Digital Commons Network^™

Missing Wedge Completion Via Unsupervised Learning With Coordinate Networks, Dave Van Veen, Jesús G Galaz-Montoya, Liyue Shen, Philip Baldwin, Akshay S Chaudhari, Dmitry Lyumkis, Michael F Schmid, Wah Chiu, John Pauly

Faculty, Staff and Students Publications

Cryogenic electron tomography (cryoET) is a powerful tool in structural biology, enabling detailed 3D imaging of biological specimens at a resolution of nanometers. Despite its potential, cryoET faces challenges such as the missing wedge problem, which limits reconstruction quality due to incomplete data collection angles. Recently, supervised deep learning methods leveraging convolutional neural networks (CNNs) have considerably addressed this issue; however, their pretraining requirements render them susceptible to inaccuracies and artifacts, particularly when representative training data is scarce. To overcome these limitations, we introduce a proof-of-concept unsupervised learning approach using coordinate networks (CNs) that optimizes network weights directly against input …

High-Dimensional Data Analysis Using Parameter Free Algorithm Data Point Positioning Analysis, S. M. F. D. Syed Mustapha

All Works

Clustering is an effective statistical data analysis technique; it has several applications, including data mining, pattern recognition, image analysis, bioinformatics, and machine learning. Clustering helps to partition data into groups of objects with distinct characteristics. Most of the methods for clustering use manually selected parameters to find the clusters from the dataset. Consequently, it can be very challenging and time-consuming to extract the optimal parameters for clustering a dataset. Moreover, some clustering methods are inadequate for locating clusters in high-dimensional data. To address these concerns systematically, this paper introduces a novel selection-free clustering technique named data point positioning analysis (DPPA). …

Enhancing Electrical Network Vulnerability Assessment With Machine Learning And Deep Learning Techniques, M Mishkatur Rahman, Ayman Sajjad Akash, Harun Pirim, Chau Le, Trung Le, Om Prakash Yadav

Northeast Journal of Complex Systems (NEJCS)

This research utilizes advanced machine learning techniques to evaluate node vul-
nerability in power grid networks. Utilizing the SciGRID and GridKit datasets, con-
sisting of 479, 16,167 nodes and 765, 20,539 edges respectively, the study employs
K-nearest neighbor and median imputation methods to address missing data. Cen-
trality metrics are integrated into a single comprehensive score for assessing node
criticality, categorizing nodes into four centrality levels informative of vulnerability.
This categorization informs the use of traditional machine learning (including XG-
Boost, SVM, Multilayer Perceptron) and Graph Neural Networks in the analysis.
The study not only benchmarks the capabilities of these …

Computationally Assessing Suspicion, Wesley M. Oliver, Morgan A. Gray, Jaromir Savelka, Kevin D. Ashley

University of Cincinnati Law Review

Law enforcement officers performing drug interdiction on interstate highways have to decide nearly every day whether there is reasonable suspicion to detain motorists until a trained dog can sniff for the presence of drugs. The officers’ assessments are often wrong, however, and lead to unnecessary detentions of innocent persons and the suppression of drugs found on guilty ones. We propose a computational method of evaluating suspicion in these encounters and offer experimental results from early efforts demonstrating its feasibility. With the assistance of large language and predictive machine learning models, it appears that judges, advocates, and even police officers could …

A Framework For Filtering Irrelevant Notams, Roudha Abdulrahman

Theses

The Notice to Airmen (NOTAM) system is essential for aviation safety, giving critical information on risks and operating limitations. However, the volume and complexity of NOTAM data complicate interpretation, which could compromise safety and efficiency. This study tries to address these issues by creating a predictive algorithm for analyzing NOTAM data and predicting whether to keep or remove them. The methodology combines effective machine learning algorithms to reveal insights that improve safety and operational efficiency. Using a dataset of NOTAM entries, multiple prediction algorithms are tested to forecast possible problems and enable proactive risk management. The study's findings help to …

Robust Prediction Of Charpy Toughness Of Additively Manufactured Kovar Using Deep Convolutional Neural Networks, Nathan R. Bianco

Mathematics & Statistics ETDs

Understanding the reason for mechanical failures of manufactured parts in their operating environments is critical to prevention of future failures. However, in-situ post-mortem evaluation of physical properties, such as fracture toughness, is time consuming and alters the condition of the material, leading to potentially misleading findings. In this study, additively manufactured test coupons were produced over a wide range of process conditions to test the impact toughness of a material. The Charpy V-Notch toughness was measured on over 200 samples alongside corresponding optical images of both sides of the fracture surface. Convolutional neural network models were trained to correlate fracture …

Improving Automatic Refactoring Candidate Identification, Ryan Devoe

Theses

Extract method refactoring is pivotal for enhancing code readability, maintainability, and modularity by segmenting complex code into clearer, isolated methods. Identifying opportunities for such refactorings necessitates a deep understanding of the codebase’s evolution and its intricate relationships. Current methodologies utilize developer commit messages, advanced graph analysis, and diverse machine learning approaches to automate this identification process. This research delves into the application of deep learning-based Large Language Models (LLMs) to tackle the complexities inherent in extract method refactoring. We introduce innovative approaches, including the use of LLMs to cluster code blocks based on complex patterns and dependencies, and the analysis …

Predicting The Pathway Involvement Of Metabolites Based On Combined Metabolite And Pathway Features, Erik D. Huckvale, Hunter N. B. Moseley

Markey Cancer Center Faculty Publications

A major limitation of most metabolomics datasets is the sparsity of pathway annotations for detected metabolites. It is common for less than half of the identified metabolites in these datasets to have a known metabolic pathway involvement. Trying to address this limitation, machine learning models have been developed to predict the association of a metabolite with a “pathway category”, as defined by a metabolic knowledge base like KEGG. Past models were implemented as a single binary classifier specific to a single pathway category, requiring a set of binary classifiers for generating the predictions for multiple pathway categories. This past approach …

A Hierarchical Decision Model For Evaluating The Strategy Readiness Of Quantitative Machine Learning/Data Science-Driven Investment Strategies, Mohammadsaleh Saadatmand

Dissertations and Theses

Big data and computational technologies are increasingly important worldwide in asset and investment management. Many investment management firms are adopting these data science methods and technologies to improve performance across all investment processes. Researchers actively use these methods to develop more effective systematic investment strategies and produce more reliable outcomes less vulnerable to human decision-making biases. However, the success of such a strategy depends heavily on the scientific rigor applied throughout the process. "Best practices involve understanding how to make better decisions in the research design process. A good question is whether we can make better decisions in developing quantitative …

Improving Iol Calculators’ Total Corneal Astigmatism Predictions With Deep Learning, Jason Fink, Mathieu Gauvin, Chirag Shah, Mark Cohen, Avi Wallerstein

Research Day

Precise intraocular lens (IOL) selection plays a critical role in optimizing surgical outcomes for patients undergoing IOL surgery. Traditionally, only the anterior corneal astigmatism (ACA) was considered during IOL selection. However, the posterior corneal astigmatism (PCA), also significantly impacts vision, though it is more challenging to measure directly. Advancements in deep learning offer a promising solution. By analyzing existing ACA measurements, deep learning algorithms can predict total corneal astigmatism (TCA) with high accuracy, effectively inferring PCA and enabling more precise IOL selection.

We employed a multivariate linear regression neural network (NN) and evaluated its performance through R-squared and root mean …

Data-Driven And Cell-Specific Determination Of Nuclei-Associated Actin Structure, Nina Nikitina, Nurbanu Bursa, Matthew Goelzer, Madison Goldfeldt, Chase Crandall, Sean Howard, Janet Rubin, Anamaria Zavala, Aykut Satici, Gunes Uzer

Mechanical and Biomedical Engineering Faculty Publications and Presentations

Quantitative volumetric assessment of filamentous actin (F-actin) fibers remains challenging due to their interconnected nature, leading researchers to utilize threshold-based or qualitative measurement methods with poor reproducibility. Herein, a novel machine learning-based methodology is introduced for accurate quantification and reconstruction of nuclei-associated F-actin. Utilizing a convolutional neural network (CNN), actin filaments and nuclei from 3D confocal microscopy images are segmented and then each fiber is reconstructed by connecting intersecting contours on cross-sectional slices. This allows measurement of the total number of actin filaments and individual actin filament length and volume in a reproducible fashion. Focusing on the role of F-actin …

A Statistical Look Into How Common Soccer Metrics Influence Expected Goal Measures In The Professional Game, Tristan George Rumsey

Undergraduate Honors Thesis Collection

The advent of sports analytics has ignited a fervor across all sporting disciplines, particularly soccer, where clubs are sprinting to harness vast data reserves to elevate team performance, spearhead effective marketing endeavors, and bolster financial gains crucial for club expansion. Much like Billy Beane's transformative "Moneyball" approach, soccer clubs are in pursuit of innovative strategies to transcend financial limitations and achieve triumph. In soccer, where goals are scarce commodities, heightened offensive efficacy becomes imperative. Presently, one metric stands out as pivotal in gauging a team's goal-scoring success: expected goals (xG). This metric quantifies the likelihood of a given shot or …

Advancing Learning Models For High-Dimensional Data: From Molecular Modeling To Motion Planning, Tuan Nguyen Anh Tran

Legacy Theses & Dissertations (2009 - 2024)

Processing and analyzing high-dimensional data, particularly in domains like protein research and robotics, introduces unique challenges. In protein analysis, high-dimensional data often arises from complex molecular structures and their interactions. Analyzing protein data involves intricate computational models and algorithms that must deal with the large size of molecular datasets and the need to extract biologically relevant information. This makes it challenging to discern significant protein structure-activity relationships, understand complex biological interactions, or predict protein behavior accurately. Similarly, in robotics, high-dimensional data is prevalent when dealing with sensory inputs, the dimensionality of the robot, or the complexity of the operational environments. …

In-Training Explainability Frameworks To Make Black-Box Machine Learning Models More Explainable., Asuman Cagla Acun

Electronic Theses and Dissertations

Despite ongoing efforts to make black-box machine learning models more explainable, transparent, and trustworthy, there is growing advocacy for using only inherently interpretable models for high-stake decision-making. Post-hoc explanations have recently been criticized for learning surrogate models that may not accurately reflect the actual mechanisms of the original model and for adding computational burden at prediction time. We propose two novel explainability approaches to address these limitations: pre-hoc explainability and co-hoc explainability. These approaches integrate explanations derived from an inherently interpretable white-box model into the learning stage of the black-box model without compromising accuracy. Unlike post-hoc methods, our approach does …

Traffic Prediction In 5g Networks Using Machine Learning, Hossein Mehri

Boise State University Theses and Dissertations

The advent of 5G technology promises a paradigm shift in the realm of telecommunications, offering unprecedented speeds and connectivity. However, the efficient management of traffic in 5G networks remains a critical challenge. It is due to the dynamic and heterogeneous nature of network traffic, including bursty patterns, varying user behaviors, and diverse applications, all of which demand highly accurate and adaptable prediction models to optimize network resource allocation and management. This dissertation investigates the intricate domain of traffic prediction within 5G networks, addressing the specific challenges posed by both massive machine type communication (mMTC) networks and 5G cellular networks.

The …

Neural Network Characterization Methods For Noisy Quantum Hardware Results, Daniel Richard Breslawski

Theses

As the nascent quantum computing paradigm matures and quantum devices become widely available, discrete "eras" are emerging which characterize evolving quantum computing technologies, much like semiconductor technology node classifications. The current period has been coined the Noisy Intermediate-Scale Quantum (NISQ) era, as devices above 50 qubits in size exist but computations using these devices accumulate error quickly from classical interference sources and state decoherence. NISQ machines may surpass the capabilities of modern classical computers in ideal circumstances, but the accumulation of error from physical noise limits the size and implementability of reliable quantum algorithms. Because of these limitations, strategies are …

Predicting Student Attrition In Higher Education Institutions In The Uae Using Machine Learning, Dezzil M. Castelino

Theses

UAE has made significant progress in the field of education, including Higher Education, by attracting students from all around the world to various colleges and universities. Student dropout or attrition is a major issue that is faced by Higher Education Institutions (HEI). Hence, we need effective techniques to identify student data that affects attrition. This research aims to explore the use of machine learning algorithms to predict student attrition at Rochester Institute of Technology- Dubai (RIT Dubai). In this research, the CRISP-DM (Cross-Industry Standard Process for Data Mining) methodology was used as it is widely used in projects based on …

An Analysis Of Machine Learning Hardware From Tenstorrent, Matthew Stephen Doerner

Theses

Machine learning is growing at an exponential rate in industry today, with contemporary machine learning models having parameter counts in the high hundred millions to trillions paired with a huge computing requirement. As the field continues to grow and the models get larger and more computationally expensive, there is a greater need for specialized hardware that can accelerate the inference and training computations that are required of these machine learning models. The design of a custom hardware device specifically made to accelerate machine learning models is very intricate, with software and hardware development that needs to work together at the …

Techniques To Overcome Energy Storage Limitations In Electric Vehicles, Matthew J. Hansen

All Graduate Theses and Dissertations, Fall 2023 to Present

Electric vehicles are becoming increasingly popular, battery limitations (cost, size, and weight) complicate electric vehicle adoption. While important research on battery development is ongoing, this dissertation discusses two main approaches to overcome those limitations within the existing battery technology paradigm. Those thrusts are: improving battery health through an optimal charging strategy and minimizing necessary battery size through dynamic wireless power transfer. In this dissertation, relevant literature is discussed, with opportunities for further development considered. Within the two thrusts, three objectives sharpen the focus of the research presented here. First, a planning tool is defined for a battery electric bus fleet. …

Ethical Imperatives And Challenges: Review Of The Use Of Machine Learning For Predictive Analytics In Higher Education, Emily Barnes, James Hutson, Karriem Perry

Faculty Scholarship

The escalating integration of machine learning (ML) in higher education necessitates a critical examination of its ethical implications. This article conducts a comprehensive review of the application of ML for predictive analytics within higher education institutions (HEIs), emphasizing the technology's potential to enhance student outcomes and operational efficiency. The study identifies significant ethical concerns, such as data privacy, informed consent, transparency, and accountability, that arise from the use of ML. Through a detailed analysis of current practices, this review underscores the need for HEIs to develop robust ethical frameworks and technological infrastructures to navigate these challenges effectively. The findings reveal …

Artificial Intelligence's Ability To Detect Online Predators, Olatilewa Osifeso

Electronic Theses, Projects, and Dissertations

Online child predators pose a danger to children who use the Internet. Children fall victim to online predators at an alarming rate, based on the data from the National Center of Missing and Exploited Children. When making online profiles and joining websites, you only need a name, an email and a password without identity verification. Studies have shown that online predators use a variety of methods and tools to manipulate and exploit children, such as blackmail, coercion, flattery, and deception. These issues have created an opportunity for skilled online predators to have fewer obstacles when it comes to contacting and …

Machine Learning Assisted Discovery Of High-Performance Perovskites For Hydrogen Production, Ximei Zhai

All Dissertations

Hydrogen has significantly contributed to the global energy transition as a clean energy carrier. However, traditional hydrogen production involves energy-intensive processes with heavy carbon emissions. High-temperature water splitting to generate hydrogen has become a promising route for clean hydrogen production. Directly utilizing concentrated solar heat to perform solar thermochemical hydrogen (STCH) production is considered one of the most eco-friendly hydrogen production options. At the same time, high-temperature electrolysis in protonic ceramic electrolysis cells (PCECs) can efficiently utilize intermittent and cost-effective renewable electricity to produce hydrogen. Perovskite oxides with good redox capability and tolerance for oxygen deficiency are considered the most …

Coarctation Duration And Severity Predict Risk Of Hypertension Precursors In A Preclinical Model And Hypertensive Status Among Patients, Arash Ghorbannia, Hilda Jurkiewicz, Lith Nasif, Abdillahi Ahmed, Jennifer Co-Vu, Mehdi Maadooliat, Ronald K. Woods, John F. Ladisa Jr.

Mathematical and Statistical Science Faculty Research and Publications

BACKGROUND:

Coarctation of the aorta (CoA) often leads to hypertension posttreatment. Evidence is lacking for the current >20 mm Hg peak-to-peak blood pressure (BP) gradient (BPGpp) guideline, which can cause aortic thickening, stiffening, and dysfunction. This study sought to find the BPGpp severity and duration that avoid persistent dysfunction in a preclinical model and test if predictors translate to hypertension status in patients with CoA.

METHODS:

Rabbits (n=75; 5–12/group) were exposed to mild, intermediate, or severe CoA (≤12, 13–19, ≥20 mm Hg BPGpp) for ≈1, 3, or 22 weeks using dissolvable and permanent sutures with thickening, stiffening, contraction, and endothelial …

Texture-Based Classification To Overcome Uncertainty Between Covid-19 And Viral Pneumonia Using Machine Learning And Deep Learning Techniques, Omar Farghaly, Priya Deshpande

Electrical and Computer Engineering Faculty Research and Publications

The SARS-CoV-2 virus, responsible for COVID-19, often manifests symptoms akin to viral pneumonia, complicating early detection and potentially leading to severe COVID pneumonia and long-term effects. Particularly affecting young individuals, the elderly, and those with weakened immune systems, the accurate classification of COVID-19 poses challenges, especially with highly dimensional image data. Past studies have faced limitations due to simplistic algorithms and small, biased datasets, yielding inaccurate results. In response, our study introduces a novel classification model that integrates advanced texture feature extraction methods, including GLCM, GLDM, and wavelet transform, within a deep learning framework. This innovative approach enables the effective …

Looking Beyond Mortality Prediction: Primary Care Physician Views Of Patients' Palliative Care Needs Predicted By A Machine Learning Tool, Lisa Rotenstein, Liqin Wang, Sophia N Zupanc, Akhila Penumarthy, John Laurentiev, Jan Lamey, Subrina Farah, Stuart Lipsitz, Nina Jain, David W Bates, Li Zhou, Joshua R Lakin

Faculty, Staff and Students Publications

Objectives: To assess primary care physicians' (PCPs) perception of the need for serious illness conversations (SIC) or other palliative care interventions in patients flagged by a machine learning tool for high 1-year mortality risk.

Methods: We surveyed PCPs from four Brigham and Women's Hospital primary care practice sites. Multiple mortality prediction algorithms were ensembled to assess adult patients of these PCPs who were either enrolled in the hospital's integrated care management program or had one of several chronic conditions. The patients were classified as high or low risk of 1-year mortality. A blinded survey had PCPs evaluate these patients for …

Identifying Temporomandibular Disorder Morphological Risk Factors Via Explainable Deep Learning And Multiscale Biomechanical Modeling, Shuchun Sun

All Dissertations

Clarifying multifactorial musculoskeletal disorder etiologies supports risk analysis and development of targeted prevention and treatment modalities. Deep learning enables comprehensive risk factor identification through systematic analysis of disease datasets but does not provide sufficient context for mechanistic understanding, limiting clinical applicability for etiological investigations. Conversely, multiscale biomechanical modeling can evaluate mechanistic etiology within the relevant biomechanical and physiological context. We propose a hybrid approach combining 3D explainable deep learning and multiscale biomechanical modeling; we applied this approach to investigate temporomandibular joint (TMJ) disorder etiology by systematically identifying risk factors and elucidating mechanistic relationships between risk factors and TMJ biomechanics and …

Development And Evaluation Of A Modeling Platform For Evaluating Immunotherapeutic Efficacy In The Tumor Microenvironment., Dylan Andrew Goodin

Electronic Theses and Dissertations

The tumor microenvironment (TME) represents the complex outcome of numerous tumor, stromal, and immune interactions, and whose composition can significantly affect treatment response. Particularly, immunotherapeutic efficacy is subject to multiple tumor-specific TME interactions that may be difficult to evaluate/predict clinically. Mathematical modelling has been formulated to evaluate specific aspects of the TME, including vasculature, ECM deposition, and immune-tumor interactions. However, the computational challenge of simulating multiple TME interactions has led to sacrificing varying degrees of model generalizability and clinical relevance. This work describes increased computational performance of a 3D continuum model that simulates tumor tissue, ECM, and vasculature using a …

Classifying Forecasts, Michael S. Drake, James R. Moon Jr., James D. Warren

Faculty Publications

We employ a novel machine learning technique to classify analysts’ forecast revisions into five types based on how the revision weighs publicly available signals. We label these forecast types as quant, sundry, contrarian, herder, and independent forecasts. Our tests reveal that a greater diversity of forecast types within the consensus is associated with increased consensus dispersion and improved consensus accuracy. Additionally, consensus diversity is associated with an improved information environment for firms, as reflected in reduced earnings announcement information asymmetry and volatility, higher earnings response coefficients, and faster price formation. Our study sheds light on how analysts revise their forecasts …

Semantic Segmentation Of Point Cloud Sequences Using Point Transformer V3, Marion Sisk

Master's Theses

Semantic segmentation of point clouds is a basic step for many autonomous systems including automobiles. In autonomous driving systems, LiDAR sensors are frequently used to produce point cloud sequences that allow the system to perceive the environment and navigate safely. Modern machine learning techniques for segmentation have predominately focused on single-scan segmentation, however sequence segmentation has often proven to perform better on common segmentation metrics. Using the popular Semantic KITTI dataset, we show that by providing point cloud sequences to a segmentation pipeline based on Point Transformer v3, we increase the segmentation performance between seven and fifteen percent when compared …

An Intelligent Water Source Discrimination Method For Water Inrushes From Coal Seam Roofs In The Inner Mongolia-Shaanxi Border Region, Wang Hao, Sun Junqing, Zeng Yifan, Shang Hongbo, Wang Tiantian, Qiao Wei

Coal Geology & Exploration

Water hazard on the coal seam proof induced by high-intensity coal mining are increasingly prominent in the Inner Mongolia-Shaanxi border region. The effective, accurate water-source discrimination of the water inrushes is the key to water hazard prevention. This study investigated three typical mines in the Inner Mongolia-Shaanxi border region. To this end, principal component analysis (PCA) was employed to extract principal components from 80 groups of groundwater samples. Then, with inorganic indicators K⁺+Na⁺, Ca²⁺, Mg²⁺, Cl⁻, SO₄ ²⁻, HCO₃ ⁻ and TDS and organic indicators UV_{254 …}