 3 5 4 6 3 2 4; mso-font-charset:0; mso-generic-font-family:roman; mso-font-pitch:variable; mso-font-signature:-536870145 1107305727 0 0 415 0;}p.MsoNormal, li.MsoNormal, div.MsoNormal {mso-style-unhide:no; mso-style-qformat:yes; mso-style-parent:""; margin:0in; mso-pagination:widow-orphan; font-size:12.0pt; font-family:"Times New Roman",serif; mso-fareast-font-family:"Times New Roman";}.MsoChpDefault {mso-style-type:export-only; mso-default-props:yes; mso-font-kerning:0pt; mso-ligatures:none;}div.WordSection1 {page:WordSection1;}

Swimming in beaches water contaminated with high levels of bacteria can make you sick. Current monitoring at the public beaches on Tybee Island consists of weekly monitoring and enumeration of fecal indicator bacteria that takes 24 hours for results. If the number of bacteria exceed regulatory limits, a public health advisory is issued, and affected waters are retested until …

Security And Energy-Efficiency In Federated Learning, Priyesh Ranjan

Doctoral Dissertations

Federated Learning (FL), which facilitates collaborative model training and protects users' privacy, has drawn great interest from the research community. With FL, the participants train their models on local data and submit the corresponding updates for aggregation to a server. While concealing the participants' identities, FL may attract adversaries aiming to hamper the underlying model. These adversaries aim to submit malicious weight updates that corrupt the performance of the server model. Further, these models when communicated to the participating clients extend the behavior which is undesirable. Additionally, FL suffers from increased energy consumption at the edge device level due to …

Trust-Aware Algorithms For Stackelberg Ground Vehicle Routing, Doris Evelyn Meredith Brown

Doctoral Dissertations

Despite decades of research focused on reducing ground vehicle traffic congestion, urban traffic networks worldwide continue to experience traffic flows that lead to increased network travel times, largely resulting from the routing decisions of individual vehicles. To address this challenge, this work leverages a Stackelberg game framework to model the interaction between a vehicle agent and a routing authority as a leader–follower game, in which the routing authority proposes routing interventions to which the agent responds. This research contributes to existing traffic mitigation literature by exploring the role of trust in route decision-making and providing trust-aware algorithms that influence vehicle …

A Knowledge-Driven, Ai-Assisted Cyber Defence Framework For Iomt Remote Patient Monitoring, Kulsoom S. Bughio, David M. Cook, Abdul M. Unar

Research outputs 2022 to 2026

The rapid adoption of Internet Medical Things (IoMT) technologies in remote patient monitoring has reshaped healthcare delivery by enabling continuous, real-time clinical observation outside traditional care settings. However, this shift has also expanded the cyber-attack surface across heterogeneous, resource-constrained medical devices, wireless networks, cloud services, and third-party platforms. In cyber warfare, healthcare has become an incorporated target of geopolitics, with hospitals, remote monitoring systems, and emergency health systems being used to broaden the attack surface for adversaries to exploit. Existing security approaches for IoMT environments remain largely manual, fragmented, and reactive, limiting their effectiveness in dynamically assessing vulnerabilities and supporting …

Can An Experienced Qualitative Researcher Distinguish Ai From Human Qualitative Content Analysis?, Alexandra T. Lucas, Jianna Ramos, Maria Bajwa, Aaron Calhoun, Mark W. Scerbo, Janice C. Palaganas

Psychology Faculty Publications

Background

Artificial intelligence (AI) has become increasingly embedded in research workflows. Large language models (LLMs) are being used to code segments of text, organise codes into themes and interpret patterns within contexts. Recent comparisons between human and AI analyses demonstrate up to 80% thematic overlap, yet humans consistently exhibit deeper interpretive integration and contextual understanding. This study assesses whether experienced researchers can distinguish between entirely human-generated and AI-generated qualitative content analyses of a simulation debriefing.

Methods

We conducted a qualitative descriptive study comparing human-generated qualitative content analysis (QCA) with ChatGPT-4o-generated QCA using a single focus group transcript on emotion management …

A Unified Framework For Evaluating Training Efficiency In Deep (Bayesian) Neural Networks: Metrics, Overtraining, Stopping Criteria, And Grokking Computer Science, Eduardo Cueto Mendoza

Doctoral

Measuring training efficiency for artificial neural networks is an open research problem, current literature reports several attempts to define measures or create reporting frameworks. Current methods lack generality as they require measurements of the hardware or software thus, comparing efficiency between different systems can be difficult. Similarly, current metrics or frameworks generally do not propose the use of the metrics to directly improve training efficiency. This thesis presents three main contributions: (1) a novel framework that quantifies the training efficiency of a neural architecture on a learning task as the average ratio of model accuracy to total energy consumption during …

Motivation System Evoked By High Performance Work System In Chinese Universities And Colleges - An Examination Of The Relationship Between Human Resource Practices, Basic Psychological Needs, And Job Performance, Juan Chen

Doctoral

High Performance Work Systems (HPWS) are strategic, integrated Human Resource Practices (HRPS) designed to enhance employee abilities, motivation, and opportunities for participation, driving sustained organisational performance. While early research emphasized commitment HRPS, recent studies recognise that control and commitment HRPS are complementary rather than mutually exclusive, with neither holding an absolute advantage. Their success depends on contextual factors such as country, industry,organisation, economic environment, and workforce. In Chinese higher education institutions, transitioning from a high-commitment “iron rice bowl” system to a more control-oriented HR model has created tensions between efficiency and academic freedom. Based on the resource-based view, social exchange …

A Systematic Review And Characterization Of Privacy Noncompliance In Real-World Applications, Alexander E. Charkiewicz

Graduate Studies Theses and Dissertations 2026

Software applications increasingly rely on user data to provide their functionality, but improper handling of such data can lead to serious privacy noncompliance with applicable regulations and policies. A prominent example is the Facebook–Cambridge Analytica scandal, in which a third-party application collected the personal data of approximately 87 million Facebook users without users' consent. Despite growing attention to privacy compliance, two key challenges hinder the systematic understanding and analysis of privacy noncompliance. First, unlike security vulnerabilities, which have been systematically categorized through taxonomies such as the Common Weakness Enumeration (CWE), privacy noncompliance lacks a technical taxonomy describing how it manifests …

An Empirical Comparison Of K-Nearest-Neighbors And Logistic Regression Classification Models, Jackson Cushing

Graduate Studies Theses and Dissertations 2026

This thesis presents an empirical comparison of two classification methods: Logistic Regression and K Nearest Neighbors (KNN). The primary objective of this research is to evaluate the strengths and limitations of each method when applied to real-world datasets. Several publicly available datasets on diabetes, breast cancer, heart attack risk, and cardiovascular disease, were analyzed. For each dataset, K Nearest Neighbors models were implemented in the same way logistic regression had already been applied. The results demonstrate that while logistic regression offers interpretable parameter estimates and performs well when the underlying predictor and outcome relationship is approximately linear, however KNN can …

Event-Based Predefined-Time Synchronization For Complex Networks With Deception Attacks: An Asynchronously Intermittent Strategy, Lei Xue, Jiong Yu, Haoyu Zhou, Yongbao Wu, Jian Liu, Donald C. Wunsch

Electrical and Computer Engineering Faculty Research & Creative Works

This article studies the practical predefined-time synchronization (PPTS) for complex networks (CNs) under deception attacks based on the asynchronously intermittent event-triggered control (AIE-TC). Notably, AIE-TC effectively integrates the advantages of asynchronously intermittent control (AIC) and event-triggered control, where AIC provides each subsystem node with independent control and rest intervals. Besides, all synchronization errors of the CNs converge to an adjustable neighborhood within the predefined time by designing a bounded time-varying function into the controller. Moreover, this article considers that the transmission network is subjected to stochastic deception attacks modeled by a Markov process, which captures the state-driven dynamic transition characteristics …

An Incidental Standard For Medical Ai, Nicholson W. Price Ii

Articles

Medical AI is poised to make a major difference in the provision of health. It brings major challenges, though: how can developers and implementers ensure that it will work safely and effectively—especially within the context of complex and highly variable health-care systems? Standards provide one key tool, potentially providing guidelines for everything from privacy to accuracy to how AI interacts with human clinicians. This Essay considers the last of these, describing a powerful quasi-standard from a surprising source: an FDA guidance document that tells developers when certain AI systems are not considered medical devices, and are therefore not regulated by …

A General Algorithm For Assortment Optimization Under Random Utility Choice Models, Tien Mai, Andrea Lodi

Research Collection School Of Computing and Information Systems

This work concerns the assortment optimization problem that refers to selecting a subset of items that maximizes the expected revenue in the presence of the substitution behavior of consumers specified by a random utility choice model. The key challenge lies in the computational difficulty of finding the best subset solution, which often requires exhaustive search. The literature on constrained assortment optimization lacks a practically efficient method that is general to deal with different types of customer choice models (e.g., the multinomial logit, mixed logit or general multivariate extreme value models). In this work, we propose a new approach that allows …

Decentralized Linear Solvers: Communication Cost And Privacy, Nelson G. Brasil, Vinay A. Vaishampayan

Publications and Research

We consider the problem of multiple parties iteratively solving a system of linear equations in a decentralized manner. Specifically, we solve for $\negr{x} \in \R^n$, the $n \times n$ system of linear equations $M\negr{x} = \negr{b}$ when each party only knows their row of the matrix $M$ and a single component of $\negr{b} \in \R^n$. Our objective is to determine the tradeoff between the accuracy of the solution and the total communication cost measured in bits. A fully connected, reliable mesh network is assumed to connect the different parties. We develop a general formulation that applies to a large class …

Large Language Model-Assisted Research Question Development In Public Health: A Case Study In The Special Supplemental Nutrition Program For Women, Infants, And Children (Wic), Qi Zhang, Bidusha Neupane, Priyanka Patel, Futun N. Alkhalifah, Yi He, Leslie Hodges

Health Behavior, Policy & Management Faculty Publications

Objective:

To assess the feasibility of using large language models (LLMs) to develop research questions about changes to the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) food packages.

Design:

We conducted a controlled experiment using ChatGPT-4 and its plugin, MixerBox Scholarly, to generate research questions based on a section of the USDA summary of the final public comments on the WIC revision. Five questions weekly for three weeks were generated using LLMs under two conditions: fed with or without relevant literature. The experiment generated 90 questions, which were evaluated using the FINER criteria (Feasibility, Innovation, Novelty, Ethics, …

The Crucial Role Of Machine Learning Models In Predicting Current Childhood Asthma: Model Comparison, Calibration, And Shap-Based Interpretation, Aditya Chakraborty, A. K.M. Raquibul Bashar

Epidemiology, Biostatistics, & Environmental Health Faculty Publications

Background: Asthma is one of the most prominent chronic diseases in children and one of the most challenging ailments to diagnose in infants and preschoolers in the United States. Predictive models can be instrumental in improving early diagnosis, personalized treatment strategies, and disease progression. By utilizing nationalized data, this study focuses on building and comparing high-performing analytical predictive models based on the relevant risk factors and identifying the most influential predictors.

Methods: We analyzed cross-sectional BRFSS Asthma Call-Back Survey data (2011-2020; N = 9,813) and randomly split participants into training and testing sets. An XGBoost model (hyperparameters tuned via grid …

Pahina: Precision-Aware Hierarchical In-Network Aggregation For Edge Distributed Training, Yingpu Nian, Bo Yi, Qiang He, Xingwei Wang, Geyong Min, Keqin Li, Sajal K. Das

Computer Science Faculty Research & Creative Works

The rise of edge intelligence is driving distributed machine learning toward a new paradigm of edge-collaborative computing. To overcome the severe communication bottleneck in this paradigm, In-Network Aggregation is a critical enabling technology. However, its effectiveness is fundamentally undermined by the profound resource heterogeneity of edge networks. Specifically, edge devices, adapting to hardware constraints, operate at varying numerical precisions, leading to significant data inflation as gradients are aggregated. Compounding this, unevenly distributed network resources and traditional, precision-oblivious routing strategies often misallocate critical, high-precision gradients to low-quality paths. This mismatch creates severe network congestion, crippling the efficiency of distributed training. To …

Future Mining: Learning For Safety And Security, Md Sazedur Rahman, Mizanur Rahman Jewel, Sanjay Madria

Computer Science Faculty Research & Creative Works

Mining industry is rapidly transforming into an AI-driven cyber-physical ecosystem where safety and operational reliability depend on robust perception, resilient communication, trustworthy distributed intelligence and continuous monitoring of miners and equipment. Real-world mining environments impose severe constraints like poor illumination, dust, occlusion, GPS-denied conditions, irregular underground topologies, and intermittent connectivity. These factors degrade perception quality, disrupt situational awareness, impair trajectory prediction and weaken the reliability of distributed learning systems. Emerging cyber-physical threats, including backdoor triggers, sensor spoofing, label-flip attacks and poisoned model updates, further jeopardize operational safety, particularly as mines increasingly adopt autonomous vehicles, humanoid assistance, and federated learning for …

Qura: Reinforcement Learning Based Routing For Quantum Networks, Tasdiqul Islam, Engin Arslan, Md Arifuzzaman

Computer Science Faculty Research & Creative Works

Quantum routing deals with identifying a set of quantum repeaters to use to create entanglement between distant endpoints. Previous approaches proposed shortest-path and linear programming methods to find a solution to this problem. While the shortest path approach results in suboptimal performance, linear programming takes too long to find a solution as the network size and constraints increase. In this paper, we apply Deep Q-Reinforcement Learning (DQRL) to optimize routing in quantum networks both in terms of execution time and performance. The proposed Quantum Routing Algorithm (QuRA) first chooses which request to schedule among all requests. It then determines which …