Physical Sciences and Mathematics Commons^™

Prune: A Patching Based Repair Framework For Certifiable And Privacy-Robust Unlearning Of Neural Networks, Xuran Li, Jingyi Wang, Xiaohan Yuan, Peixin Zhang

Research Collection School Of Computing and Information Systems

Machine unlearning has emerged as a key mechanism for enabling the “right to be forgotten” in neural network models, allowing the selective removal of specific training data upon request. Existing approaches typically rely on retraining models with the remaining data, which is computationally expensive and difficult to verify, especially when deployed models are distributed or resource-constrained. To address this challenge, our prior conference work introduced PRUNE, a patching-based framework that formulates unlearning as a neural network repair problem. PRUNE achieves targeted forgetting by learning lightweight patch networks that redirect model predictions on the data to be unlearned while preserving performance …

Machine Learning For Functional Outcome Prediction After Vestibular Schwannoma Surgery: A Systematic Review And Diagnostic Test Accuracy Meta-Analysis, Shiva Nischal, Shaan Patel, Musa China, Kush Kale, Yi Hein Chai, Santosh Guru, William Muirhead, Patrick Grover

Department of Neurosurgery Faculty Papers

PURPOSE: Machine learning (ML) models have been increasingly applied to predict postoperative facial nerve dysfunction and hearing preservation after vestibular schwannoma (VS) surgery. However, reported performance varies substantially, and the overall diagnostic accuracy and clinical reliability of these models remain uncertain. We conducted a systematic review and diagnostic test accuracy meta-analysis to characterise the current state and methodological readiness of ML-based prediction of these outcomes.

METHODS: PubMed, Embase, and CENTRAL were searched from inception to February 2026. Studies evaluating ML-based prediction of facial nerve function or hearing preservation following VS surgery were included. Diagnostic performance metrics were pooled using random-effects …

Set-Valued Conjugate Calculus And Duality In Optimization And Machine Learning Via Generalized Relative Interiors, Gary Lee Sandine

Dissertations and Theses

Convex analysis, optimization, and duality theory are broadly applicable to a multitude of real-world problems. Theoretical frameworks ensuring strong duality and optimality are fundamentally tied to convex separation under qualification conditions frequently involving topological or relative interiors. These can be empty in convex sets that appear naturally in infinite-dimensional models, yet guarantees of optimality and strong duality are often attainable. This dissertation presents frameworks for applying convex separation via generalized relative interiors to develop generalized calculus rules as well as conditions for strong duality and optimality for infinite-dimensional constrained convex optimization problems.

The first major contribution is the introduction of …

Understanding And Predicting Precipitation Characteristics In The United States Through Machine Learning, Numerical Modeling And Measurements, Cody Luther Ratterman

All Graduate Theses and Dissertations, Fall 2023 to Present

Precipitation is one of the most important yet uncertain variables in the climate system. It varies dramatically in terms of timing, accumulation, rate, and phase, depending on location, season, circulation patterns, and atmospheric conditions. Precipitation forecasts, especially snowfall, are essential to supporting drought mitigation and water management in the Intermountain West. Because rainfall and snowfall lead to opposite effects on snowpack, accurately partitioning rain and snow is important to estimate snowpack levels, winter recreation, mountain ecosystems and runoff. The research findings in this dissertation have advanced the understanding and prediction of precipitation and snowpack in the U.S. by addressing the …

Applications Of Machine Learning To Gas Plume Analysis In Longwave Infrared Hyperspectral Images, Scout C. Jarman

All Graduate Theses and Dissertations, Fall 2023 to Present

Each pixel from a hyperspectral camera measures the intensity of light over a continuous range of wavelengths, which is in contrast to traditional color cameras, which just measure the intensity of red, green, and blue wavelengths of light. Longwave infrared hyperspectral images can be used to detect gases from a distance by measuring how different materials emit and absorb heat. This makes them useful for applications such as monitoring industrial emissions or locating hazardous gas leaks. In practice, however, gas signatures in these hyperspectral images are often weak and easily obscured by variations in the background scene, making reliable identification …

Evaluating Machine Learning Models On Classification Of Novel Cyber Attacks In The Healthcare Domain, Promise Ehimen

Dissertations, Theses, and Projects

The increasing adoption of the Internet of Medical Things (IoMT) has improved healthcare delivery through connected medical devices while simultaneously expanding the cybersecurity risks facing healthcare organizations. Although machine learning based intrusion detection systems have demonstrated high detection accuracy, their ability to respond reliably to previously unseen cyberattacks remains uncertain. This study investigated how a Neural Network model and a Logistic Regression model classified novel cyberattacks within the IoMT environment. The Neural Network and Logistic Regression models were both trained and tested using a subset of the CICIoMT2024 benchmark dataset. The Neural Network achieved 99.82% test accuracy and a 0.94 …

New Distance Distributions Of Asymptotic Giant Branch Stars And Their Role In Tracing Galactic Structure, Rajorshi Bhattacharya

Physics & Astronomy ETDs

Asymptotic giant branch (AGB) stars are among the final evolutionary stages of low- and intermediate-mass stars and among the most luminous cool stellar populations in the Milky Way (MW). Their strong infrared emission allows them to be observed through regions of high interstellar extinction, making them useful tracers of the inner MW. However, their use is limited by unreliable distances because many are dust-obscured and variable, while geometric parallaxes are often unavailable or uncertain. This thesis develops statistical distance-estimation methods for large samples of oxygen-rich AGB stars. The resulting distances broadly agree with literature estimates, supporting their statistical reliability for …

Identifying Textual Predictors Of Early Termination In Clinical Trials In Medicine: An Explainable Machine-Learning Study, Rohan Ramnarain

Dissertations, Theses, and Capstone Projects

About one in five clinical trials in medicine ends early, wasting valuable resources and reducing the evidence available for developing life-saving medical treatments. This project uses a method called Trial2Vec, which is a self-supervised machine-learning method that converts clinical trial documents into dense numerical representations that capture their key design and clinical characteristics, to turn each proposed clinical trial’s written protocol into a compact numerical profile (a process referred to as embedding). These profiles are then paired with a predictive machine learning models to identify the words and phrases in the trial documents that can signal a higher risk of …

Evaluating Soil Health And Crop Yield In Louisiana Agricultural Systems: Impacts Of Best Management Practices And Prediction Models, Hector J. Mendoza Lagos

LSU Doctoral Dissertations

The adoption of conservation management practices is critical for improving soil health, enhancing nutrient use efficiency, and sustaining crop productivity in row crop systems in Louisiana. This study evaluated the role of conservation agronomic practices, soil biochemical indicators, and machine learning predictive models to improve soil nutrient dynamics, soil health indicators, microbial communities (MC), and crop productivity on a corn (Zea mays L.) research plot scale and in a commercial forty-hectare cotton (Gassypium hirsutum L.)-corn-soybean (Glycine max L.) rotation system in northeast Louisiana. The objectives of the study were to evaluate soil nutrient dynamics and MCs under …

Machine Learning For Modeling In An Elementary Differential Equations Class, Nathan Albin, Andrew G. Bennett, Abhinav Chand

CODEE Journal

Mixing machine learning with modeling is an area of increasing importance. This paper presents a lesson where students model a spring-mass system both using traditional analysis with linear damping and using machine learning to learn the damping from real data. The machine learning is implemented in a Jupyter notebook hosted on Google Colab, allowing students to train the neural network without requiring the students to carry out coding. Students get experience with how machine learning can fail, how it can work, and the time and data requirements for machine learning to succeed, and are asked to apply this knowledge to …

Mapping Homogeneous Configuration States For Learning Based Motion Planners, Yazied Hasan

Computer Science ETDs

Reinforcement learning (RL) excels at solving complex tasks, but training times can become prohibitively large for challenging motion-planning problems. Methods that address this cost often require additional training or tuning, counteracting the goal of reducing training time. A more effective approach is to exploit inherent task equivalences: many elements of the state space, dynamics, or structure are functionally interchangeable, enabling simplification or knowledge reuse. We present learning solutions that leverage these equivalences to enhance the RL process. First, we leverage the symmetry of homogeneous multi-agent teams to simplify the task to a single strategy. Second, we map correspondences between distinct …

Computational Design Of Peptides And Proteins Through Machine Learning Approaches, Emily J. Hendrix

Chemistry and Chemical Biology ETDs

Advancements in machine learning have emerged as a pivotal tool in computational biochemistry, offering new advancements to address challenges in protein structure and function. However, current machine-learning approaches offer limited insight in understanding protein dynamics. The purpose of this work is to combine traditional physics-based computational tools, such as molecular dynamics and coarse-grained simulations, with recently developed AI-driven computational tools to bridge gaps and advance the understanding of proteins in both structural and dynamic aspects. I investigated several approaches such as (i) traditional physics-based methods to study protein conformation and ensembles; (ii) identifying a peptide inhibitor for the PICK1 PDZ …

Stridevision: Automated Detection Of Running Form Deviations From 2d Pose Estimation And Machine Learning, Paulina Eguibar Ortega

Honors Theses

Running gait analysis plays a critical role in injury prevention and performance optimization, however, existing approaches often rely on specialized laboratory equipment or wearable sensors with limited interpretability. Recent advances in computer vision, particularly 2D human pose estimation, enable markerless motion analysis from standard video. However, progress remains constrained by the lack of publicly available datasets designed for running form analysis.

In this work, we introduce a preliminary dataset and benchmark for stride-level running gait analysis. The dataset consists of 73 treadmill running videos from 15 participants with varying experience levels, annotated with over 4,600 stride-level labels across multiple biomechanical …

Security Assessment Of A Machine Learning Approach To Generate And Validate Digital Signatures, Juan Ortiz Couder

Doctoral Dissertations and Master's Theses

Cybersecurity has become a global concern as cyber-attacks have become more common, and the cost of the damage caused by them continues to increase. There are several approaches to improve the cyber security of systems such as Digital Signatures, hashing, watermarking, and encryption among others. Digital Signatures are a cryptographic technique used to verify the authenticity and integrity of digital messages or documents. Digital Signatures use a combination of hashing and public-private key encryption to verify the authenticity and integrity of videos, just as they are used for documents and messages. As a result of using a combination of other …

Ais26s: Ai In Biomedicine, Shiqian Shen

Paul English Applied Artificial Intelligence (AI) Institute Publications

This presentation explores the role of artificial intelligence in advancing biomedical research and clinical practice from the perspective of a physician-scientist. Dr. Shiqian Shen discusses current challenges in neuroscience and medicine, including limitations in human observation, data analysis, and decision-making across complex biological systems. The talk highlights how AI-driven approaches such as computer vision, neural signal processing, and large-scale data modeling can improve the understanding of disease mechanisms, enhance experimental workflows, and enable more precise, patient-centered care. Specific applications include automated classification of neural activity, behavioral analysis in animal models, and multimodal data integration. A key concept introduced is “shadow …

Machine Learning For Handwritten Character Recognition, Hannah Freitag

Honors Capstones

Handwritten character recognition remains a challenging problem in machine learning due to the high variability of handwriting across individuals and the visual similarity between certain character classes. This project explores whether Singular Value Decomposition (SVD)-based dimensionality reduction can serve as an effective preprocessing step for a fully connected neural network trained on the EMNIST Balanced dataset, a 47-class benchmark of handwritten digits and letters. By projecting 784- dimensional pixel inputs onto the top 70 principal components, approximately 90% of the total variance is preserved while reducing input dimensionality by 91%. The resulting SVD-based model achieves approximately 94% test accuracy, outperforming …

Ai Institute Summer Camp Academic Preview Webinar: Curriculum, Research, And Outcomes, Paul English Applied Artificial Intelligence Institute

Paul English Applied Artificial Intelligence (AI) Institute Publications

This webinar presents an academic preview of the AI Institute Summer Camp hosted by the Paul English Applied Artificial Intelligence Institute at the University of Massachusetts Boston. The session introduces the program’s curriculum, structure, and student outcomes, providing insight into a hybrid learning model that combines faculty-led lectures, hands-on labs, and guided project development. The webinar highlights the program’s five-week structure, covering topics such as machine learning, neural networks, computer vision, speech and language processing, and generative AI. Participants learn how students engage in real-world AI applications, complete portfolio-ready projects, and develop research and presentation skills. This session is designed …

Escaping The Promotion Trap: A Machine Learning Framework For Brand Equity Preservation In Beverage Cpg, Lucas P. Jones

Data Science Undergraduate Honors Theses

When companies acquire beverage brands, they typically value them based on total sales revenue. This traditional approach treats all sales equally over time, whether they are driven by genuine consumer demand or temporary discounts. This is important because while promotions can boost short-term sales, they tend to erode brand value over long periods of time. The measurement problem extends to acquisitions, where buyers lack the tools to distinguish real consumer demand from artificial promotional inflation.

This thesis develops a framework to separate genuine baseline demand from promotional dependence using Nielsen scanner data covering 189 beverage brands across 188,304 weekly observations …

The Quality Assurance Machine – A Software Quality Assurance Architecture For Ml-Enabled Systems, Shane E. Downing

All-Inclusive List of Electronic Theses and Dissertations

This dissertation evaluates whether a reusable assurance architecture, the Quality Assurance Machine (QAM), can provide effective product and process quality assurance for ML-enabled software platforms. The QAM is a system-level SQA architecture that turns plans and policies into versioned configurations, executes them in controlled environments, and produces preserved run evidence that supports traceability, auditability, and controlled change. The study follows Design Science Research and evaluates the instantiated artifact using eight assurance requirements (AR1–AR8) synthesized from standards-based guidance, including IEEE 730 and ISO/IEC/IEEE 15026. A four-year longitudinal evaluation combines two methods. First, operational evidence from routine regression and release-validation runs, defect …

Using Ai For Data Loss Prevention, Camden A. Wright

Theses/Capstones/Creative Projects

Data Loss Prevention (DLP) systems play a critical role in protecting modern systems that handle sensitive information from both accidental and malicious exposure. Traditional DLP approaches often rely on static rules and methods that can struggle to adapt to complex and evolving data patterns. This paper presents a hybrid DPL system that integrates machine learning-based message classification, rule based policy enforcement, and context-aware access control to improve both detection accuracy and decision reliability. In addition, the system introduces a second stage access control model that evaluates user context, including role of clearance level and job title to determine whether access …

Evaluating Modern Neural Network Architectures For Suicide Prediction, Kyle Brown

Master's Theses

Suicide remains a leading cause of death among adolescents despite more access to healthcare information than ever before. Medical professionals struggle to make accurate diagnoses and catch warning signs with the overwhelming amount of data available. Machine learning algorithms, including neural networks, have previously been employed for this task, yet it remains an understudied domain.

This research aims to evaluate the capabilities of Multi-Layer Perceptron (MLP) and a selection of its successors, ResNet and MLP with a category embedding layer, at the task of predicting suicidal ideation among high-school students. This research finds ResNet to be the most capable at …

Geospatial Science And The Changing Environment: Applied Methods For Sustainable Agriculture And Landscape Management, Harrison Wakefield Smith

Graduate Theses and Dissertations

Rapid environmental change is disrupting agricultural productivity, ecological function, and the long-term resilience of managed landscapes. At the same time, the rapid growth of geospatial data science has improved our understanding of environmental change and increasingly is being used to improve sustainability in agricultural and environmental management. However, critical gaps remain that limit the applicability of data-driven insights in landscape management. This dissertation investigates the potential of geospatial analytics for sustainable agriculture and landscape management, with a focus on applied methods that operate across spatial and temporal scales. Using field, landscape, regional, and national datasets, it explores the capabilities and …

Genre Prediction Using Rnns And Llm-Enhanced Video Game Review Data, Gabriel Young

Graduate Theses and Dissertations

LLMs (Large Language Models) are powerful tools for engaging with textual data, carrying many advantages over classical NLP (Natural Language Processing) and ML (Machine Learning) approaches. However, a classical ML model can still be faster, more efficient to run, and accessible than an LLM. We seek to gain the benefits of LLM text comprehension and preserve them in a classical ML model, a hybrid approach. The LLM operates on text to surface relevant information and associations in our problem space, then the ML model trains on the LLM output. The model may learn from the LLM and provide a more …

Gradient Based Optimization Methods For Robust Learning And Biomedical Signal Modeling, Jarrod Mau

All Graduate Theses and Dissertations, Fall 2023 to Present

This dissertation explores how modern artificial intelligence techniques can be used to better understand complex biological data. Specifically, it develops new machine learning based methods and applies them to two important biomedical problems: analyzing brain signals and studying protein behavior.

The first part of the work introduces a new machine learning approach designed to improve how computers classify structured data. Traditional neural networks are powerful but can sometimes generalize poorly. This research proposes a method that combines the flexibility of neural networks with the reliability of ensemble techniques, leading to more robust and accurate predictions across different types of datasets. …

Integrating Multi-Source Data With Machine Learning Techniques To Upscale Wetland Carbon Dioxide Fluxes, Abdullah Sulaiman Abdullah Al Fazari

Electronic Theses and Dissertations

Accurate quantification of atmospheric carbon dioxide (CO₂) fluxes in wetland ecosystems is essential for understanding their role in both regional and global carbon dynamics, particularly in the context of climate change. However, the spatial and temporal heterogeneity of wetlands presents major challenges for developing reliable upscaling models. This research developed and validated a comprehensive framework to upscale CO₂ fluxes across the Everglades National Park (ENP) and Big Cypress National Preserve (BCNP) in South Florida through the integration of multi-source datasets, including AmeriFlux eddy covariance (EC) tower measurements, NASA’s BlueFlux airborne CO₂ data, and multispectral satellite imagery from Landsat 8 OLI …

Machine Learning Based Models For Simulation And Analysis Of Bulk Earth Melt System, Abin Shakya

LSU Doctoral Dissertations

Understanding the segregation of bulk Earth melt systems into metallic (core) and silicate (mantle) phases under high-pressure and high-temperature conditions is central to modeling Earth’s interior, yet relevant experimental and computational studies remain limited. This work develops a machine learning–based simulation pipeline that iteratively couples first-principles (quantum mechanical) calculations with neural network training to generate high-fidelity force fields. Using major-element Fe–Mg–Si–O melt systems, with and without H and N, as testbeds, we demonstrate that this framework enables large-scale molecular dynamics simulations at near first-principles accuracy. We further introduce a sequence of phase identification methods, progressing from statistical binning of elemental …

Developing Machine Learning Algorithms For Highly Imbalanced Neonatal Disorder Data, Ali Nawaz

Thesis/ Dissertation Defenses

Neonatal disorders such as low birth weight, very low birth weight, extremely low birth weight, preterm birth, and very preterm birth increase the likelihood of high neonatal morbidity or mortality and call for early identification. However, the rarity of occurrence of these conditions in the clinical datasets has resulted in a severe class imbalance, raising questions about the application of binary classification models to them. Therefore, this thesis proposes a sequential methodological framework for neonatal disorder detection under different assumptions related to the availability of labels. Initially, binary classification experiments are conducted to analyze the behaviour of commonly used classification …

Ais26s: Genai And Llms For Cybersecurity, Political Sciences, Transportation Security, And Resiliency, Latifur Khan

Paul English Applied Artificial Intelligence (AI) Institute Publications

This presentation examines the role of generative artificial intelligence (GenAI) and large language models (LLMs) in addressing complex challenges across cybersecurity, political science, and transportation security. Dr. Latifur Khan discusses how advanced AI methods can be applied to threat detection, data analysis, and decision-making in high-risk and data-intensive environments. The talk highlights interdisciplinary applications of LLMs, emphasizing their ability to extract insights from large-scale data, improve system resilience, and support intelligent infrastructure. Emerging research directions and practical implications for real-world deployment are also discussed.

A View Under The Hood: Duquesne Kline's Law And Computing Program, Wesley M. Oliver, Katherine L.W. Norton, Martin Mckown, David Horrigan

West Virginia Law Review

No abstract provided.

Bayesball : A Comprehensive Framework For Predicting Ucl Injury, Brady M. Pinter, Will Best Ph.D.

SPARK Symposium Presentations

Ulnar Collateral Ligament (UCL) reconstruction, commonly referred to as Tommy John Surgery, has seen a significant rise among Major League Baseball (MLB) pitchers, prompting growing interest in identifying the mechanical and performance-based factors that contribute to injury risk. While previous studies have examined these relationships using traditional frequentist approaches separately, this study combines multiple different model techniques to present a broad framework for finding significant predictors of UCL Surgery. These models include Lasso and Ridge Regression,  Principal Component Regression (PCR) , Partial Least Squares Regression (PLS) , Random Forest, Multiple Linear Regression, and a Bayesian Statistical Model. Using these models, …