Computer Sciences Commons^™

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 …

Improving Medical Diagnostics With Vision-Language Models: Convex Hull-Based Uncertainty Analysis, Ferhat Ozgur Catak, Murat Kuzlu, Taylor Patrick, Michel Audette

Engineering Technology Faculty Publications

In recent years, vision-language models (VLMs) have been applied to various fields, including healthcare, education, finance, and manufacturing, with remarkable performance. However, concerns remain regarding VLMs' consistency and uncertainty, particularly in critical applications such as healthcare, which demand a high level of trust and reliability. This paper proposes a novel approach to evaluate uncertainty in VLMs' responses using a convex hull approach on a healthcare application for visual question answering (VQA). For any VLM, temperature refers to a sampling parameter used in probabilistic generation, which controls the randomness of the model's output. The LLM-CXR model is selected as the medical …

Machine Learning Classification Of Prostate Cancer Genomic Sequences Using K-Mer And Sequence-Derived Features, Kuldeep Rawat, Hirendra Nath Banerjee, Jamie Noble, Saa Naudia Deloatch, Satyendra Banerjee, Sachin Shetty, Soumya Banerjee

VMASC Publications

Prostate cancer disproportionately impacts African American men, who experience significantly higher mortality rates and earlier disease onset than other populations. Current diagnostic approaches, including prostate-specific antigen testing and biopsy, lack sufficient specificity and sensitivity, underscoring the need for accurate, molecular-level classification tools. This paper presents a machine learning framework for binary classification of genomic DNA sequences as cancerous or healthy. A dataset of 1684 FASTA-formatted sequences obtained from the National Library of Medicine - GenBank was analyzed, with 1662 sequences retained after quality control filtering. Feature engineering yielded 67 attributes, including GC content, Shannon entropy, sequence length, and trinucleotide k-mer …

Beyond Full Fine-Tuning: The New Playbook For Adapting Deep Neural Networks, Cristian S. Mcgee

Honors Undergraduate Theses

Fine-tuning is the process of teaching and specializing a pre-trained neural network on a downstream task. Fine-tuning is a rapidly growing topic in artificial intelligence domains; however, many fine-tuning endeavors are highly specialized without a coherent framework connecting them. This work presents a unified perspective on fine-tuning methods and performance metrics. Our perspective organizes the methods in terms of how they are applied to fine-tuning. This framework showcases methods that (i) update effective subspaces of the pre-trained model, (ii) change the adaptation optimization procedure, and (iii) alter the representations of the embedded input. Additionally, we present unconventional metrics such as …

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 …

Law Librarianship And Legal Information Science In The Age Of Genai, Paul D. Callister

Faculty Works

This article examines the relationship between law librarianship and legal information science in the age of generative AI (GenAI), arguing that closer integration between the two is essential to navigating a rapidly evolving legal information landscape. It contends that law librarianship—long grounded in stable classification systems and cognitive authority—must adopt the analytical methods of legal information science to remain effective in the digital era. Together, these fields can reinforce the rule of law by improving the organization, retrieval, and stability of legal information. The article identifies emerging subfields of legal information science that support this integration and develops several concepts …

An Empirical Framework For Evaluating Semantic Preservation Using Hugging Face, Nan Jia, Anita Raja, Raffi Khatchadourian

Publications and Research

As machine learning (ML) becomes an integral part of high-autonomy systems, it is critical to ensure the trustworthiness of learning-enabled software systems (LESS). Yet, the nondeterministic and run-time-defined semantics of ML complicate traditional software refactoring. We define semantic preservation in LESS as the property that optimizations of intelligent components do not alter the system's overall functional behavior. This paper introduces an empirical framework to evaluate semantic preservation in LESS by mining model evolution data from HuggingFace. We extract commit histories, $\textit{Model Cards}$, and performance metrics from a large number of models. To establish baselines, we conducted case studies in three …

Llm-Assisted Legal Propositions Identification From Party Arguments In The U.S. Supreme Court Briefs, Heng Zheng, Alex Zhang

Faculty Scholarship

Merits briefs are central to U.S. litigation, serving as the primary means for parties to present arguments and persuade judges. Legal propositions in these merits briefs are the atomic units of arguments, whose relationships evolve throughout litigation and inform court decisions and precedent. Large language models (LLMs) have been applied to legal document review, but there is limited evidence on their ability to identify legal propositions in merits briefs. Given the laborintensive nature of the task, we evaluate a human-AI collaborative approach to identifying legal propositions in the U.S. Supreme Court merits briefs, in which legal annotators review and revise …

Inside Out: Improving Large Model Safety, Wei Zhao

Dissertations and Theses Collection (Open Access)

While Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) are at the frontier of current advancements in artificial intelligence, demonstrating remarkable capabilities across diverse applications, there are growing concerns about their reliability and security. LLMs remain vulnerable to adversarial attacks through carefully crafted prompts that circumvent safety mechanisms, while MLLMs face additional security challenges stemming from their multimodal nature. Despite considerable efforts in reinforcement learning from human feedback (RLHF) and supervised fine-tuning, existing safeguards have proven inadequate in addressing these critical vulnerabilities. This inadequacy stems from the fact that these models are inherently blackboxes that do not provide …

Af-Xray: Visual Explanation And Resolution Of Ambiguity In Legal Argumentation Frameworks, Yilin Xia, Heng Zheng, Shaun Bowers, Bertram Ludäscher

Computer Science Faculty Scholarship

Argumentation frameworks (AFs) provide formal approaches for legal reasoning, but identifying sources of ambiguity and explaining argument acceptance remains challenging for non-experts. We present AF-XRAY, an open-source toolkit for exploring, analyzing, and visualizing abstract AFs in legal reasoning. AF-XRAY introduces: (i) layered visualizations based on game-theoretic argument length revealing well-founded derivation structures; (ii) classification of attack edges by semantic roles (primary, secondary, blunders); (iii) overlay visualizations of alternative 2-valued solutions on ambiguous 3-valued grounded semantics; and (iv) identification of critical attack sets whose suspension resolves undecided arguments. Through systematic generation of critical attack sets, AF-XRAY transforms ambiguous scenarios into grounded …

Enlem: Ensemble Learning-Based Model To Detect Phishing Websites, Most Nilufa Yeasmin, Md Abu Rumman Refat, Bikash Chandra Singh, Zulfikar Alom, Zeyar Aung, Mohammad Azim

School of Cybersecurity Faculty Publications

Phishing involves manipulating individuals into revealing private data, e.g., user IDs, bank details, and passwords. The observed surge in fraud is related to increased deception, impersonation, and advanced online attacks. Thus, effective phishing detection methods are required to mitigate escalating global phishing threats. Existing methods (e.g., heuristics-based, signature-based, and visual similarity-based methods) attempt to detect phishing sites, and machine learning (ML) and deep learning (DL) methods are effective in the cybersecurity context in terms of learning from data, offering insights, and forecasting. However, independent ML algorithms are limited when handling complex data, and DL techniques surpass traditional ML methods in …

Toward Secure And Practical Machine Learning-Based Access Control: A Framework With Real-World Constraints And Adversarial Analysis, Olusesi Balogun, Mohammad Ghasemigol, Zhipeng Cai, Daniel Takabi

School of Cybersecurity Faculty Publications

Attribute-Based Access Control (ABAC) frameworks coordinate access requests based on subject, object, and environment attributes, as well as policy rules, and are widely used in corporate security systems. Recently, machine learning has been applied to ABAC to address policy-generation imbalances, misassigned privileges, and attribute leakages. However, existing MLBAC techniques do not consider the structural constraints and attribute interdependencies present in traditional ABAC systems. Moreover, these frameworks have not been extensively evaluated under black-box attack scenarios. To address these gaps, we propose extensions to MLBAC that integrate structural constraints, attribute dynamism, and attribute weighting into the MLBAC objective function. Additionally, we …

Exploring The Synergy Between Very Large Transformer And Lstm Models For Effective Medical Captioning From Videos To Text: The Impact Of Captioning In Healthcare, R. V. Aswiga, Moin Ahmed Zahir

Data Science Faculty Publications

In today’s rapidly evolving digital landscape, the demand for accurate and contextually relevant subtitles for image and video content, particularly in the medical domain, is increasingly critical. Despite the proliferation of visual data across various platforms, existing captioning systems often struggle due to variations in visual settings, complex temporal relationships, and nuanced semantics. Additionally, challenges such as limited datasets, privacy issues, and specialized annotation requirements make medical image captioning particularly difficult. To tackle these challenges, we investigate cutting-edge deep learning methodologies, specifically Transfer Learning and Transformer models, through a comparative analysis. Specifically, we focus on Transfer Learning through the MedVisionCapturer …

A Hybrid Cnn-Lstm Surrogate Model For Hyper-Resolution Spatiotemporal Flood Forecasting In Norfolk, Virginia, Yidi Wang, Jonathan L. Goodall, Chetan Kumar, Diana Mcspadden, Sergio A. Barbosa, Binata Roy, Ali Shahabi, Navid Tahvildari

Data Science Faculty Publications

Study region

Norfolk, Virginia, United States

Study focus

Accurate and timely flood forecasting is essential for enhancing resilience in coastal urban areas in the context of increasing frequency and intensity of rainfall, sea level rise and urbanization. This study presents a hybrid deep learning-based surrogate model that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to enable real-time spatiotemporal flood forecasting. The model leverages CNN to capture spatial features from inputs such as elevation and Topographic Wetness Index (TWI), while LSTM processes time-series inputs of rainfall and tide data to capture temporal features.

New hydrologic insights for …

Predicting Next-Day Eur/Usd Direction Using News Sentiment And Technical Indicators With Finbert And Xgboost, Darshan Sanjaybhai Khunt

Selected Full-Text Master Theses 2021-

Forecasting exchange-rate movements is a challenging task because currency prices are influenced not only by macroeconomic and financial variables but also by market sentiment reflected in financial news. This thesis examines whether financial news headlines can be used to predict the next-day directional movement of the EUR/USD exchange rate by applying finance-specific natural language processing and machine learning techniques.

The study uses a dataset of approximately 466,000 finance-related English-language news headlines collected between 2021 and 2025, aligned with daily EUR/USD closing prices. After preprocessing and temporal alignment, the data are used to construct a binary classification task in which the …

Game-Based Learning For Asynchronous Ai Literacy Course: Approach To Improve Students' Cognitive, Behavioural, Affective, And Ethical Learning Of Ai, Jinhee Kim, Guang Yang, Wing Sha Chan, Xi Lin, Yukyeong Song

STEMPS Faculty Publications

Educators in higher education face persistent challenges in scaling AI literacy across disciplines and helping novice learners understand abstract AI concepts. Although research on game-based learning (GBL) reports mixed outcomes, few studies have examined its large-scale use in mandatory, asynchronous AI literacy courses for diverse undergraduate populations. Addressing this gap, this study investigates a scalable GBL-based AI literacy course delivered to 4898 first-year undergraduates across disciplines. Using a mixed-methods design with 311 valid pre- and post-survey responses and 20 interviews, the study evaluates students' cognitive, behavioural, affective, and ethical learning of AI. Quantitative results show significant improvements in overall AI …

Simulating Social Attitudes With Llms: Accuracy, Demographic Effects, And Refusal Behavior In The Sensitive Domain Of Suicide Prevention, Cristina J. Perez, Michael P. Vasquez Jr., Philippe J. Giabbanelli, Patrick Y. Wu

VMASC Publications

Large language models (LLMs) are increasingly used to simulate public opinion, yet their validity in sensitive policy domains remains underexplored. We evaluate whether LLMs can reproduce attitudes toward suicide prevention policies using 32 questions drawn from seven nationally representative U.S. surveys (2023-2025). We systematically vary demographic conditioning (race/ethnicity, gender, age, education, income, party), prompt framing (direct elicitation, respondent embodiment, specialist embodiment), and model architecture (GPT-5 Nano, DeepSeek V3.2, Meta Llama 3.1 8B, Mistral Small 24B). Across 811,560 prompts, the mean absolute error—the average gap between predicted and human response distributions—is 23 percentage points. We also find that LLM responses to …

Reliable And Label-Efficient Learning For Open-World Visual Perception And Robot Learning Under Uncertainty, Zongyao Lyu

Computer Science and Engineering Dissertations

Modern learning systems deployed in open-world environments must make reliable decisions despite predictive uncertainty, previously unseen classes, limited annotations, and distribution shifts. This dissertation develops methods for reliable and label-efficient learning in visual perception and robot control.

First, this work studies uncertainty in object detection by representing semantic and spatial predictions probabilistically. A deep-ensemble framework aggregates detections into class-probability distributions and probabilistic bounding boxes, while a subsequent extension combines deep ensembles with Monte Carlo dropout to further investigate predictive uncertainty. Second, this dissertation addresses open-set recognition, where classes absent during training may appear at inference time. An empirical study shows …

Advancing Hate Speech Detection: Binary And Multiclass Approaches From Traditional Methods To Parameter-Efficient And Ontology-Guided Language Models, Mahmoud Abusaqer

Graduate Theses/Dissertations

The widespread proliferation of hate speech on social media platforms poses significant challenges for content moderation and user safety, requiring automated systems that are simultaneously accurate, efficient, and capable of fine-grained distinctions. This thesis investigates hate speech detection through five published manuscripts organized into two complementary threads: binary detection (hateful vs. non-hateful) and multiclass detection across demographic targeting categories. The binary thread progresses from a broad 38-model baseline spanning traditional machine learning, deep learning, and transformer architectures (where RoBERTa reaches 91.48% accuracy and CatBoost remains competitive at 88.60%) to parameter-efficient adaptation, in which Low-Rank Adaptation (LoRA) of large language models …

Students-Generative Ai Interaction Patterns And Its Impact On Academic Writing, Jinhee Kim, Sang-Soog Lee, Rita Detrick, Jialin Wang, Na Li

STEMPS Faculty Publications

Considering both the transformative opportunities and challenges presented by generative AI (GenAI) in academic writing, effectively integrating GenAI into the academic setting becomes a significant need requiring prioritization. Yet, there is limited understanding regarding the nature of interactions between different types of students, what behavioral patterns students exhibit during a student-GenAI interaction (SAI) on a given task, and how these different SAI patterns relate to the actual writing task performance. This study, therefore, aimed to identify SAI patterns of academic writing tasks depending on students’ level of AI literacy and examine the differences in academic writing performance between the identified …

Students' Perception Of Generative Ai-Assisted Collaborative Argumentation, Jinhee Kim, Seongryeong Yu, Rita Detrick, Liangjie Fan, Na Li

STEMPS Faculty Publications

The rapid scaling of generative artificial intelligence (GenAI) technology presents opportunities for personalised learning experiences and facilitates collaborative learning, including collaborative argumentation (CA). However, empirical research examining students' perceptions of GenAI-assisted CA within classroom contexts remains limited. This study explored university students' experiences with GenAI-assisted CA through in-depth interviews with 36 students following a CA activity using a ChatGPT4-embedded argumentation platform developed by the research team. Findings indicate that students viewed GenAI as serving multiple roles, including tool, facilitator, teaching assistant and machine buddy. Students perceived that GenAI-assisted CA could empower task performance and create a collaborative learning environment. Meanwhile, …

Standardization Of Neuromuscular Reflex Analysis—Role Of Fine-Tuned Vision-Language Model Consortium And Openai Gpt-Oss Reasoning Llm-Enabled Decision Support System, Eranga Bandara, Ross Gore, Sachin Shetty, Ravi Mukkamala, Christopher K. Rhea, Brittany S. Samulski, Amin Hass, Atmaram Yarlagadda, Shaifali Kaushik, Malith De Silva, Andriy Maznychenko, Inna Sokolowska, Kasun De Zoysa

VMASC Publications

Background/Objectives: Accurate assessment of neuromuscular reflexes, such as the Hoffmann reflex (H-reflex), plays a critical role in sports science, rehabilitation, and clinical neurology. Conventional interpretation of H-reflex electromyography (EMG) waveforms is subject to inter-rater variability and interpretive bias, limiting reliability and standardization. This study aims to develop an automated, interpretable, and robust agentic AI–driven framework for H-reflex waveform analysis. Methods: We propose a fine-tuned Vision–Language Model (VLM) consortium combined with a reasoning Large Language Model (LLM)–enabled decision support system for automated H-reflex interpretation. Multiple VLMs were fine-tuned on curated datasets of H-reflex EMG waveform images annotated with expert clinical observations, …

News Media Sentiment Toward Chinese Ai: A Comparative Analysis With Belt And Road Initiative Involvement And Public Opinion On China, Asya Vaisberg

Pomona Senior Theses

This study evaluates different countries' news media’s sentiment towards Chinese AI, between May 2023 and May 2024, by using Microsoft Azure NLP Sentiment Analysis. The results are then compared with the country’s public opinion on China and its involvement in the Belt and Road Initiative (BRI). For this study 12 countries have been selected which are USA, Australia, Pakistan, Peru, Russia, Romania, Italy, Greece, Portugal, Philippines, Brazil, and Egypt. For each one, GNews Application Programming Interface (API), which has access to more than 60,000 global news sources, was used to aggregate relevant news articles based on queried keywords. The collected …

Generative Ai And Llm Applications In Renewable Energy And Smart Grids: A Systematic Review For The Sustainable Energy Transition, Umit Cali, Ugur Halden, Merlinda Andoni, Ferhat Ozgur Catak, Si Chen, Benoit Couraud, Emre Kantar, Samuel Knapper, Ibrahim Kucukdemiral, Huseyin Kusetogullari, Murat Kuzlu, Yashar Mousavi, Sonam Norbu, Taha Selim Ustun, David Flynn

Engineering Technology Faculty Publications

The global energy transition toward decarbonization and digitalization requires advanced methods to manage decentralized, data-intensive cyber-physical energy systems. This systematic review analyzes 106 research studies on Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) in renewable energy and smart grids, organized into seven application clusters covering forecasting, system design, operation, reliability, data and cybersecurity, and energy markets. The review situates these applications within a Cyber-Physical-Social Systems (CPSS) framework. Results show that GANs dominate current applications (47.2%), followed by LLMs (10.4%) and VAEs (9.4%), with growing adoption of diffusion and score-based models (7.5% each). Selected studies report improved probabilistic forecasting …

Adaptive Multi-Grade Deep Learning For Highly Oscillatory Fredholm Integral Equations Of The Second Kind, Jie Jiang, Yuesheng Xu

Mathematics & Statistics Faculty Publications

This paper studies the use of Multi-Grade Deep Learning (MGDL) for solving highly oscillatory Fredholm integral equations of the second kind. We provide rigorous error analyses of continuous and discrete MGDL models, showing that the discrete model retains the convergence and stability of its continuous counterpart under sufficiently small quadrature error. We identify the DNN training error as the primary source of approximation error, motivating a novel adaptive MGDL algorithm that selects the network grade based on training performance. Numerical experiments with highly oscillatory (including wavenumber 500) and singular solutions confirm the accuracy, effectiveness and robustness of the proposed approach.

Artificial Intelligence And Machine Learning In Smart Vaginal Formulation Development, Deborah A. Ogundemuren, Vivek Agrahari, Andrew P. Wong, Carolina Herrera, Margaret O. Ilomuanya, Gustavo F. Doncel

CONRAD Publications

Vaginal drug delivery in women's health remains underutilized and insufficiently studied, largely due to the complexity and dynamic nature of the vaginal microenvironment. Variations in vaginal pH, hormonal levels, and microbiota composition introduce significant biological variability, complicating formulation design and contributing to inconsistent therapeutic outcomes and poor patient adherence. Conventional vaginal formulations often fail to account for these individual differences, highlighting the need for more adaptive and predictive approaches. Emerging advances in artificial intelligence (AI) and machine learning (ML) offer promising strategies to address these challenges by enabling multi-parameter, data-driven formulation development that explicitly considers biological variability. Despite their transformative …

The Ai Revolution In Virtual Try-Ons: A Means-End Chain Model Perspective, Ju-Young M. Kang, Ji Young Lee, Dooyoung Choi, Sumin Helen Koo, Jeehyun Song, Youngjin Bahng

Educational Leadership & Workforce Development Faculty Publications

Leading brands have begun to implement artificial intelligence-driven virtual try-on (AI VTO) technology, which helps reduce returns and increase conversion rates, repeat purchases, and customer loyalty. This research aimed to examine how retail user experience with specific perceived quality factors of AI VTOs influences users’ value equity and downstream loyalty and to investigate the moderating effects of clothing in relation to the self as structure and concern for physical appearance, based on a Means–End Chain model. Data were collected from 509 U.S. online apparel shoppers using a consumer panel. Structural equation modeling and multigroup analysis were used for data analysis. …

Designing Ai Systems To Support A Productive-Failure-Based Learning: Insights From Adult Learners On Ai Applications And Ai System Design Principles, Jinhee Kim, Xi Lin, Seongryeong Yu, Rita Detrick

STEMPS Faculty Publications

Emerging capabilities of generative artificial intelligence (GenAI) offer significant potential to support productive failure (PF)-based learning, which engages adult learners (ALs) in exploring problems before instruction and learning from their initial attempts. However, the effective use of AI to support multifaceted areas of PF-based learning, including problem generation, exploration, consolidation, and knowledge assembly, is limited. Furthermore, AI design principles to support PF-based learning remain under-researched. This study, therefore, aims to investigate ALs’ perceptions of AI applications in enhancing PF-based learning and to explore the essential design principles of AI systems for PF-based learning. To achieve these aims, the study conducted …

Srgan-Based Deep Learning Framework For Wind Turbine Damage Detection From Sentinel-2 Imagery, Kübra Çakir, Onur Elma, Murat Kuzlu

Engineering Technology Faculty Publications

The operational reliability of wind turbines is critical for sustainable energy production in smart grids. This study proposes a remote monitoring approach using perceptually enhanced satellite imagery. Sentinel-2 multispectral data (10 m resolution) has been processed with a Super-Resolution Generative Adversarial Network (SRGAN) to improve visual quality to a perceptual resolution of 30 cm. Although true spatial refinement is not achieved, the sharper structural details enhance classification accuracy. The data set comprises 15,000 images—10,000 SRGAN-enhanced and 5000 augmented through rotation, zoom in, increasing brightness, noise addition, and blurring. A custom Convolutional Neural Network (CNN) has been trained to classify turbines …

Interpretable Machine Learning For Bridge-Pier Scour Prediction And Flood Resilience, Adil Khan, Dalya Ismael

Engineering Technology Faculty Publications

Bridge-pier scour is a leading cause of flood-induced bridge failure, yet practice still lacks transparent, physics-informed tools that link data-driven prediction with design guidance. This study develops an interpretable, physics-aware machine-learning framework to predict equilibrium scour depth and translate those predictions into actionable strategies for flood-resilient infrastructure. Using the 2014 U.S. Geological Survey Pier-Scour Database (569 laboratory cases), five models: Gradient Boosting, AdaBoost (Tree), XGBoost, Gaussian Process (RBF kernel), and Kernel Ridge (polynomial), were trained and evaluated with K-fold cross-validation. Model performance was evaluated using R², RMSE, and MAE. Gradient Boosting performed best, achieving training and testing R² of 0.99 …