Digital Commons Network^™

Towards Reliable And Trustworthy Deep Learning Through Explainability And Interpretability, Dipkamal Bhusal

Theses

Deep neural networks achieve state-of-the-art performance across many domains, yet their deployment in high-stakes settings is constrained by two challenges: opaque decision-making and vulnerability to adversarial manipulation. This thesis investigates explainability and interpretability as principled mechanisms for improving the reliability and trustworthiness of deep learning models. First, we develop new post-hoc explanation methods that improve feature attribution and concept-based explanations. These methods provide faithful decision cues by modeling meaningful feature interactions and extracting faithful coherent concepts, enabling more reliable understanding of why a model predicts a given label. Second, we show that explanation quality is not solely a property of …

Hardware Integrity Checking On An Fpga Through Power Side-Channel Analysis, Ethan Vuong

Theses

FPGAs have seen extensive usage in applications such as cloud-computing, hardware acceleration, mobile devices, and military alike. While the reconfigurability of these devices allow them to be as adaptable as they are fast, it raises concerns of adversaries modifying not mere software, but hardware itself. Moreover, designers face an IP trust issue where they cannot be sure that a third-party IP was not modified in transaction, programming, or even post-programming. Cloud computing centers are hesitant to rent fabric on multi-tenant FPGAs due to the plethora of vulnerabilities and uncertainties that come with allowing users to reconfigure hardware. This thesis aims …

Software Vulnerability Recidivism In Open-Source Projects, Brandon Keller

Theses

Software vulnerabilities present a major threat to businesses and individuals alike and it is therefore critical that a culture exists among software engineers to encourage the discovery and patching of security flaws. Vulnerability counts are a common way of evaluating a project’s security. However, this metric can run counter to building a developer culture of fault recognition if more vulnerabilities is always seen as a bad thing. While these counts can present a rough idea of a project’s history with security, they provide no insight into how the development team improves and learns as a result of a vulnerability. A …

Leveraging Volatile Ecram Dynamics For Short-Term Plasticity In Neuromorphic Circuits, Sean Borkholder

Theses

Short Term Plasticity (STP) is fundamental for information processing and computational efficiency within biological neural systems. STP has previously been implemented at the circuit level using complex designs with high power and area overheads, resulting in designs that are not scalable in neuromorphic systems. Electrochemical random-access memory (ECRAM) devices naturally exhibit STP behavior through volatile ion dynamics, creating transient conductance modulation.  In previous ECRAM implementations, this behavior was seen as an undesirable artifact of device programming when implemented as a Compute in Memory (CIM) device; however, this thesis proposes leveraging the volatile  behavior to instead act as a computational resource. …

Efficiency Evaluation Of Water Pumping Stations Using Data Envelopment Analysis (Dea), Rowdha Abdullah Alblooshi

Theses

Water pumping stations are a critical component of water transmission systems, ensuring reliable delivery of potable water while maintaining operational requirements and international standards. With the increasing focus on sustainability and energy optimization, improving the efficiency of pumping stations has become a key priority. However, there is a lack of structured benchmarking approaches to evaluate the relative performance of pumping stations across multiple operational factors. This study addresses this gap by applying a Data Envelopment Analysis (DEA) framework to evaluate the efficiency of water pumping stations. A multi-model approach is adopted, including the CCR and BCC models, along with the …

Optimizing Urban Commute Quality Through Traffic Congestion Analysis And Predictive Modeling, Obaid Almansoori

Theses

Urban traffic congestion imposes significant economic, environmental, and social costs on rapidly growing cities worldwide. This research investigates how predictive analytics and machine  learning can be leveraged to classify and forecast traffic congestion severity in real time,  enabling data-driven decision-making for transportation planning, signal optimization, and  congestion management. A real-world traffic monitoring dataset comprising 5,952 observations collected over two months via  computer vision sensors at an urban intersection was analysed under the CRISP-DM frame- work. The  dataset records counts of four vehicle classes including cars, bikes, buses, and trucks at  15-minute intervals, alongside temporal variables such as time of day, …

A Data-Driven Machine-Learning Framework For Intermittent Demand Classification And Forecasting Of Electrical Spare Parts In Dubai’S Water Pumping Stations, Ayesha Khamis

Theses

Irregularity in spare parts demand has been a recurring problem in many critical industries. The same problem is found in Dubai's water pumping stations, where demand is highly intermittent, with long periods of no usage followed by sudden increases. These irregularities are usually caused by maintenance activities or equipment failures. Forecasting such demand is challenging, as irregular patterns can lead to stockouts or overstocking. In this research, a machine learning (ML) forecasting framework is developed to handle intermittent demand for electrical spare parts in Dubai's pumping stations. The framework includes demand classification and prioritization, as well as the application of …

Imaging Performance Analysis Of Euv Mask Stacks For Sub-24 Nm Features At High And Hyper Numerical Aperture, Abdusame Omran N. Berish

Theses

As extreme ultraviolet (EUV) lithography advances toward high numerical apertures (NAs) of 0.55 and 0.75, mask-3D effects degrade image fidelity, limiting resolution and the process window. This work investigates the performance of alternative EUV absorber materials for sub-12 nm half-pitch patterning under high-NA and hyper-NA conditions. The lithographic simulations were conducted using the PROLITH Maxwell rigorous model under dipole illumination optimized for sub-24 nm resolution dense line/space features. The EUV mask stacks were derived from five fabricated and tested absorbers: TaBN, Ru/Ta, Pt2W, Ni/CrN, and TaSi2.  The absorber thickness was recalibrated to an optimal value, ensuring a maximum normalized image …

Full-Field Experimental Analysis Of 3d Printed Open-Hole Plates Reinforced With Isotropic And Concentric Carbon Fibers, Mozah Alyammahi

Theses

Additive Manufacturing technology (AM) has become an attractive process for innovative products in recent decades. The applications of AM materials widen to include composite materials. The fabrication of composite materials via AM makes the process more cost-effective than traditional fabrication methods. Nowadays, there are 3D printers that can produce composite materials with different strategies which results in different mechanical properties. This study examines the impact of carbon fiber reinforcement, both isotropic and concentric, on 3D printed open-hole plates utilizing full-field experimental analysis.  Problem Statement  The research on investigating different fiber reinforcement is increasing rapidly. However, most of the studies are …

Cancer Detection System Using Binary Neural Network On Dna Based Architecture, Antar Narayan Chowdhury

Theses

Deoxyribonucleic acid (DNA) is among the most durable chemical storage media, capable of encoding the fundamental instructions for protein synthesis (the central dogma). Each human cell contains a unique DNA sequence characterized by identifiable markers that facilitate pattern recognition. These molecular features offer significant potential for personalized drug development and disease identification. Recent advancements in DNA based research have demonstrated that fundamental arithmetic operations can be executed directly through molecular interactions, bypassing the need for silicon-based computational assistance. These biochemical applications can be further scaled to support Binary Neural Network (BNN) models, which are particularly well-suited for mitigating stochastic noise …

Effect Of Geometry, Cell Size, And Carbon Fiber Reinforcement On The Charpy Impact Strength Of Additively Manufactured Tpms Lattice Structures, Ahmed Yousuf Mohammad Bin Yaroof (Alsuwaidi)

Theses

Scientific research indicates a growing utilization of lightweight, high-performance materials across various disciplines, including engineering, driven by the demands of modern technological advancements. Recent developments within these fields include the creation of advanced lattice structures through additive manufacturing (AM) processes. One such lattice structure is the triply periodic minimal surface (TPMS) structure. TPMS structures possess unique mechanical properties and energy absorption characteristics that distinguish them from conventional AM lattice structures. However, published research to date has largely focused on the quasi-static behavior of TPMS structures, with limited attention given to the impact performance of composite- reinforced TPMS structures. The purpose …

Preparing Future-Ready Graduates For Technological And Workforce Transformations, Ghia El Dirani

Theses

The United Arab Emirates (UAE) is undergoing rapid economic transformation driven by technolog- ical innovation and national strategies such as Vision 2031 and UAE Centennial 2071, positioning STEM (Science, Technology, Engineering, and Mathematics) education as critical to building a knowledge-based economy. However, a persistent gap exists between the competencies developed in STEM higher education programs and the skills demanded by emerging sectors such as artificial intelligence, renewable energy, and advanced manufacturing. While the UAE has introduced pro- gressive education policies and invested in digital infrastructure, most curriculum reforms remain reactive and disconnected from long-term workforce projections. This research applies strategic …

Investigation Of Machining (Drilling) Of Bio-Composite Reinforced With Jute Fibers Under Different Machining Conditions, Mohammed Abdul Mujeeb Ansari

Theses

This study investigates the drilling performance of jute fibre reinforced bio-composites under different machining conditions to reduce performance parameters like surface roughness, tool type and delamination damage (entry and exit). Jute composites were fabricated using a process called hand layup. Drilling experiments were carried out using full factorial method focusing on three parameters: Feed Rate (0.05mm/rev, 0.1mm/rev and 0.15mm/rev), Drill bit Type (Titanium Nitride-TiN, High Speed Steel-HSS and High-Speed Steel Cobalt HSS-Co) and Lubrication condition (Dry, Minimum Quantity Lubrication-MQL and Cryogenic-LN2). A total of 27 experimental trials were performed, and the responses were measured for each condition. Statistical analysis was …

Predicting Property Sale Prices In Dubai Using Machine Learning Regression Models, Mohammad Buabdulla

Theses

It has been observed that the fastness of digital real estate’s application has resulted in availability of large-scale property data which forms an opportunity to more precise and open property valuation practices. A lot of conventional real estate appraisal methodologies with much emphasis on manual evaluation and historical comparative value find it difficult to respond to the market dynamics or the dynamics that exist within the market at an alarming rate. This research will help mitigate these shortcomings by generating and testing machine learning regression models to determine the price of residential property at sale in Dubai by using real-life …

The Best Practices For Prefabrication Of Industrial Buildings, Derek Gines

Theses

Prefabrication has demonstrated measurable and repeatable advantages in productivity, cost certainty, and environmental performance, yet its adoption within industrial building typologies remains largely inconsistent. Existing research largely evaluates prefabrication through downstream performance outcomes, while offering limited insight into the upstream design and organizational decisions that enable or undermine its reliability. This thesis reframes prefabrication as a design‑led methodology rather than a construction optimization, arguing that successful hybrid prefabrication is determined primarily by early decision timing, governance structures, and the control of spatial and logistical interfaces. This study adopts a qualitative design‑research approach that combines comparative case study analysis with expert …

Predicting Luxury Car Sales Using Machine Learning: A Comparative Study Of Linear Regression, K-Nearest Neighbors, And Support Vector Machines, Khalifa Jamal Mohammad Saleh Alblooshi

Theses

This paper explores the use of machine learning to predict the sales of luxury cars in the globe, and BMW as a case study of its sales data in the global market between the period 2010 and 2024. The study focuses on three regression algorithms, which include the Linear Regression, K-Nearest Neighbours (KNN) and Support Vector Machines (SVM) and sees their predictive accuracy, generalisation performance and business applicability. In Python with the help of the Google Colab, an end-to-end analytical pipeline was developed entailing data preprocessing, outlier management, feature engineering, and time-sensitive traintest division. RMSE, MAE, MAPE, and R2 were …

Toward Reliable Computational Social Science: Inconsistency-Aware Methods For Human Annotation And Ai Inference, Sujan Dutta

Theses

As artificial intelligence (AI) becomes increasingly common in computational social science, \textit{inconsistency} has emerged as a key challenge. AI models often contradict themselves when given equivalent inputs, disagree with other models on the same data, and diverge from human judgments in seemingly opaque ways. Human annotators exhibit their own inconsistencies, both within individuals and across groups shaped by differing values and identities. Rather than treating these inconsistencies simply as noise, this dissertation argues that they contain meaningful signals that can be leveraged to improve learning efficiency, strengthen evaluation, and increase the reliability of large-scale social measurement. To study this phenomenon, …

Toward A Unified Framework For Open World Visual Learning, Yuansheng Zhu

Theses

Artificial intelligence systems have achieved remarkable performance across a wide range of visual tasks. However, most existing models operate under the unrealistic closed-world assumption, where training and test data are drawn from the same distribution. In real-world applications such as anomaly detection, autonomous driving, and medical diagnosis, learning systems frequently encounter novel or out-of-distribution scenarios. These settings require models that can recognize unknown inputs, adapt to new information over time, and maintain reliable performance under evolving conditions. This dissertation studies the problem of Open World Visual Learning, a paradigm that enables visual learning systems to operate robustly in dynamic and …

Llm-Driven Mission Control And Autonomous Planning For Search-And-Rescue Uavs: A Simulation-Based Evaluation, Naser Bader Alsaedi

Theses

Unmanned aerial vehicles (UAVs) are increasingly used in search‑and‑rescue (SAR) missions, yet many systems still rely on fragmented software where mission design, perception, and flight control are configured separately. This thesis examines whether a unified AI‑driven framework can reduce configuration effort and operator workload in UAV‑based SAR operations. The proposed system integrates natural‑language mission specification using a large language model (LLM) (LLaMA 3.1), autonomous coverage planning, YOLOv8‑based victim detection, and PX4/MAVSDK control within a single architecture. Operators describe missions through free‑form text or a graphical interface; the model converts these descriptions into structured mission parameters that are automatically planned and …

Efficient Fpga Implementation Of A 1 Million-Point Fft, Jwaher Abdulqader Al Tamimi

Theses

The one-million-point Fast Fourier Transform is implemented using a radix-2 single-path delay feedback pipeline architecture. To minimize the computational overhead, twiddle factors were pre-computed and stored in memory. The design uses a fixed-point representation with two integer bits and seven fractional bits, achieving a measured signal-to-noise ratio of 37.98. Given the substantial memory requirements, a memory partitioning approach was used. It mapped the delay buffers in each stage lookup table memory, block random-access memory, or ultra random-access memory, based on word width and memory depth.

The implementation operates successfully at 100 megahertz on a mid-scale field-programmable gate array. Post-implementation reported …

Optimizing Virtual Scrolling Performance In Angular: A Comparative Study Of Cdk And Custom Implementation, Guri Sokoli

Theses

Modern web applications often display large datasets with tens of thousands of items, such as e-commerce catalogs, data tables, and social media feeds. Rendering all items in the Document Object Model (DOM) at once causes browser freezing, high memory use, and slow interfaces. Virtual scrolling solves this problem. It is widely adopted but rarely studied through direct performance comparison. Few empirical studies measure how different implementations behave under varying dataset sizes, devices, or browsers. This research conducts a comparative analysis of Angular CDK Virtual Scroll as an industry-standard baseline and develops an optimized implementation incorporating framework-specific enhancements: OnPush change detection …

An Artist’S Playground: Building Bridges Between Children’S Education And Their Knowledge Of Art Through Multi-Modal Learning, Alice J. Cauchi

Theses

Not all museums welcome children as part of their main audience with open arms. Research suggests that experiencing museums as a child supports classroom learning and development, but only when the museums engage young children properly. This paper explores how incorporating multi-modal interactives into art museums can effectively engage children, ages 4 to 8, and build a foundation that supports lifelong learning. Drawing on ideas from early-childhood education, museum visitor-engagement studies, and interactivity sciences, this project results in a framework that guides art museums on how they can overlap play, interactives, and content to incorporate this age group effectively to …

Nlp Crowdsourcing For Predominantly Oral Languages: The Case Of Bambara, Allahsera Auguste Tapo

Theses

Predominantly oral languages (POLs) face a significant "digital divide," as they are often excluded from the benefits of modern natural language processing (NLP) technologies, due to a lack of extensive, readily available machine learning (ML) datasets. We investigate methods to overcome this data scarcity for Bambara, a Manding language, spoken primarily in Mali, with a rich oral tradition but limited digital presence.     The research leverages crowdsourcing and community engagement to build high-quality ML ready dataset resources. Key contributions include methods for automatic speech recognition (ASR) and machine translation (MT) dataset collection and curation and for educational resource creation.      Our findings …

Beyond Anomaly Detection: Classifying Attacker Automation Level From Ssh Honeypot Behavioral Signatures, Ashley Alt

Theses

The proliferation of AI-assisted offensive tools has introduced a new category of cyber attacker that combines the speed of automation with the adaptive reasoning previously associated only with skilled human operators. Despite the richness of behavioral data captured by SSH honeypots, existing analyses treat interaction logs primarily as evidence of malicious activity rather than as a dataset capable of distinguishing between attacker types. This thesis investigates whether human-driven, traditionally automated, and AI-assisted attackers produce distinguishable behavioral signatures within SSH honeypot interactions, and whether machine learning techniques can reliably classify attacker behavior from session-level features. A controlled experimental architecture was developed …

Traditional And Machine-Learning Equalization Techniques For Bandwidth-Limited Short-Reach Optical Communication Channels, Abdullah Khawatmi

Theses

This thesis investigates equalization techniques for bandwidth-limited short-reach optical communication systems, with a focus on Visible Light Communication (VLC) and Step-Index Plastic Optical Fiber (SI-POF) links. Commercial light-emitting diodes and photodiode receivers impose severe bandwidth constraints, inter-symbol interference, and noise sensitivity, which fundamentally limit achievable data rates. The work addresses these impairments through systematic evaluation of traditional digital signal processing–based equalizers and modern machine-learning-based post-equalization methods. The primary aim of this thesis is to enhance the achievable data rate and reliability of commercial short-reach optical links while maintaining practical computational complexity. Specifically, the objectives are to (i) design and experimentally …

A Computational Assessment Of Halobacterium Salinarum Glutamate Dehydrogenase Enzymes, Maaz Abdalla

Theses

Glutamate dehydrogenase (GDH) is a hexameric enzyme. GDH is involved in several pathways and cellular processes such as oxidation-reduction homeostasis, ammonia metabolism, lipid biosynthesis, insulin and lactate production, and acid-base equilibrium. The main objective of this thesis is to understand the structural and biochemical properties of this enzyme and why some organisms, for e.g., Halobacterium salinarum, have more than one GDH with different coenzyme specificities. The catabolism of glutamate is linked to NAD+ specific GDHs, meanwhile, NADP+ specific GDHs play an anabolic role in ammonia assimilation. Molecular docking and binding free energy calculations were employed followed by long-scale (500 nanoseconds) …

Analysis Of A Coaxial Transmission Line Filled With An Orthorhombic Dielectric-Magnetic Medium, Fathima Manikunnath Abdul Akbar

Theses

Coaxial transmission lines are fundamental means for Transverse Electromagnetic (TEM) wave propagation in RF, microwave and high-speed electronic systems. The study of transmission lines is often familiar when they are filled with isotropic materials; however modern engineered direction dependent materials reshape field distributions. In this thesis, we consider a coaxial transmission line of an inner radius and outer radius b filled with an orthorhombic dielectric-magnetic material, which is described by two anisotropy parameters αx and αy. The potential and field distributions are studied in relation to the ratio b/a as well as the anisotropy parameters αx and αy. Due to …

Breath-Based Detection Of Liver Cancer Biomarkers Using An Swcnt-Fet Nano-Biosensor: Quantumatk, Mohamed Mohieb Rashdan

Theses

Early detection of liver cancer remains limited by the slow pace and invasiveness of current testing methods. This study proposes a single-walled carbon nanotube field-effect transistor (SWCNT-FET) designed to detect hexanal—a volatile organic compound (VOC) elevated in liver cancer—directly from exhaled breath. The device is modeled in QuantumATK using a semi-empirical Extended Hückel Hamiltonian within the non-equilibrium Green's function (NEGF) framework, emphasizing realistic contact physics by employing metallic SWCNT electrodes instead of conventional metal films. Zigzag channels with (11,0) and (12,0) chiralities are examined to analyze how geometry and contact matching influence charge transport. Simulations include current–voltage (I–V) characteristics and …

Development Of A Novel Cfd Approach For Predicting Multiphase Flow Enhanced By Machine Learning, Abdulrahman Elazazi Salem

Theses

Accurate estimation of interface orientation is crucial for ensuring both the accuracy and robustness of Volume of Fluid (VOF) schemes in multiphase flow simulations, especially when using non-uniform Cartesian meshes. Conventional gradient reconstruction approaches, such as least-squares (LSQ) methods, often suffer from significant errors and strong oscillations on highly stretched grids. This work proposes a grid-transferable, learning-based methodology that predicts interface unit normal vectors directly from the local volume-fraction field on nonuniform structured Cartesian grids using a feedforward neural network. The methodology extends earlier work originally developed for uniform grids, which is first reproduced to assess how performance deteriorates in …

Modeling And Autonomous Control Systems For A Delivery Drone Application, Dana Helal Alnuaimi

Theses

This thesis focuses on the development of a multi-drone navigation and control system, aiming to enhance payload capacities beyond the limits of single-drone systems. By integrating multiple drones to work collaboratively as a unit, this study addresses the challenges associated with lifting and transporting heavier payloads.

The primary objective is to design and evaluate a multi-drone system capable of working in tandem to transport larger payloads efficiently. The research aims to develop robust control algorithms, supported by system identification for dynamic modeling, and navigation strategies to enable effective coordination between drones.

The study employs a combination of simulation and real-world …