Digital Commons Network^™

First-Order And Second-Order Adjoint Method And Stochastic Approximation For Inverse Problems, Adrian Heldt

Theses

We present a unified framework for estimating stochastic parameters in general variational problems. This nonlinear inverse problem is formulated as a stochastic optimization problem using the output least-squares (OLS) objective, which minimizes the discrepancy between observed data and the computed solution. A key challenge in OLS-based formulations is the efficient computation of first- and second-order derivatives of the OLS functional, which depend on the corresponding derivatives of the parameter-to-solution map—often costly and difficult to evaluate, especially in stochastic settings. To address this, we develop a rigorous computational approach based on first- and second-order adjoint methods for inverse problems governed by …

Silicon Photonics For Quantum Information, Evan Manfreda-Schulz

Theses

Quantum information science and technology (QIST) combines principles of quantum mechanics and information theory to develop new approaches for processing, transmitting, and sensing information. Practical realization of applications in QIST requires generation and validation of a significant amount of entanglement, which is not only a property but a logical resource. Silicon photonic integrated circuits (PICs) are an attractive platform for applications in QIST and for generating the entanglement resource because they are foundry-fabricated, are compatible with existing fiber optic infrastructure, have a small spatial footprint, and afford access to several high-dimensional (HD) degrees of freedom. This dissertation covers three projects …

Resonating Patterns: Adaptive Resonance Theory And Self-Organizing Maps, Meeti Dixit

Theses

Adaptive Resonance Theory (ART) represents a powerful neural network architecture designed to address the stability-plasticity dilemma. Its primary objective is to enable rapid learning without compromising retention. ART embodies characteristics of self-organization and self-stabilization, distinguishing it from traditional neural networks. Unlike error-based learning prevalent in conventional models, ART employs competitive learning mechanisms. One of the distinguishing features of ART is its involvement in hypothesis testing, a departure from the approach of deep learning. This capability allows ART to learn dynamically in real-time environments. Over time, various ART models have emerged that cater to different types of data, such as binary …

Predicting Student Dropout Risk Using Machine Learning, Fatma Alameri

Theses

Student dropout remains a persistent challenge in higher education, undermining institutional performance, reducing workforce preparedness, and limiting students’ academic and economic opportunities. Accurately identifying students at risk of attrition is complex, due to the interplay of academic, financial, and behavioral factors. This thesis addresses this challenge by applying a combined machine learning framework—integrating both unsupervised and supervised techniques—to predict student dropout using structured, first-year academic and financial data. The study utilizes a comprehensive dataset of 4,424 undergraduate student records from a European higher education institution, covering ten academic years and comprising 35 variables related to academic performance, enrollment behavior, and …

From The Bleachers To The Browser: Redefining Fan Experience With Ar And Ai In Smaller Teams, Jennifer Lee Wunder

Theses

This project documents the creation and deployment of HootyHoo, an interactive augmented reality (AR) mascot experience designed for the O’Fallon Hoots, a small-scale collegiate summer baseball team. Built using accessible, open-source tools such as WebXR, Mixamo, Meshy, Botpress, Claude and ChatGPT, this prototype merges AI-driven conversation with animated 3D avatar interaction—redefining how fans engage with sports organizations digitally. Unlike enterprise-level applications used by professional franchises, HootyHoo is entirely browser-based, eliminating the need for app downloads and ensuring maximum accessibility for families and new fans with smartphones. The experience centers on Hooty, the team mascot, who answers questions about baseball and …

Knot Alone: A 24/7 Live Chat App For Sewing Enthusiasts, Erica Frankenhoff

Theses

Knot Alone innovates support in the sewing industry by providing a 24/7 chat experience driven by community engagement. Unlike other forums or workshops with time limits, this platform guarantees interactive, personalized, and immediate guidance. Regardless of whether it’s a new beginner sewer learning how to thread a sewing machine, or a sewing enthusiast who is more advanced and is just figuring out issues with fabric tension, Knot Alone users can, at any hour, seek out professional advice. Professional advice will be provided by sewing educators and instructors, post graduate fashion students, and established sewing bloggers and influencers. By joining together …

Forecasting Load At Residential, Industrial, And Commercial Stations For Dubai Electricity And Water Authority: A Machine Learning Approach To Managing Generation During Peak And Off-Peak Times Over The Coming Years, Mohammad Anoohi

Theses

In Dubai, with the rapid growth of energy demand, efficient demand side management is necessary to optimize the distribution of electricity, reduce peak loads, and improve grid stability. Traditional DSM strategies are based on historical data and reactive control mechanisms that cannot adapt to evolving consumption patterns. This research uses Machine Learning techniques to enhance DSM in residential, commercial, and industrial sectors by developing predictive models that forecast energy demand based on seasonal variations. The dataset used was from Dubai Electricity and Water Authority (DEWA), covering the consumption patterns across Summer, Winter, Transition from Winter to Summer, and Transition from …

Using Generative Ai For Tutoring Data Science, Yusra Khalid

Theses

A large increase in the use of Generative AI has been observed in the last few years. Data science is also a rapidly growing field with a high demand for skilled professionals. The goal of this thesis is to explore the potential of Generative AI, specifically ChatGPT, in facilitating data science education. The focus is on how ChatGPT can be used as a tutor to help solve practical exercises. The capabilities of Generative AI were explored along with the comparison of a few different models in terms of data science. Exploratory analysis was conducted to compare Generative AI models and …

Stochastic Variational Autoencoder, Shounak Desai

Theses

This work explores a novel generative modeling approach inspired by variational autoencoders (VAEs). Traditional VAEs rely on a recognition model (encoder) that approximates the latent posterior with a single gaussian distribution for each input, limiting their flexibility in capturing complex data distributions. In contrast, we propose a modified recognition model that utilizes stochastic mixtures of gaussians, allowing for a more expressive latent representation. By leveraging stochastic neural networks within the VAE framework, we aim to achieve a tighter evidence lower bound (ELBO) on the log-likelihood of the data. Our approach is a preliminary investigation to enhance the latent space structure …

Groundwater Storage Dynamics In Abu Dhabi Emirate Using In-Situ And Grace Data, Tala Jalal Maksoud

Theses

Water scarcity presents a significant challenge to the sustainable development; especially in arid regions like Abu Dhabi Emirate. Addressing concerns over nonrenewable aquifer depletion, caused by agricultural expansion and lifestyle advancements, has, therefore, become crucial in such regions. This thesis investigates Groundwater Storage (GWS) dynamics in Abu Dhabi Emirate by integrating satellite data from the Gravity Recovery and Climate Experiment (GRACE) with in-situ data from 257 wells over two decades (2002-2022), employing GIS and RS technologies. The main aim of this study is to provide a comprehensive approach to evaluate GWS dynamics considering multiple perspectives including: dealing with missing datasets, …

Elliptic Curves And Their Applications In Cryptography, Mercedes Danange Moll

Theses

This thesis explores the mathematical foundations and cryptographic applications ofelliptic curves. To understand elliptic curves, we start by introducing group theory, ring theory, and field theory. Then, we dive into elliptic curves, their algebraic structure, and the group law defined on their points. We then examine elliptic curves over finite fields, leading to the discrete logarithm problem and key cryptographic applications such as Weil Pairings, the Diffie-Hellman key exchange, Digital Signature Algorithm, and Elliptic Curve Digital Signature Algorithms. Using the theory and its application, this thesis highlights the significance of elliptic curves in modern cryptography.

Customer Flow Prediction At Emirates Id Centers, Humaid Ahmed Saeed Alkhuroosi

Theses

Emirates ID centers face significant resource management challenges due to fluctuating customer traffic, leading to long wait times, customer dissatisfaction, and inefficient resource use. This thesis explores the application of time series analysis to predict customer traffic at Emirates ID centers, focusing on the Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) models. The primary research question is: “Can historical queue data from the Qmatic system be effectively used to forecast customer traffic at Emirates ID centers?” To answer this question, 800,000 observations of historical ticket issuance data from the Qmatic queue management system were analyzed. The study …

Beyond Regex – Heuristic-Based Secret Detection, Jesse Burdick-Pless

Theses

Accidental token or credential leakage presents a significant concern within digital environments. Current detection methods of such secrets employ ruleset-based techniques to identify secret information. These methods use pattern recognition within strings (i.e., regex rules) to pinpoint characteristics that resemble various types of secrets. However, regex does not allow for detection of secrets that lack specific patterns, such as passwords. This research addresses the possibility of using heuristics and machine learning to develop a reliable and accurate method for determining if a given string is merely a piece of inconsequential data or a leaked secret requiring timely attention, without the …

Modelling And Optimization Of A Desiccant Cooling System For Industrial Applications In Dubai, Ahmad Ababneh

Theses

The study examines the modelling and optimization of solar-assisted desiccant cooling systems (SADCS) specifically designed for industrial applications in Dubai. Four system configurations were evaluated under Dubai's extreme climate using TRNSYS 18 simulation software: classic ventilation, variable percentage recirculation, ventilation with a sensible heat exchanger that utilizes exhaust air to preheat the feed of the auxiliary heating air, and recirculation with a sensible heat exchanger that also utilizes exhaust air for preheating the auxiliary heating air feed. A parametric study comprising 60 simulation cases was performed, examining variations in desiccant wheel effectiveness, airflow rates (3–5 ACH), regeneration temperatures (50–80 °C), …

Wide Lock-In Energy Harvesting From Vortexinduced Vibrations Of A Deformable Cylinder, Ahmed Raafat Mostafa

Theses

Energy harvesting from ambient sources has gained attention due to increasing energy demands. Despite VIV-based harvesters showing significant potential, their lock-in region, where significant power is generated, is narrow. Given the continuously varying ambient conditions of fluid currents, harvesters can easily fall into de-synchronization, yielding low energy output. Existing solutions like tunable masses or multiple degrees of freedom systems increase complexity and weight, limiting practical applications. This work introduces a novel variable diameter cylinder mechanism—a practical technique that actively tunes the cylinder’s geometry in real time to enhance energy harvesting efficiency from VIV. The mechanism employs an expanding pulley system …

Energy Forecasting Inaccuracies And Their Direct Impact On Grid Performance, Rashed Khalid Alboom

Theses

Accurate energy forecasting is critical for the stability, efficiency, and cost-effectiveness of modern power grids, particularly as renewable energy sources like solar and wind become more prominent. The variability of these sources presents challenges for traditional forecasting models, which struggle with non-linearity, external dependencies, and evolving grid conditions. These limitations lead to operational inefficiencies, grid instability, and increased financial risks. This study investigates the effectiveness of traditional statistical models, advanced machine learning techniques, and hybrid forecasting approaches to enhance prediction accuracy and grid performance. Using real-world datasets from Kaggle, including historical energy generation, consumption, pricing, and weather variables, this research …

Uvm Based Verification Of Ee621 Risc Processor Module, Collin Neidel

Theses

In the process of digital design, the majority of efforts are focused on verification. This is because a proper verification environment can help engineers discover bugs that otherwise would have flown under the radar. On the other hand, a poorly designed verification environment could fail to fully confirm the device’s functionality and leave the customer with a buggy mess. For obvious reasons, this should be avoided at all costs, hence the emphasis on functional verification and validation efforts. Throughout time, the complexity of the average design under test (DUT) has increased dramatically, making verification a significant challenge for today’s verification …

Enhancing Influence Estimation In Gradient Boosted Decision Trees Through Hierarchical Analysis, Anurag Choubey

Theses

This report presents a novel influence function for gradient-boosted decision trees (GBDTs), a widely-used class of predictive models. Influence estimation aims to quantify how individual training samples affect a model’s predictions, offering valuable insights for model debugging, data quality analysis, and in- terpretability. Existing influence functions for GBDTs—such as LeafInfluence, LeafInfluenceSP, and BoostIn—have shown varying degrees of success, with BoostIn currently recognized as the state-of-the-art in terms of estimation quality and computational efficiency. In this work, we propose BoostInLCA, a new influence function that extends BoostIn by incorporating information from non-leaf nodes via the Lowest Common Ancestor (LCA) path, thereby …

Classification Of Linear Systems Of Equations For Quantum Computing Implementation, Mark Danza

Theses

Drawing nearer to an error-corrected era of quantum computing, it is necessary to understand the suitability of certain post-NISQ algorithms for practical problems. One of the most promising, applicable, and yet difficult to implement in practical terms is the Harrow, Hassidim and Lloyd (HHL) algorithm for linear systems of equations. An enormous number of problems can be expressed as linear systems of equations, from machine learning to fluid dynamics to electrical circuit analysis. However, in most cases, HHL will not be able to provide a practical, reasonable solution to these problems. This work seeks to determine whether problems can be …

Sales Opportunities Lead Qualification In B2b Market, Hajar Hussain Ahli

Theses

This thesis addresses the challenge of inefficient lead qualification in the business-to-business (B2B) market by applying machine learning techniques to predict the likelihood of winning a sales opportunity. Using a real-world dataset of over 78,000 records and 17 variables, the study aims to improve how sales teams identify and prioritize high-conversion leads. A thorough data preparation process was conducted, including handling of missing values, outlier detection, and under-sampling to resolve class imbalance between won and lost opportunities. After thorough data cleaning, preprocessing, and under-sampling to address class imbalance, five machine learning models were developed: Logistic Regression, Random Forest, Neural Network, …

Intellimad: A Framework For Secure Machine Learning Models Evaluation And Fine-Tuning In Federated Setting, Dmitrii Korobeinikov

Theses

In contemporary artificial intelligence (AI) appli- cations, Machine Learning (ML) models are core components enabling AI-driven functionalities, yet selecting and fine-tuning a model and its hyperparameters remains challenging. ML model architecture, as well as key model training parameters, such as the number of training epochs, batch size, and learning rate, are highly dependent on both the dataset modalities and the specific task resolved in a particular application. More sophisticated execution setups may require determination of additional environment-related parameters, such as identifica- tion of computational capabilities required for execution of a particular AI-driven task, or discovery and establishment of desired security-related …

Developing Meld-Accelerated Molecular Dynamics Protocols To Simulate The Binding Of The P53-Derived Ligand To The Mdm-2, X Protein, Maria Ciko

Theses

In this study, we focus on developing computational methods to predict protein-ligand binding affinities, with applications in peptide drug discovery. Molecular Dynamics (MD) simulations can capture the complex conformational behavior of proteins, but their high computational cost limits their efficiency. MELD, or Modeling Employing Limited Data, is a Bayesian approach that integrates external information to accelerate sampling of low- energy, high-probability conformations. Building on previous work by Morrone et al., which successfully applied MELD to P53-MDM2 complexes, we hypothesize that we can effectively compute the relative binding affinities while reducing steric clashes and mitigating the effect of slowed diffusion on …

Predicting The Probability Of Crime Related Danger In Los Angeles, Joshua Oghenetega Okpako

Theses

This thesis examines the use of advanced machine learning to predict crime danger in Los Angeles, where 2023 violent crime rates (503 per 100,000) surpass the national average (363.8 per 100,000). Rooted in theories like social disorganization and victim vulnerability, it addresses the lack of combined victim-centric modeling by focusing on three questions: (1) Do blended ensemble models outperform individual models in predicting crime danger? (2) Can unsupervised learning techniques enhance supervised models’ accuracy through label generation or augmentation? (3) How can we interpret accurate machine learning models' decision-making processes? Using historical crime data from Los Angeles (2020–2025) and various …

Methods Of Quantum Circuit Simulation: A Comprehensive Comparative Analysis, Anthony Peter Bacchetta

Theses

The advent of quantum computing has ushered in a new era of computational capabilities, promising to solve problems that were previously intractable for classical computers. Quantum circuit simulation is an essential aspect of harnessing the potential of quantum algorithms, and understanding quantum systems. This thesis presents an exploration and comparative analysis of various methods used in the simulation of quantum circuits, aimed at providing a comprehensive understanding of their strengths, weaknesses, and practical applications. This starts with a dive into the foundational concepts of quantum circuits, quantum gates, and the fundamental principles underlying quantum computation. Quantum circuit simulation emphasizes the …

Modelling And Load Frequency Control (Lfc) Design Of Microgrid Frame-Worked As Technological Norm To Multi Energy System (Mes), Mekonnen Shewarega Worku

Theses

In today's world, the trend of using renewable energy sources as an alternative energy source is becoming more and more common due to various driving factors, such as energy scarcity (fossil fuel depletion, etc.) and environmental issues (carbon footprint). This trend is leading to a steady increase in the penetration of renewable energy sources (RESs) through microgrid (MG), which is formed from combination of various distribution energy sources (DERs). The inherent intermittent nature of RESs coupled with abrupt load changes can instigate sustained frequency fluctuations and can keep the frequency deviation out of the allowable range that leads to an …

An Investigation Of Electrical Surface Impact As A Three-Dimensional Low-Energy Defibrillation Method, Rhiannan Ruef

Theses

Ventricular fibrillation is a medical emergency that leaves the heart unable to beat properly, and is fatal within minutes if left untreated. Furthermore, it’s the leading cause of sudden cardiac arrest according to the NIH, killing over 400,000 Americans every year. Previously, we’ve proposed a new low-energy defibrillation method to restore the heart to its normal rhythm during fibrillation. Our method required further investigation due to the presence of several important dynamical processes. In this dissertation, we report the results of these investigations. First, we will present our results related to a novel behavior called “flopping,” which occurs when a …

A High-Efficiency Ldo Regulator With Adaptive Psrr And Transient Enhancements, Daniel Zeznick

Theses

Reliable power delivery is essential for all computing systems, particularly those operating with limited or constrained energy sources, such as batteries. Power management circuits must provide energy efficiency, stability, and resilience to disturbances to support accurate and consistent system performance. This thesis presents a high-efficiency low-dropout (LDO) regulator featuring adaptive power supply rejection ratio (PSRR) and transient response enhancement techniques. By leveraging adaptive analog design strategies, the proposed regulator dynamically boosts performance at high load currents, maintaining optimal efficiency across the full load range while circumventing key trade-offs inherent to conventional LDO architectures. The design incorporates several novel circuit techniques …

Assessing Readiness For Transformation From Rulebased To Ai-Based Chatbot In Uae Healthcare: A Case Study Of A Rehabilitation Hospital In Abu Dhabi, Mubarak Alketbi

Theses

This study investigates the readiness for transforming rule-based chatbots to AI-based chatbots in UAE healthcare, examining a rehabilitation hospital in Abu Dhabi through quantitative research involving healthcare professionals (N=96) and technical analysis. Findings revealed positive perceptions of the current system alongside enhancement opportunities through AI capabilities, with perceived usefulness strongly correlating with behavioural intention, high service quality ratings for empathy and responsiveness, midcareer professionals demonstrating the highest AI acceptance levels, and system integration identified as the highest priority implementation area.

The research contributes to healthcare technology transformation knowledge in the UAE by providing a structured implementation framework addressing technical requirements, …

A Data-Driven Recommendation System For Selecting The Appropriate Mode Of Learning And Instructional Tools Based On Course Characteristics, Ayisha Manzoor

Theses

The rapid transformation of educational delivery methods during the COVID-19 pandemic required institutions to transition between online, hybrid, and offline learning approaches, creating both challenges and opportunities for educators and students. While online and hybrid learning modes ensured continuity, their effectiveness across different course types remained uncertain. This thesis addresses this gap by developing a datadriven recommendation framework that predicts Course Learning Outcome (CLO) achievement scores using regression, and recommends the most appropriate learning mode (online, hybrid, or offline) along with instructional tools based on course characteristics. This study analyzed 100 undergraduate and postgraduate courses from the College of Information …

Monitoring Sand Migration In Al Ain City Utilizing Remote Sensing Techniques, Sanad Salem Mohammad Fareaa

Theses

Sand migration significantly impacts urban development, infrastructure, and ecosystems in arid regions such as Al Ain city, United Arab Emirates (UAE). This study employs advanced remote sensing techniques to monitor and predict sand dune migration across the Sieh Al Hama dune field, a critical area west of Al Ain. The research objectives include quantifying dune migration rates over three years (2018–2020) using monthly Sentinel-2 satellite imagery, identifying distinct dune fields, and analyzing the textural and mineralogical properties of dune sediments to infer their provenance. Field sampling from four dunes (Large Sabra Dune, Dune 1, Dune 2, and Dune 3) was …