Data Science Commons^™

Federated Learning For Early Medical Diagnosis: Enhanced Diabetic Retinopathy Detection In Smart Healthcare, Mohammad Nasajpour Esfahani

Master's Theses

This thesis investigates the role of federated learning as a privacy-preserving solution for modern healthcare challenges. In traditional machine learning, sensitive medical data must be centralized for model training, raising concerns about privacy, security, and regulatory compliance. Federated learning offers an alternative by allowing hospitals, clinics, and personal health devices to collaboratively train shared models without exchanging raw patient data. The study first explores how federated learning is being used across various healthcare domains, including cancer detection, medical imaging, and disease prediction— highlighting its potential to support secure collaboration across institutions. It addresses key benefits such as data privacy, scalability, …

Predicting Student Belonging In Computing Education: A Multimodal Machine Learning Approach Using Eeg And Survey Data, Hannah Moshtaghi

Master's Theses

Measuring students’ sense of belonging, characterized by feelings of acceptance, inclusion, and encouragement from teachers, remains a significant challenge in computing education. Prior research has associated this multidimensional construct with positive academic outcomes and has identified instructors’ growth- and fixed-mindset messaging as a potential influence. However, belonging is a complex and deeply personal experience that is difficult to capture through direct observation alone. Current measurement methods rely on self-report surveys, which may not capture every aspect of an experience that can also involve emotional and cognitive responses.

This thesis investigates whether combining EEG data recorded during a belonging questionnaire with …

Developing A Humpback Whale Vocalization Detector Using Machine Learning Models, Lucas Kantorowski

Master's Theses

Humpback whale songs are notoriously complex. Identification of humpback whale song units requires bioacousticians to tediously listen, analyze, and annotate collected sound data. Even sparse data requires listening to the entirety of the collected acoustic data. In this study, three hours of audio containing over one-thousand humpback whale song units was collected in Monterey Bay, California.

Prior studies have seen success using convolutional neural networks by performing image classification on hundreds of hours worth of spectrograms. Our study uses traditional machine learning models, as they are less computationally demanding, and require less data.

We use time splitting and Mel-frequency cepstrum …

Determining K Clusters In K-Means Clustering With The Crab Algorithm, Jasmine Kristine S. Cabrera

Master's Theses

Unsupervised clustering often faces the challenge of determining the correct number of clusters in the absence of a true target variable. Traditional methods such as the Elbow Method and the Silhouette Score can produce ambiguous results and rely on assumptions about cluster shape or separation. To address this, we created the Clustering Rivals and Buddies (CRAB) algorithm which evaluates clusters based on stability across multiple subsamples. CRAB uses pairwise classifications to identify points that consistently group together called “Buddies” and points that remain separated called “Rivals.” Applied with K-means, CRAB accurately recovers underlying cluster structures in both spherical and non-spherical …

Crab: A Novel Clustering Score Using Clustering With Rivals And Buddies For Unsupervised Learning, Allen Choi

Master's Theses

Unsupervised clustering algorithms today are used across a wide variety of fields such as biology, engineering, and industry in order to classify observations into groups where labels are not provided. This can provide important latent information regarding the observations within groups, as well as insight regarding the groups themselves. In order to judge the optimal number of clusters for an unsupervised clustering algorithm, many methods exist such as the Elbow Method and Silhouette Score; however, these methods come with drawbacks and are not necessarily flexible across many unsupervised methods. We present a novel clustering score framework relying on a resampling-based …

Using Ensemble Disagreement To Stabilize Conformal Prediction Under Distribution Shift, Patrick D. Murphy

Master's Theses

Semantic segmentation of eelgrass from drone imagery is crucial for coastal habitat monitoring, restoration, and management, as these habitats continue to see rapid changes due to climate change and human influence. However, the reliability of generalizing a deployed classification model relies on both high-accuracy segmentation as well as robust uncertainty quantification that holds up when conditions change over years or locations. Conformal prediction (CP) is a method that converts a classifier's output into prediction sets with a guaranteed average coverage level for in-distribution data. However, the “vanilla” conformal score can often under-cover in hard or out-of-distribution (OOD) regions under drift. …

Generalized Detection Of Animal Behavior Using Accelerometers, Alexander J. Arrieta

Master's Theses

Animal mounted sensors are becoming increasingly used to passively monitor both domestic and wild animals. Advances in lightweight accelerometer and GPS technology have allowed many animals to be fitted with high accuracy sensors for extended periods of time. This leads to new opportunities to study animal behavior without direct observation. However, interpreting the raw data is difficult due to the high volume and missing context of the information. Machine learning techniques excel at extracting information from raw data streams and are excellent candidates for processing the sensor data. However, due to large variance in how different animals execute the same …

A Decision Support System For Conference Session Selection Using Natural Language Processing, Tillman E. Erb

Master's Theses

Conference attendees are faced with selecting from hundreds to thousands of presentations and sessions in pursuit of new findings and methods relevant to their area of interest, an overwhelming amount of information from which to clearly make a decision. To address this, we developed a decision support system leveraging natural language processing (NLP) techniques such as semantic matching. By creating and matching embeddings of conference presentation abstracts and titles, the application provides improved query matching compared to keyword searching. We introduce Session Scout, a novel conference decision support system built upon a semantic retrieval framework. Session Scout is designed to …

Density-Based And Model-Based Clustering With Tidyclust In R, Brendan S. Callender

Master's Theses

Clustering is a fundamental technique in unsupervised learning that can be used to find hidden patterns and structures within unlabeled data. The tidyclust package in R provides a unified interface for applying various clustering techniques to data. This paper outlines the addition of density-based clustering with DBSCAN, and model-based clustering using Gaussian mixture models (GMMs) to the tidyclust package. DBSCAN can be performed using the db_clust() function and makes use of the dbscan package implementation as its engine. GMMs can be fit using the gm_clust() function which makes use of the mclust package implementation. This paper highlights the changes made …

Opening The Black Box With Regal: A Novel Explainable Ai Approach To Uncover Key Predictors In Search And Rescue Success, Brandon Hyunjun Kim

Master's Theses

The outcome of a search and rescue (SAR) operation is influenced by a complex, non-linear interplay among numerous factors, including geographic context, subject-specific characteristics, and environmental conditions. The high dimensionality and intricate dependencies among these variables pose significant challenges to traditional exploratory modeling approaches, limiting their ability to uncover meaningful patterns and relationships associated with mission success. This study introduces Rules Based Explanations for Generated neighborhoods Around Localized cases (REGAL), a novel adaptation of the Local Interpretable Model-agnostic Explanations (LIME) framework to explain deep multimodal neural networks and what key features it assesses to determine search and rescue success. REGAL …

Topodino: Self-Supervised Topological Representation Learning For Neuronal Morphologies, Yasser Binbisher

Master's Theses

Neuronal cell types are categorized by transcriptomic identity, yet their morphological heterogeneity defies this classification. In response, researchers have adopted unsupervised graph representation learning as a tool to reveal morphological variation within single-class transcriptomic types. However, the complex geometry of neuronal morphology—especially long axons and dense dendrites—challenges graph neural networks, which struggle with message propagation across extended structures. To mitigate this, current approaches enforce sub-sampling on neuronal graphs and omit axons entirely, sacrificing critical biological features for computational efficiency. To overcome this trade-off, this thesis introduces TopoDINO, a self-supervised, topology-aware representation learning model designed to preserve the full hierarchical organization …

Frequent Itemset Mining With Tidyclust In R, Andrew D. Kerr

Master's Theses

Unsupervised learning is closely associated with clustering, however other methods fall under this umbrella such as data mining. In R, the tidyclust package provides a unified interface for clustering models, yet lacks support for data mining. This thesis addresses this gap by introducing the Apriori and ECLAT algorithms into tidyclust, with a focus on frequent itemset mining. Unlike traditional clustering models, frequent itemsets produce groupings of column variables, rather than cluster labels or partitions of observations. To address this, a novel clustering approach is proposed: items (columns) are grouped based on their ”dominant” frequent itemset. A key contribution is a …

Data Driven Analysis Of Samara Seed Kinematics And Dynamics, Shashwat Sparsh

Master's Theses

Samara Seeds are a class of fruit most famously belonging to the Acer species and are characterized by their single-bladed geometry and their auto-rotation response during descent. This steady-state auto-rotation response is the subject of aerodynamic analysis which aim to quantify the performance. The period prior to the beginning of steady-state auto-rotation is classified as the transition regime and has not been the subject of intense scrutiny.

This thesis employs a data-driven approach to analyzing the kinematic and dynamic response of these seeds during both the transition and auto-rotation stages of flight to quantify the performance with respect to the …

Dynamic Knowledge Elicitation: Leveraging Student Feedback For Improved Language Model Distillation, Reuven Muller

Master's Theses

Large Language Models (LLMs) have significantly advanced the field of natural language processing but remain resource-intensive and impractical for many organizations. Specialist models offer a viable alternative, often developed through Knowledge Distillation (KD) techniques. However, traditional KD methods rely on predefined static datasets to elicit knowledge from the teacher model, failing to dynamically address the weaknesses of the student model during training. This research introduces two novel methods for adaptive knowledge elicitation: Feedback-Driven Question Generation and Agent-Based Targeted Question Generation. These methods iteratively expand the training dataset based on the student model’s performance, leveraging a teacher model to generate targeted …

Semantic Structuring Of Digital Documents: Knowledge Graph Generation And Evaluation, Erik E. Luu

Master's Theses

In the era of total digitization of documents, navigating vast and heterogeneous data landscapes presents significant challenges for effective information retrieval, both for humans and digital agents. Traditional methods of knowledge organization often struggle to keep pace with evolving user demands, resulting in suboptimal outcomes such as information overload and disorganized data. This thesis presents a case study on a pipeline that leverages principles from cognitive science, graph theory, and semantic computing to generate semantically organized knowledge graphs. By evaluating a combination of different models, methodologies, and algorithms, the pipeline aims to enhance the organization and retrieval of digital documents. …

Design And Implementation Of A Vision-Based Deep-Learning Protocol For Kinematic Feature Extraction With Application To Stroke Rehabilitation, Juan Diego Luna Inga

Master's Theses

Stroke is a leading cause of long-term disability, affecting thousands of individuals annually and significantly impairing their mobility, independence, and quality of life. Traditional methods for assessing motor impairments are often costly and invasive, creating substantial barriers to effective rehabilitation. This thesis explores the use of DeepLabCut (DLC), a deep-learning-based pose estimation tool, to extract clinically meaningful kinematic features from video data of stroke survivors with upper-extremity (UE) impairments.

To conduct this investigation, a specialized protocol was developed to tailor DLC for analyzing movements characteristic of UE impairments in stroke survivors. This protocol was validated through comparative analysis using peak …

Accessible Real-Time Eye-Gaze Tracking For Neurocognitive Health Assessments, A Multimodal Web-Based Approach, Daniel C. Tisdale

Master's Theses

We introduce a novel integration of real-time, predictive eye-gaze tracking models into a multimodal dialogue system tailored for remote health assessments. This system is designed to be highly accessible requiring only a conventional webcam for video input along with minimal cursor interaction and utilizes engaging gaze-based tasks that can be performed directly in a web browser. We have crafted dynamic subsystems that capture high-quality data efficiently and maintain quality through instances of user attrition and incomplete calls. Additionally, these subsystems are designed with the foresight to allow for future re-analysis using improved predictive models, as well as enable the creation …

Contrastive Filtering And Dual-Objective Supervised Learning For Novel Class Discovery In Document-Level Relation Extraction, Nicholas Hansen

Master's Theses

Relation extraction (RE) is a task within natural language processing focused on the classification of relationships between entities in a given text. Primary applications of RE can be seen in various contexts such as knowledge graph construction and question answering systems. Traditional approaches to RE tend towards the prediction of relationships between exactly two entity mentions in small text snippets. However, with the introduction of datasets such as DocRED, research in this niche has progressed into examining RE at the document-level. Document-level relation extraction (DocRE) disrupts conventional approaches as it inherently introduces the possibility of multiple mentions of each unique …

Characterization And Estimation Of Musculoskeletal Pain Using Machine Learning, Boluwatife Faremi

Master's Theses

Traditional scales utilized for recording pain are known to be highly subjective and biased due to inaccuracies in recollecting actual pain intensities. As a result, machine learning (ML) models that are trained using these scores as ground truth are reported to have low performance for objective pain classification because of the huge disparity between what was felt in moments of pain and the scores recorded afterward.

In the present study, two devices were designed for gathering real-time, continuous in-session subjective pain scores and the recording of the autonomic nervous system (ANS) altered endodermal (EDA) activity. 24 participants were recruited to …

Analyzing Tortuosity In Patterns Formed By Colonies Of Embryonic Stem Cells Using Topological Data Analysis, Jackie Driscoll

Master's Theses

Pluripotent stem cells have been observed to segregate into Turing-like patterns during the early stages of Dox-inducible hiPSC differentiation. In this thesis, we de- velop a tool to quantify the tortuosity in the patterns formed by colonies of pluripo- tent stem cells using methods from topological data analysis. We use clustering techniques and the mapper algorithm to create simplicial complexes representing samples of cells and detail a method of evaluating the tortuosity of these complexes. We use the resulting persistence landscapes and their associated norms to evaluate experimental data and simulated data from an agent based model. This thesis finds …

Tempers Rising: The Effect Of Heat On Spite, Jake C. Cosgrove

Master's Theses

The relationship between heat and harmful outcomes is well documented, with research connecting various adverse economic outcomes to the climate. In the presence of increasing global warming and climate change, understanding why the climate leads to negative economic outcomes is essential for forming peaceful institutions of the future. We study how behavioral economic outcomes change in the presence of heat through a lab experiment involving 1,110 observations conducted in five different countries. This paper specifically focuses on the social preference outcome of spite. We find that increased time exposure to the treatment effect of heat is required to elicit an …

Deep Learning For Detecting Trees In The Urban Environment From Lidar, Julian R. Rice

Master's Theses

Cataloguing and classifying trees in the urban environment is a crucial step in urban and environmental planning. However, manual collection and maintenance of this data is expensive and time-consuming. Algorithmic approaches that rely on remote sensing data have been developed for tree detection in forests, though they generally struggle in the more varied urban environment. This work proposes a novel method for the detection of trees in the urban environment that applies deep learning to remote sensing data. Specifically, we train a PointNet-based neural network to predict tree locations directly from LIDAR data augmented with multi-spectral imaging. We compare this …

Legislative Language For Success, Sanjana Gundala

Master's Theses

Legislative committee meetings are an integral part of the lawmaking process for local and state bills. The testimony presented during these meetings is a large factor in the outcome of the proposed bill. This research uses Natural Language Processing and Machine Learning techniques to analyze testimonies from California Legislative committee meetings from 2015-2016 in order to identify what aspects of a testimony makes it successful. A testimony is considered successful if the alignment of the testimony matches the bill outcome (alignment is "For" and the bill passes or alignment is "Against" and the bill fails). The process of finding what …

Take The Lead: Toward A Virtual Video Dance Partner, Ty Farris

Master's Theses

My work focuses on taking a single person as input and predicting the intentional movement of one dance partner based on the other dance partner's movement. Human pose estimation has been applied to dance and computer vision, but many existing applications focus on a single individual or multiple individuals performing. Currently there are very few works that focus specifically on dance couples combined with pose prediction. This thesis is applicable to the entertainment and gaming industry by training people to dance with a virtual dance partner.

Many existing interactive or virtual dance partners require a motion capture system, multiple cameras …

Investigating Daily Fantasy Baseball: An Approach To Automated Lineup Generation, Ryan Smith

Master's Theses

A recent trend among sports fans along both sides of the letterman jacket is that of Daily Fantasy Sports (DFS). The DFS industry has been under legal scrutiny recently, due to the view that daily sports data is too random to make its prediction skillful. Therefore, a common view is that it constitutes online gambling. This thesis proves that DFS, as it pertains to Baseball, is significantly more predictable than random chance, and thus does not constitute gambling.

We propose a system which generates daily lists of lineups for Fanduel Daily Fantasy Baseball contests. The system consists of two components: …

Soarnet, Deep Learning Thermal Detection For Free Flight, Jake T. Tallman

Master's Theses

Thermals are regions of rising hot air formed on the ground through the warming of the surface by the sun. Thermals are commonly used by birds and glider pilots to extend flight duration, increase cross-country distance, and conserve energy. This kind of powerless flight using natural sources of lift is called soaring. Once a thermal is encountered, the pilot flies in circles to keep within the thermal, so gaining altitude before flying off to the next thermal and towards the destination. A single thermal can net a pilot thousands of feet of elevation gain, however estimating thermal locations is not …

Modeling The Spread Of Covid-19 Over Varied Contact Networks, Ryan L. Solorzano

Master's Theses

When attempting to mitigate the spread of an epidemic without the use of a vaccine, many measures may be made to dampen the spread of the disease such as physically distancing and wearing masks. The implementation of an effective test and quarantine strategy on a population has the potential to make a large impact on the spread of the disease as well. Testing and quarantining strategies become difficult when a portion of the population are asymptomatic spreaders of the disease. Additionally, a study has shown that randomly testing a portion of a population for asymptomatic individuals makes a small impact …

A Performance Survey Of Text-Based Sentiment Analysis Methods For Automating Usability Evaluations, Kelsi Van Damme

Master's Theses

Usability testing, or user experience (UX) testing, is increasingly recognized as an important part of the user interface design process. However, evaluating usability tests can be expensive in terms of time and resources and can lack consistency between human evaluators. This makes automation an appealing expansion or alternative to conventional usability techniques.

Early usability automation focused on evaluating human behavior through quantitative metrics but the explosion of opinion mining and sentiment analysis applications in recent decades has led to exciting new possibilities for usability evaluation methods.

This paper presents a survey of modern, open-source sentiment analyzers’ usefulness in extracting and …

Application Of Machine Learning Techniques To Forecast Harmful Algal Blooms In Gulf Of Mexico, Bala Tripura Sundari Yerrapothu

Master's Theses

The Harmful Algal Blooms (HABs) forecast is crucial for the mitigation of health hazards and to inform actions for the protection of ecosystems and fisheries in the Gulf of Mexico (GoM). For the sake of simplicity of our application we assume ocean color satellite imagery from the National Oceanic and Atmospheric Administration as a proxy for HABs.

In this study we use a deep neural network trained on the 2-Dimensional time series proxy data to provide a forecast of the HABs’ manifestations in the GoM.Our approach analyzes between both spatial and temporal features simultaneously. In addition, the network also helps …

Node Classification On Relational Graphs Using Deep-Rgcns, Nagasai Chandra

Master's Theses

Knowledge Graphs are fascinating concepts in machine learning as they can hold usefully structured information in the form of entities and their relations. Despite the valuable applications of such graphs, most knowledge bases remain incomplete. This missing information harms downstream applications such as information retrieval and opens a window for research in statistical relational learning tasks such as node classification and link prediction. This work proposes a deep learning framework based on existing relational convolutional (R-GCN) layers to learn on highly multi-relational data characteristic of realistic knowledge graphs for node property classification tasks. We propose a deep and improved variant, …