Data Science Commons^™

Edge Co-Occurrence Regularization For Node Classification, Kadir Altunel

Theses

We propose a simple yet effective regularization technique for node classification on graphs that leverages edge-based label co-occurrence patterns. We first train an MLP on node features to produce class probability distributions, then compute a fixed penalty matrix from edge-based co-occurrence statistics of these predictions. This penalty matrix, which captures unlikely class combinations on connected nodes, is then used to regularize GNN training without further updates. We evaluate this approach across multiple homophilic datasets (Cora, CiteSeer, PubMed, ogbn-arxiv) and heterophilic benchmarks (Chameleon, Squirrel, Actor, Roman-Empire) using three GNN architectures: GCN, GraphSAGE, and H2GCN. Results show consistent improvements on homophilic graphs, …

A Generative Ai-Driven Computational Framework For Industry-Scale Discovery Of Novel Battery Materials, Joy Datta

Dissertations

The growing demand for sustainable, high-energy-density electrochemical storage has motivated the exploration of multivalent-ion batteries based on earth-abundant elements such as aluminum, calcium, magnesium, and zinc. While multivalent charge carriers offer higher theoretical energy density than lithium, their practical deployment is hindered by sluggish ion transport, strong ion-host interactions, and structural degradation of electrode materials. Identifying host materials that can reversibly accommodate multivalent ions while maintaining structural integrity remains a fundamental challenge. The dissertation develops a scalable, end-to-end computational framework that integrates density functional theory (DFT), machine learning (ML), and generative artificial intelligence (GenAI) to accelerate the discovery of next-generation …

Toward Learning-Based Reconstruction And Part Decomposition Of Man-Made 3d Geometry: Neural Implicit Representations And Scalable Supervision, Shen Fan

Dissertations

Digital three-dimensional (3D) models are central to engineering design, analysis, and manufacturing, but learning pipelines for man-made geometry often operate on sampled carriers that do not preserve all of the structure present in exact CAD representations. This dissertation studies learning-based reconstruction and part decomposition for structured man-made 3D geometry, from general object benchmarks to CAD-derived datasets, with a focus on neural implicit representations trained from signed-distance samples, point clouds, and tessellated meshes. The goal is to make these models more accurate, more part-aware, and more consistently supervised.

First, signed distance function (SDF) reconstruction with implicit neural representations is improved through …

Semi-Supervised Learning For Annotation And Representation Of Single-Cell Rna Sequencing And Spatial Transcriptomics Data, Haoran Liu

Dissertations

Semi-supervised learning has emerged as a powerful paradigm for analyzing single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics (ST) data, where full annotation is often costly or impractical. scRNA-seq technologies measure the expression of thousands of genes across tens of thousands of cells, whereas ST additionally captures the spatial coordinates of gene expression within intact tissue sections. Annotation is a key step in both scRNA-seq and ST analysis pipelines, aiming to identify cell types, spatial domains, and latent biological structures. However, most existing annotation approaches rely on separate clustering methods that are typically fully unsupervised and fail to leverage side information …

Beyond Words: A Systematic Multimodal Framework For Text, Images, And Extreme Helpfulness In Online Reviews, Alvaro J. Aguado Marin

Dissertations

Online product reviews have become increasingly multimodal, combining text with media-rich elements such as images. However, academic research has largely examined textual features in isolation, overlooking how visual content and its interaction with text shape perceived helpfulness. This dissertation addresses that gap by developing and empirically validating a comprehensive framework capturing how textual, visual, and contextual features collectively influence review evaluation. Grounded in the Elaboration Likelihood Model (ELM) and extended through the Text-Image Elaboration Likelihood Model (TI-ELM), the framework advances understanding of how consumers process content from both user- and business-generated sources. It also lays the foundation for examining emerging …

Harnessing Graphs For Knowledge Representation In Natural Language Processing, Uras Varolgunes

Dissertations

This work proposes innovative methods for integrating domain-specific knowledge into natural language processing tasks through the use of graphs, aiming to enhance the performance of models across various domains, including finance and healthcare. Several novel approaches are proposed that fuse graph structures with modern deep learning techniques, addressing the challenges of missing word embeddings, label prediction, and graph representation learning for large language models.

First, a powerful embedding method built on top of the recent advances in latent graph learning is introduced to address the critical problem of word embedding imputation. Second, a graph-enhanced label attention model designed for medical …

Machine Learning And Optimization For Intelligent Decision-Making, Elson Cibaku

Dissertations

This dissertation presents a series of innovative machine learning and optimization model designs that address complex operational challenges across logistics and power systems. By integrating advanced neural architectures with robust optimization techniques, the work delivers scalable solutions designed to improve efficiency, reliability, and decision-making in dynamic and real-world environments. The first study introduces a two-stage approach to effective vaccine distribution. This framework tackles the capacitated vehicle routing problem by combining adaptive clustering techniques with reinforcement learning and a simulated annealing pickup policy. Through extensive computational experiments, the approach demonstrates substantial improvements in routing efficiency, reducing both computational time and logistical …

Towards Explainable Ai On Graph Neural Networks: Xaig, Jiaxing Zhang

Dissertations

In the evolving landscape of artificial intelligence (AI), Graph Neural Networks (GNNs) have garnered growing prominence for their adeptness in processing graph-structured data. Despite this, the interpretability of their predictions often remains elusive. The demand for transparency and explainability in complex prediction models has reached unprecedented levels. To address this, post-hoc instance-level explanation techniques have emerged, aiming to unveil the rationale behind GNN predictions. These techniques endeavor to unearth substructures that elucidate the predictive behavior of trained GNNs.

This dissertation embarks on an exploration of Explainable AI (XAI) technologies within the realm of GNNs. Amid the challenges posed by the …

A Novel Framework For Dynamic Graph Representation Learning With Mamba, Ashish Pandey

Theses

Dynamic graph embedding is a key technique for modeling temporal dependencies in evolving networks. While transformer-based models perform well, their quadratic complexity limits scalability on long graph sequences. This thesis compares transformer approaches with the Mamba architecture-a linear-complexity state-space model—for temporal graph embedding.

Two frameworks are proposed: DG-Mamba and GDG-Mamba. DG-Mamba uses standard GCN-based spatial encoding, while GDG-Mamba incorporates domain-aware edge features using Graph Isomorphism Network with Edge Convolution (GraphGINE). Experiments on UCI, Reality Mining, Slashdot, Bitcoin-OTC, and SBM datasets show that Mamba-based models match or exceed transformer performance, especially on graphs with high temporal variability.

The thesis also applies …

Interactive Visualization Workflows For Mitigating Analytical Uncertainty, Kaustav Bhattacharjee

Dissertations

This dissertation takes a process-centric and stakeholder-first perspective for handling analytical uncertainty: the form of uncertainty that confronts data analysts' insight-generation processes in high-consequence decision-making scenarios. The cost of an incorrect decision when data is used for movie recommendations as opposed to when personal data is used to drive insights or when data-driven modeling is used to drive real-time decisions for maintaining the health of a grid are vastly different in terms of consequences. This dissertation looks at analytical uncertainty in two real-world scenarios: i) how sensitive information leakage can be prevented during the open data release process with data …

Visual Analytic Techniques For Interpretable Algorithmic Ranking Systems, Jun Yuan

Dissertations

Rankings have a profound impact on the increasingly data-driven society. From leisurely activities like the movies to watch, the restaurants to patronize; to highly consequential decisions, like making educational and occupational choices or getting hired by companies— these are all driven by sophisticated yet mostly opaque algorithmic rankers. A small change in how these rankers order the data items can have profound consequences, like deterioration of the prestige of a university or a job applicant missing out on being on the list of the top candidates for an organization. These scenarios necessitate data-driven and human-centered innovation to make rankers accessible, …

Advancing Prediction Of Stimulant Medication Misuse Through Graph Representation Learning, Hamid Razavi

Theses

The misuse of stimulant prescription medications poses a significant and escalating public health concern in the United States, particularly among young adults. Addressing this issue requires sophisticated methodologies capable of uncovering complex patterns and relationships in data. Geometric Deep Learning, a paradigm designed to analyze data with non-Euclidean structures, has achieved remarkable success across various domains, offering a powerful framework for tackling complex graph structure data challenges.

This study leverages Graph Convolutional Networks (GCNs) to predict the likelihood of stimulant medication misuse using data from the National Survey on Drug Use and Health (NSDUH). Individuals are represented as nodes in …

Data Driven Decision Making For Sustainable Planning And Operations Of Large Scale Networks, Bahareh Kargar

Dissertations

This dissertation explores data-driven decision-making networks, focusing on sustainable planning and operations for large-scale systems such as healthcare supply chains and power systems. One significant application in healthcare is the optimization of vaccine supply chains. An agent-based simulation-optimization modeling framework is developed to enhance the efficiency and sustainability of vaccine distribution. First, an agent-based epidemiological model of COVID-19 is extended to capture disease transmission dynamics and forecast the number of susceptible individuals and infections. Then, a sustainable vaccine supply chain considering the impacts of greenhouse gases is developed and integrated with the simulation model to minimize total costs and environmental …

A Methodological Framework For Ontology Development, Enrichment, And Application In Natural Language Processing Tasks, Navya Martin Kollapally

Dissertations

Electronic Health Records (EHRs) have been widely used in healthcare to record demographics, vital signs, test results, immunizations, medical imaging reports, differential diagnoses, etc. It is now accepted that non-clinical (e.g., social) factors have a substantial influence on health outcomes. Hence, it is desirable to record these Social and Commercial Determinants of Health (SDoH & CDoH) in an EHR. The "non-text parts" of EHR notes (e.g., data tables) rely on coded terms from underlying ontologies or terminologies to facilitate semantic interoperability. Ontologies help define concepts, the relationships between them, and instances that can be utilized in research.

The first accomplishment …

Machine Learning-Based Design Of Doppler Tolerant Radar, Kyle Peter Wensell

Dissertations

In this work, machine learning theory is applied to the design of a radar detector in order to train a machine learning-based detector that is robust against Doppler shifts. The radar system is designed to work with data that would be otherwise intractable to conventional optimal detector design, such as transmitted noise waveforms and the effects of one-bit quantization at the receiver. The detection performance of the one-bit receiver is shown to match the performance of the derived square-law sign correlator detector. The resulting learning-based detector also introduces Doppler tolerance to the system, which allows for the successful detection of …

Financial Time Series Fusion, Completion, And Prediction With Deep Neural Networks, Dan Zhou

Dissertations

Time-series analysis is essential for a wide range of financial applications, including but not limited to bond valuation, firm earnings forecasts, firm fundamentals predictions, and firm characteristics imputations. Given its considerable value, the financial community has shown a strong interest in refining and advancing time-series analysis techniques. The study in this dissertation contributes to this field by employing advanced machine learning approaches, specifically graph neural networks, deep neural networks, and matrix/tensor methods. The primary objectives are twofold: first, to reveal complex correlations within financial time series to improve prediction accuracy, and second, to enhance the process of integrating and imputing …

Human-Ai Complex Task Planning, Sepideh Nikookar

Dissertations

The process of complex task planning is ubiquitous and arises in a variety of compelling applications. A few leading examples include designing a personalized course plan or trip plan, designing music playlists/work sessions in web applications, or even planning routes of naval assets to collaboratively discover an unknown destination. For all of these aforementioned applications, creating a plan requires satisfying a basic construct, i.e., composing a sequence of sub-tasks (or items) that optimizes several criteria and satisfies constraints. For instance, in course planning, sub-tasks or items are core and elective courses, and degree requirements capture their complex dependencies as constraints. …

Data-Driven 2d Materials Discovery For Next-Generation Electronics, Zeyu Zhang

Dissertations

The development of material discovery and design has lasted centuries in human history. After the concept of modern chemistry and material science was established, the strategy of material discovery relies on the experiments. Such a strategy becomes expensive and time-consuming with the increasing number of materials nowadays. Therefore, a novel strategy that is faster and more comprehensive is urgently needed. In this dissertation, an experiment-guided material discovery strategy is developed and explained using metal-organic frameworks (MOFs) as instances. The advent of 7r-stacked layered MOFs, which offer electrical conductivity on top of permanent porosity and high surface area, opened up new …

On Explainability Of Neural Networks, Cem Benar

Dissertations

It is widely reported that deep neural networks outperform most competitors for a range of applications. The state-of-the-art neural networks have built-in inductive bias of architectural choices, regularizations, optimizer types, and initialization methods. Using inductive bias is intuitive to enhance the model approximation. Deep neural networks are mostly dense and heavily overparameterized. They tend to be biased towards low-rank solutions to reduce complexity and improve generalization performance, known as implicit regularization. The implicit regularization as observed in specific architectures and various real-world data sets suggests to overparameterize neural networks judiciously and learn compressed representations (lower rank approximation) with improved performance. …

Trustworthy Machine Learning Through The Lens Of Privacy And Security, Thi Kim Phung Lai

Dissertations

Nowadays, machine learning (ML) becomes ubiquitous and it is transforming society. However, there are still many incidents caused by ML-based systems when ML is deployed in real-world scenarios. Therefore, to allow wide adoption of ML in the real world, especially in critical applications such as healthcare, finance, etc., it is crucial to develop ML models that are not only accurate but also trustworthy (e.g., explainable, privacy-preserving, secure, and robust). Achieving trustworthy ML with different machine learning paradigms (e.g., deep learning, centralized learning, federated learning, etc.), and application domains (e.g., computer vision, natural language, human study, malware systems, etc.) is challenging, …

Ai Approaches To Understand Human Deceptions, Perceptions, And Perspectives In Social Media, Chih-Yuan Li

Dissertations

Social media platforms have created virtual space for sharing user generated information, connecting, and interacting among users. However, there are research and societal challenges: 1) The users are generating and sharing the disinformation 2) It is difficult to understand citizens' perceptions or opinions expressed on wide variety of topics; and 3) There are overloaded information and echo chamber problems without overall understanding of the different perspectives taken by different people or groups.

This dissertation addresses these three research challenges with advanced AI and Machine Learning approaches. To address the fake news, as deceptions on the facts, this dissertation presents Machine …

Deep Hybrid Modeling Of Neuronal Dynamics Using Generative Adversarial Networks, Soheil Saghafi

Dissertations

Mechanistic modeling and machine learning methods are powerful techniques for approximating biological systems and making accurate predictions from data. However, when used in isolation these approaches suffer from distinct shortcomings: model and parameter uncertainty limit mechanistic modeling, whereas machine learning methods disregard the underlying biophysical mechanisms. This dissertation constructs Deep Hybrid Models that address these shortcomings by combining deep learning with mechanistic modeling. In particular, this dissertation uses Generative Adversarial Networks (GANs) to provide an inverse mapping of data to mechanistic models and identifies the distributions of mechanistic model parameters coherent to the data.

Chapter 1 provides background information on …

A Survey On Online Matching And Ad Allocation, Ryan Lee

Theses

One of the classical problems in graph theory is matching. Given an undirected graph, find a matching which is a set of edges without common vertices. In 1990s, Richard Karp, Umesh Vazirani, and Vijay Vazirani would be the first computer scientists to use matchings for online algorithms [8]. In our domain, an online algorithm operates in the online setting where a bipartite graph is given. On one side of the graph there is a set of advertisers and on the other side we have a set of impressions. During the online phase, multiple impressions will arrive and the objective of …

Integrated Machine Learning And Optimization Approaches, Dogacan Yilmaz

Dissertations

This dissertation focuses on the integration of machine learning and optimization. Specifically, novel machine learning-based frameworks are proposed to help solve a broad range of well-known operations research problems to reduce the solution times. The first study presents a bidirectional Long Short-Term Memory framework to learn optimal solutions to sequential decision-making problems. Computational results show that the framework significantly reduces the solution time of benchmark capacitated lot-sizing problems without much loss in feasibility and optimality. Also, models trained using shorter planning horizons can successfully predict the optimal solution of the instances with longer planning horizons. For the hardest data set, …

Analyzing Fluctuation Of Topics And Public Sentiment Through Social Media Data, Haoyue Liu

Dissertations

Over the past decade years, Internet users were expending rapidly in the world. They form various online social networks through such Internet platforms as Twitter, Facebook and Instagram. These platforms provide a fast way that helps their users receive and disseminate information and express personal opinions in virtual space. When dealing with massive and chaotic social media data, how to accurately determine what events or concepts users are discussing is an interesting and important problem.

This dissertation work mainly consists of two parts. First, this research pays attention to mining the hidden topics and user interest trend by analyzing real-world …

Efficient And Scalable Triangle Centrality Algorithms In The Arkouda Framework, Joseph Thomas Patchett

Theses

Graph data structures provide a unique challenge for both analysis and algorithm development. These data structures are irregular in that memory accesses are not known a priori and accesses to these structures tend to lack locality.

Despite these challenges, graph data structures are a natural way to represent relationships between entities and to exhibit unique features about these relationships. The network created from these relationships can create unique local structures that can describe the behavior between members of these structures. Graphs can be analyzed in a number of different ways including at a high level in community detection and at …

Representation Learning In Finance, Ajim Uddin

Dissertations

Finance studies often employ heterogeneous datasets from different sources with different structures and frequencies. Some data are noisy, sparse, and unbalanced with missing values; some are unstructured, containing text or networks. Traditional techniques often struggle to combine and effectively extract information from these datasets. This work explores representation learning as a proven machine learning technique in learning informative embedding from complex, noisy, and dynamic financial data. This dissertation proposes novel factorization algorithms and network modeling techniques to learn the local and global representation of data in two specific financial applications: analysts’ earnings forecasts and asset pricing.

Financial analysts’ earnings forecast …

Un-Fair Trojan: Targeted Backdoor Attacks Against Model Fairness, Nicholas Furth

Theses

Machine learning models have been shown to be vulnerable against various backdoor and data poisoning attacks that adversely affect model behavior. Additionally, these attacks have been shown to make unfair predictions with respect to certain protected features. In federated learning, multiple local models contribute to a single global model communicating only using local gradients, the issue of attacks become more prevalent and complex. Previously published works revolve around solving these issues both individually and jointly. However, there has been little study on the effects of attacks against model fairness. Demonstrated in this work, a flexible attack, which we call Un-Fair …

On Performance Optimization And Prediction Of Parallel Computing Frameworks In Big Data Systems, Haifa Alquwaiee

Dissertations

A wide spectrum of big data applications in science, engineering, and industry generate large datasets, which must be managed and processed in a timely and reliable manner for knowledge discovery. These tasks are now commonly executed in big data computing systems exemplified by Hadoop based on parallel processing and distributed storage and management. For example, many companies and research institutions have developed and deployed big data systems on top of NoSQL databases such as HBase and MongoDB, and parallel computing frameworks such as MapReduce and Spark, to ensure timely data analyses and efficient result delivery for decision making and business …

Private And Federated Deep Learning: System, Theory, And Applications For Social Good, Han Hu

Dissertations

During the past decade, drug abuse continues to accelerate towards becoming the most severe public health problem in the United States. The ability to detect drug­abuse risk behavior at a population scale, such as among the population of Twitter users, can help to monitor the trend of drug­abuse incidents. However, traditional methods do not effectively detect drug­abuse risk behavior in tweets, mainly due to the sparsity of such tweets and the noisy nature of tweets. In the first part of this dissertation work, the task of classifying tweets as containing drug­abuse risk behavior or not, is studied. Millions of public …