Computer Sciences Commons^™

Qlorax: Heuristic-Guided Fine-Tuning Of Llama-2 For Domain Adaptation In Entrepreneurship, Gaurob Saha

Theses and Dissertations

This thesis presents a study on the fine-tuning of large language models (LLMs) for domain-specific applications using limited data. We fine-tuned the LLaMA-2 (7B) model on a curated entrepreneurial dataset containing 3,545 human-written question-answer pairs, of which 3,095 were used for training and 450 were reserved for evaluation. A complete fine-tuning and evaluation pipeline was developed, which included clustering human-written answers, generating centroid-based summaries for each cluster, and evaluating the model's generated responses through cosine similarity.

Training was carried out over five epochs, with model performance evaluated after each epoch. The fine-tuned model demonstrated strong semantic alignment with human-written content, …

Exploratory Data Analysis (Eda) And Predictive Machine Learning (Ml) For Buildings’ Energy Fault Detection, Akshith Nukala

Theses and Dissertations

Building energy load fault detection is a critical challenge in energy usage analysis. It helps uncover energy wastage, machinery/appliance degradation or inefficiency, and failures or faults in buildings’ HVAC (heating, ventilation, and air conditioning) systems. Early identification of machinery failure and energy wastages due to operational maintenance negligence in large sites such as campus buildings is indispensable for achieving energy efficiency. This is crucial for saving patrol and minimizing the response time to restore the building appliances or systems to their optimal state.

Advancements in state-of-the-art AI/ML data-driven algorithms and techniques enabled us to build accurate, efficient and scalable fault …

Multitask Learning For Named Entity Recognition And Relationship Extraction, Adrienne D. Hembrick

Theses and Dissertations

Information Extraction (IE) is a fundamental task in Natural Language Processing (NLP), involving the identification of structured information from unstructured text. Two core components of IE—Named Entity Recognition (NER) and Relation Extraction (RE)—are widely used to extract key concepts and the relationships between them across various domains. However, the sequential dependency of RE on the output of NER makes it vulnerable to error propagation: inaccuracies in entity recognition can negatively affect downstream relation extraction.

To mitigate this issue, Multitask Learning (MTL) has been proposed as an approach that jointly models NER and RE, aiming to improve overall performance and reduce …

Exploring Brent-Kung Adders In Balanced Ternary Cmos Logic, Astrea J.C. Rojas

Theses and Dissertations

The modern computer operates on a 64-bit architecture. These devices can store large numbers and precise decimals, but more advanced devices are needed to support progressing technologies every day. A more efficient system with higher speeds and larger operable numbers would be a key to optimization of computation as we know it. The ternary device, operating in base-3, has the potential to be that optimization. However, binary technology has such precedent and research that it is a difficult gap to span to compare the ternary system to the modern binary system. With a more advanced adder and optimized gates using …

Uso De Herramientas De Inteligencia Artificial Generativa Por Parte De Docentes En Una Escuela Del Suroeste De Puerto Rico, Glorimar N. Rodríguez Guiliani

Theses and Dissertations

Esta disertación aplicada fue diseñada para investigar el nivel de conocimiento, uso y dificultades que enfrentan los docentes de quinto a duodécimo grado en una escuela privada del suroeste de Puerto Rico respecto a tecnologías emergentes las cuales presentan desafíos significativos para los docentes y los estudiantes. Se exploró la utilización de herramientas de inteligencia artificial generativa (GenAI) como ChatGPT dentro y fuera del aula para actividades pedagógicas y administrativas.

Los hallazgos revelaron una notable carencia en el conocimiento docente sobre el uso y habilidades de la inteligencia artificial. Se identificó, también, una deficiencia en la capacidad de los docentes …

Knowledge And Ontology Enhanced Approach To Natural Language Understanding (Koe-Nlu) In Computational Social Media And Healthcare, Naga Usha Gayathri Lokala

Theses and Dissertations

Natural Language Understanding (NLU) faces both opportunities and challenges as the amount of social media and healthcare data grows. This is particularly evident in context-sensitive applications such as evaluating cognitive health, identifying mental health symptoms, and monitoring drug abuse. Even though traditional NLU models work well for processing language in a wide range of areas, they often lack the ability to understand language in a specific domain, reason in context, and incorporate structured external knowledge. This dissertation talks about the Knowledge and Ontology Enhanced Approach to Natural Language Understanding (KOE-NLU), a new framework that is meant to make NLU systems …

Machine Learning Models Leveraging Patient-Similarity And Clinical Temporality For Disease Prognoses, Ahmad F. Al Musawi

Theses and Dissertations

Electronic Health Records (EHRs) constitute a comprehensive and high-dimensional repository of clinical data, encompassing a wide array of patient-level information such as diagnoses, procedures, medications, laboratory results, and unstructured clinical narratives. These data hold immense potential for advancing predictive modeling in healthcare, including tasks such as disease progression modeling, hospital readmission prediction, and length of stay (LoS) estimation. However, the intrinsic complexity of EHR data—manifested in its heterogeneity, sparsity, and temporal dynamics—poses significant analytical challenges that limit the generalizability and interpretability of conventional machine learning models. Recent methodological advancements in deep learning and graph-based learning, particularly Graph Neural Networks (GNNs), …

Relationship Extraction Using Retrieval Augmented Generation For Biomedical Dataset, Jannat .

Theses and Dissertations

With the increasing number of structured and unstructured data, obtaining reliable information effectively has become crucial. In the biomedical domain, extracting information from the scientific papers is crucial in order to stay up-to-date with accurate information, given the increased pace by which new research studies are published. This work focuses on identifying relationships between entities that are extracted from the abstracts and titles of biomedical research papers. In this work, we developed a Retrieval Augmented Generation (RAG) based system to automatically identify relations between biomedical entities. We evaluate multiple open source Large Language Models (LLMs) and the number of examples …

Learning From Non-Stationary Data Streams, Gabriel Jonas Aguiar

Theses and Dissertations

The rapid growth of data from sources such as mobile applications, sensors, and network monitoring has increased the need for machine learning algorithms capable of handling non-stationary data streams. However, learning from such streams presents significant challenges due to their evolving nature and the presence of concept drift. One of the most complex issues is learning from imbalanced data streams, where shifting data distributions, combined with feature space drifts, complicate continuous adaptation. These challenges become even more pronounced in multi-class scenarios, which are common in real-world applications. Detecting concept drift in such contexts is particularly demanding, as it requires tracking …

Microservice Architecture For Social Media Data Collection, Analysis, And Dashboarding, Sai Ram Manohar Koya

Theses and Dissertations

This research presents a novel methodology for the collection, processing, and analysis of social media data using a microservices-based architecture. The proposed system integrates multiple data streams from various social media platforms, transforming this information into a unified, JSON-based DataObject model for seamless processing and analysis. Unlike monolithic architectures, the microservices approach offers scalability and flexibility, allowing the system to handle the high velocity, variety, and volume of unstructured social media data, including text, images, and videos. By leveraging NoSQL databases like MongoDB, the methodology efficiently manages data in a semi-structured format, supporting real-time analytics such as sentiment analysis, toxicity …

Exploration Of The Gap Between The Secure Web Application Development Competencies Needed By Industry And Those Competencies Provided By Graduates Of U.S. Undergraduate Software Engineering Programs, Gary Allen Harris

Theses and Dissertations

Literature demonstrates that threats and attacks on computer systems and networks have been around since the beginning of computing, and the number, severity, sophistication, and costs of attacks and data breaches are continuing to grow. Several studies suggest that one of the most common causes of data breaches is insecure web applications that contain vulnerable application code. These studies suggest that poor secure web application development practices are a prime cause of the susceptible web applications. Additionally, studies suggest that higher education is not meeting industry’s secure software/web application development needs. Employers have reported that they are not getting the …

Advanced Models For Linking Process In Data Washing Machine, Bushra Sajid

Theses and Dissertations

Entity Resolution (ER) is a critical process in data integration and quality improvement that identifies and links multiple records referring to the same real-world entity. As data volumes and heterogeneity increase, traditional ER methods face new challenges, prompting research into more advanced techniques. The Proof-of-Concept Data Washing Machine (DWM), developed under the NSF DART Data Life Cycle and Curation research theme, aims to automatically detect and correct data quality errors through unsupervised entity resolution. Recent research focuses on enhancing DWM's effectiveness by replacing rule-based methods with machine learning and deep learning approaches, particularly in the linking process. Deep learning models, …

An Open-Source, Student-Centric Approach To The Cyber Kill Chain, Justin Lane Wooten

Theses and Dissertations

The cybersecurity landscape demands professionals with practical skills and a deep understanding of attack methodologies. However, many institutions face significant challenges in providing comprehensive cybersecurity education due to the high costs associated with commercial tools and platforms. This thesis presents a student-centric, open-source curriculum for teaching the Cyber Kill Chain, designed to bridge the gap between theoretical knowledge and real-world application while addressing the financial constraints faced by many educational institutions. Our approach leverages freely available tools and hands-on exercises to cover each phase of the Cyber Kill Chain, emphasizing ethical considerations and collaborative learning. We detail the curriculum development …

Analysis Of Cortical Evoked Auditory Response Detection In Adults Using Machine Learning, Pranavi Beerelli

Theses and Dissertations

This study focuses on the use of machine learning (ML) techniques to automate the detection of Cortical Evoked Auditory Responses (CEARs), which are key in understanding how the auditory cortex processes sound stimuli. Traditionally, analyzing these auditory responses has relied on manual interpretation by audiologists, a process that can introduce variability and human error, particularly in complex cases. To address this challenge, the research utilizes advanced deep learning models, including Convolutional Neural Networks (CNNs), Long Short Term Memory (LSTM) networks, and Bidirectional LSTM (BiLSTM) architectures, to analyze Electroencephalography (EEG) data and classify the presence or absence of auditory responses automatically. …

Improved Vector Pruning For Partially Observable Markov Decision Processes, Thomas Jonathan Bowman

Theses and Dissertations

Exact dynamic programming algorithms for planning problems that are represented as partially observable Markov decision processes rely on a subroutine that removes, or ``prunes", dominated vectors from sets of vectors that represent piecewise-linear and convex value functions. The classic vector pruning subroutine solves one linear program per vector, where the number of variables is equal to the size of the state space and the number of constraints is equal to the number of vectors shown so far to be undominated. Thus, its scalability is limited not only by the number of linear programs it solves, but especially by their size. …

A Comprehensive Performance Evaluation Of Proprietary And Open-Source Language Models In Closed And Open-Domain Tasks, Abhilash Kanduri

Theses and Dissertations

As the field of Natural Language Processing (NLP) continues to evolve, evaluating the performance of both proprietary and open-source language models has become increasingly critical. This research provides a comprehensive analysis of proprietary models like GPT-3.5 Turbo, GPT-4, and GPT-4 Turbo, alongside open-source models such as FLAN-T5, GPT-Neo, and GPT-2. By leveraging traditional metrics like ROUGE and BLEU, as well as custom metrics including ReGrAde, Contextual Precision, and Faithfulness, the study evaluates these models across closed-domain tasks (e.g., factual question-answering) and open-domain tasks (e.g., creative writing and brainstorming). The proprietary models excelled in structured, fact-based tasks, while the open-source models …

Multi-Cloud Identity Security Utilizing Self-Sovereign Identity, Morgan Lee Reece

Theses and Dissertations

With the increasing use of multi-cloud environments, security professionals face challenges in configuration, management, and integration due to uneven security capabilities and features among providers. As a result, a fragmented approach toward security has been observed, leading to new attack vectors and potential vulnerabilities. Other research has focused on single-cloud platforms or specific applications of multi-cloud environments. Therefore, there is a need for a holistic security and vulnerability assessment and defense strategy that applies to multi-cloud platforms. This dissertation explores risk and vulnerability analysis to identify attack vectors from software, hardware, and the network, as well as interoperability security issues …

Deep Learning - Based Automated Detection And Classification Of Foreign Materials In Poultry Using Color And Hyperspectral Imaging, Rohini Maram

Theses and Dissertations

This thesis explores the use of Deep learning for detection and classification of small foreign materials (FMs) in poultry meat using color and hyperspectral imagery (HSI). The study employs You only look once (YOLO) object detection models on color images for precise localization, and one-dimensional convolutional neural network (1D CNN), two-dimensional convolutional neural network (2D CNN) was used on HSI (600 – 1700 nm) for classification. Twelve different FMs commonly known as polymers including PVC, PET, LDPE and HDPE, were examined using 52 color and 52 hyperspectral images. Four YOLO models (v5x, v7x, v8x, v10x) were implemented, trained, tested and …

Enhancing Cybersecurity Strategies Through Automated Cti Extraction, Risk Prioritization, And Privacy-Conscious Information Sharing, Spencer Rian Massengale

Theses and Dissertations

Cybersecurity operations require the ability to collect and analyze large amounts of cyber threat intelligence (CTI) to assess risks and formulate defensive strategies against emerging threats. This task has become increasingly complex due to the rapid evolution of cyber threats and the growing volume of unstructured, natural-language CTI sources. The scale of data and analysis needed to utilize CTI effectively far exceeds humans' manual capacity, especially for organizations with limited resources. This research focuses on leveraging Large Language Models (LLMs) and machine learning techniques to enhance CTI extraction, risk assessment, and data sharing. We utilized LLMs to automate the extraction …

Sd-Weat: Towards Robustly Measuring Bias In Input Embeddings For Artificial Intelligence Language Models, Magnus Gray

Theses and Dissertations

Artificial intelligence (AI) is rapidly transforming industries and markets, from healthcare to entertainment, revolutionizing decision-making processes. However, as AI grow more influential, they also risk amplifying existing biases, potentially leading to harmful consequences. Recent advancements in large language models (LLMs), such as GPT-4 and Llama, have heightened concerns about bias in natural language processing (NLP) tasks, driving the need for robust methods to detect and mitigate bias. Current approaches, such as the Word Embedding Association Test (WEAT) and its sentence-level extension the Sentence Encoder Association Test (SEAT) often fall short in capturing the nuances of biases in the input embeddings …

Lstm-Transformer Based Robust Hybrid Deep Learning Model For Financial Time Series Forecasting, Md R. Kabir

Theses and Dissertations

The inherent challenges of financial time series forecasting demand advanced modeling techniques for reliable predictions. Financial data, belonging to the category of multimedia data, contains an extensive quantity of information that is widely used for data analysis and decision-making tasks. Effective financial time series forecasting is crucial for financial risk management and the formulation of investment decisions. Accurate prediction of stock prices is a subject of study in the domains of investing and national policy. This problem appears to be challenging due to the presence of multi-noise, nonlinearity, volatility, and chaotic natures in stocks. This paper proposes a novel financial …

Mitigating Code Reuse Attacks On Risc-V Binaries: Minimizing Gadget Availability Using The Compressed Extension, Heitor Vieira

Theses and Dissertations

Embedded systems are vital in civilian and military applications, requiring high performance and security. The open RISC-V Instruction Set Architecture (ISA) offers significant advantages, including security through community review and strategic independence in microchip supplies. Brazil’s recent partnership with RISC-V highlights its potential for national technological sovereignty. However, RISC-V is not inherently resistant to code reuse attacks (CRAs), highlighting the need to integrate security measures early in development. The RISC-V Compressed extension, while beneficial for optimizing performance and code flexibility, introduces security trade-offs. As RISC-V adoption grows, particularly in critical systems, addressing these security challenges from the start is crucial …

Dual-Channel Side Channel Attack: Improved Aes Key Decryption By Combining Power And Electromagnetic Side Channels With Convolutional Neural Networks, Sean P. O'Neill

Theses and Dissertations

This research introduces a novel DL approach for SCA that combines power consumption and EM signals to enhance encryption key deduction by leveraging a dual-channel CNN architecture. A new dataset, consisting of simultaneous power and EM signal collections during 128-bitAES encryption, was developed to train and evaluate the model’s effectiveness. The combined approach achieved an 88% reduction in traces needed, from 50 traces to 6, for encryption key classification, outperforming traditional methods such as random forest, DPA, DEMA,and individual side channel CNN models. These findings highlight the potential of integrating multiple side channels in SCA to improve performance without the …

Patterns Of Interactions In Human-Machine Teams, Kazuhiko Momose

Theses and Dissertations

Increasingly capable machines, including Artificial Intelligence (AI) agents are playing a more important role in a wide range of applications, including human daily activities and safety-critical systems. They can benefit even more when humans and such machines agents work together as a team by leveraging each other's strengths and complementing each other to enhance overall performance. To design high-performing teams, it is critical to analyze the team dynamics and understand how humans and machines interact with each other. Collaboration, Coordination, and Cooperation (3Cs) are terms typically used to describe the behavior of teams. However, these terms tend to be used …

Proof Of Concept: Simulating Drone Tracking In A Border Security Context, Jose Ruben Espinoza

Theses and Dissertations

Worldwide availability of drone technology has risen to unprecedented levels within the past century due to its commercial availability. While there has been various positive applications of such technology, it has additionally found usage within security critical contexts. Specifically, there have been reports of illegal drug smuggling along the Mexico-United States border in which quadrocopter based drones have been used. Within our research we aim to showcase, as a proof of concept, that autonomous drone technology can be leveraged within a defensive approach via the usage of reinforcement learning and object detection for security critical contexts. To promote the importance …

Intent Visualization In Human-Agent Teams, Rahul Tushar Mehta

Theses and Dissertations

Despite advances in autonomous systems, effective collaboration between humans and intelligent agents remains a significant challenge, particularly in shared-control scenarios. This study investigates how intent visualization affects human-agent collaboration in telecollaboration scenarios, examining its impact on team performance, operator trust, and workload. Using a custom simulation environment and Wizard-of-Oz methodology, we conducted an experiment with 13 participants who completed exploration tasks under two conditions: a baseline interface and an enhanced interface with intent visualization. Results showed that while intent visualization did not significantly improve objective performance metrics, it led to a 22.6\% increase in explicit disagreements between operators and the …

Decoding Neural Networks: An Information-Theoretic Guide To Interpretability, Error Analysis And Efficiency, Mackenzie J. Meni

Theses and Dissertations

This dissertation addresses critical challenges in neural network design by leveraging entropy-based techniques to improve model efficiency, interpretability, and bias reduction. Focusing on the unique demands of computer vision applications, particularly object detection and classification for real-time systems, this work introduces a series of innovative methods centered on information theory. At the core of these methods is the Probabilistic Explanations of Entropic Knowledge (PEEK) framework, a tool developed to analyze and visualize entropy distributions across feature maps. PEEK offers insights into information flow within neural networks, making it possible to pinpoint layers that contribute meaningfully to decision-making or identify those …

Improving Data Curation With Spectral Clustering And Shannon Entropy: An Unsupervised Approach Within The Data Washing Machine, Erin Hathorn

Theses and Dissertations

In the ever-expanding landscape of digital technologies, the exponential growth of data presents both challenges and opportunities, demanding innovative approaches to data curation. Effective data curation is pivotal for extracting meaningful insights from vast and complex datasets. This study explores the integration of spectral clustering and Shannon Entropy within the Data Washing Machine (DWM), a novel tool designed to streamline unsupervised data curation processes. The DWM incorporates Shannon Entropy into its clustering process, allowing for adaptive refinement of clustering strategies based on entropy levels observed within data clusters. Spectral clustering, known for its ability to handle complex and non-linearly separable …

Innovative Strategies For Healthcare Data Integration: Enhancing Etl Efficiency Through Containerization And Parallel Computing, Ehsan Soltanmohammadi

Theses and Dissertations

The healthcare industry is rapidly transforming due to technology adoption, resulting in an explosion of data. Extract, Transform, Load (ETL) processes are crucial for integrating and analyzing this data to support decision-making and enhance patient care. However, ETL processes face significant challenges, including data diversity, quality issues, security and compliance, and scalability. Opportunities exist to optimize ETL processes through advanced technologies like big data analytics, containerization, and parallel computing, improving data quality, and enhancing security. This literature review examines current ETL processes in healthcare, highlighting challenges and opportunities for future improvement, ultimately aiming to enhance healthcare outcomes and patient experiences. …

Multi-Scale Ai-Assisted Gene Expression Decoding, Fengyao Yan

Theses and Dissertations

Genes can be treated as a graph that can be mapped. Tremendous information is coded in genes to ensure a complex functioning organism. Decoding this information is critical to understanding our biology and developing treatments for various diseases including cancer. Deep learning, a new branch of computer science, has gained traction over the past decade. It offers more insight into the data that is processed by the deeplearning models. Our study has shown that deep-learning models can be an effective tool in decoding genetic data such as gene tissue-deconvolution, gene graph mapping and genomic imputation. In tasks such as tissue …