Physical Sciences and Mathematics Commons^™

Feature Selection In Multivariate Time Series Data For Enhanced Solar Flare Classification, Yagnashree Velanki

All Graduate Theses and Dissertations, Fall 2023 to Present

Solar flares are powerful eruptions of energy from the Sun that can cause disruptions to technology here on Earth, like communication systems, GPS, and power grids. To help manage these risks, it’s important to accurately identify and classify these solar flares before they cause problems. In our research, we focused on improving how we classify solar flares by looking at large sets of complex data collected over time. We used several techniques to find the most important factors that help us tell different types of solar flares apart. Each method has its strengths, so instead of relying on just one, …

Insights From Diverse Environments: Investigating The Fate And Transport Of Semivolatile Organic Contaminants In Alpine, Arctic, And Arid Regions, Jeffrey Perala-Dewey

All Graduate Theses and Dissertations, Fall 2023 to Present

This dissertation investigates the fate and transport of pollutants through four field-based studies in a diverse array of environments. These pollutants travel through the atmosphere and end up in places far from their original sources. In this work I explore how chemical properties of the pollutants and the environmental processes that are present change the way the pollutants travel through the environment. In the first study, pollutants from car engines were quantified in the vegetation along transects in three alpine valleys. Using these measurements, I investigated how the chemical properties of the pollutants impacted their interplay with wind systems in …

Mathematical And Statistical Methods To Harness Limited Data In Models For Ecological Space Use Under Global Change, Sarah C. Bogen

All Graduate Theses and Dissertations, Fall 2023 to Present

The dynamics of how plants and animals use space in their habitats has important implications for the fields of ecology and conservation. However, understanding and responding to these spatial and temporal dynamics is often limited by data availability, financial resources and biases. As average global temperatures increase, suitable habitats shift poleward and require local populations to move with suitable habitat, adapt to the changing environment, or risk extinction. Capacity to persist without movement may be estimated by considering changes to a combination of habitat characteristics. Capacity to track suitable habitat may be modeled through synthesizing information on species demographic mechanisms …

Using Luminescence Dating To Investigate Geomorphic And Archaeologic Features At The Wiggins Fork Bison Jump Complex, Northwestern Wyoming, Emma T. Krolczyk

All Graduate Theses and Dissertations, Fall 2023 to Present

The Wiggins Fork Bison Jump Complex, located in northwestern Wyoming, is comprised of thousands of human-placed rock piles, known as cairns, that were placed within lines to form an extensive network of drivelines and funnels leading to cliffs or steep slopes. These archaeological features were used by Native Americans as a bison hunting method, where the animals were driven off cliffs for harvest. This research uses luminescence dating, which determines when sediment was last exposed to sunlight, to date the timing of driveline construction and the deposition ages of the underlying river deposits.

Research focused on Jump #4, which had …

Assessing Khumbu Region Glaciers Mass Balance Change Using Geospatial Techniques From 2012 To 2024, Yashvi Jagatbhai Shah

Electronic Theses, Projects, and Dissertations

The glaciers in the Hindu Kush Himalaya are critical for sustaining the water supply of South Asia's major rivers, upon which over 230 million people, both in the mountainous regions and lowland areas, depend. Climate change is driving rapid glaciers melt near the summit of Mount Everest, located at an elevation of approximately 7,906 meters (25,938 feet). This has resulted in glacier thinning occurring up to 80 times faster than the original ice formation. However, due to their disposition in a complex topographic setting and inaccessible terrain, continuous glacier observations pose significant challenges.

This study evaluated the mass balance of …

A Comparative Study Of Patterns, Causes, And Impacts Of Data Breaches Across Geographical Regions And Time Frames, Bhavish Balsara

Electronic Theses, Projects, and Dissertations

The rise of digital technologies and interconnected systems has made data breaches a growing global concern. This culmination project explores the patterns, causes, and impacts of data breaches across various countries with varying levels of economic development and cybersecurity infrastructure from 2020 to 2023. This research aims to provide insights into the global landscape of data breaches and how they have evolved in recent years. The research questions are: (Q1) How do data breach patterns differ between countries with different levels of economic development and cybersecurity infrastructure? (Q2) What patterns and trends can be identified in data breaches when analyzing …

Container Runtime Vulnerability Mitigation Using User Namespace Isolation, Alexander Edsell

Electronic Theses, Projects, and Dissertations

Although containers have revolutionized application deployment by allowing for rapid and consistent deployment, their growing adoption has also raised significant security concerns. Each container is an isolated instance of an operating system that comes pre-packaged with the users desired applications. With multiple containers running on a host machine, an adversary can potentially break out of the container into the host machine. This project investigates the effectiveness of user namespace isolation as a security mechanism to mitigate container escape vulnerabilities that target the container’s runtime.

The research questions are: Question 1, does user namespace isolation mitigate container runtime vulnerabilities that target …

The Significance Of Continuous User Authentication On Mobile Devices, Mikayla Lawrence

Electronic Theses, Projects, and Dissertations

With the constant evolution of technology specifically on mobile devices, keeping personal and sensitive information safe has become increasingly vital. Continuous user authentication marks a major step forward in mobile security because it provides ongoing verification of user identity beyond the initial log in. This research explores the significance of continuous user authentication systems across mobile devices through literature-based analysis. The following research questions are addressed: (Q1) How effective are continuous user authentication methods in mitigating mobile device threats? (Q2) What are the vulnerabilities associated with continuous user authentication systems on mobile devices? (Q3) How do different continuous user authentication …

The Effects Of Covid-19 Lockdowns On Cybersecurity, Natalie Sanders

Electronic Theses, Projects, and Dissertations

The COVID-19 pandemic and subsequent lockdowns forced many Americans to quickly adapt to working from home, many of which had never done so in the past. In addition, organizations were forced to modify security policies and protocols to allow for remote access to sensitive information. The rapid change in security policies as well as the sudden growth of remote access during the pandemic presented a broader landscape for cyber criminals to attack. In this report, we review the number and type of cyberattacks reported from two (2) years before the pandemic through two (2) years after (1998 through 2023), to …

Comparison Of Statistical And Machine Learning Genomic Prediction Methods In Plant Breeding: Case Studies In Maize And Soybean, Igor Kuivjogi Fernandes

Graduate Theses and Dissertations

Plant breeding is essential to increase genetic gain and food production worldwide. This study was conducted to evaluate new ways to use machine learning (ML) to tackle plant breeding challenges, where two ideas were tested — the first chapter focuses on how to combine genetic and environmental data using ML to improve the prediction of maize grain yield in multi-environment trials, while the second chapter centers on how to couple feature selection of molecular markers with ML to enhance prediction of yield in soybean, and, in both cases, ML approaches were compared to well-established statistical methods greatly adopted by the …

Exploiting Randomness In Secret Sharing, Cailyn Bass

All Theses

Shamir's (k,n)-threshold scheme is a method for sharing a secret among n participants such that any group of k or more participants can recover the secret. Additionally, any group of participants with size less than k should learn nothing about the secret. The scheme works by distributing a share to each participant, where each share is a linear combination of the secret and k-1 random symbols. This allows any group of k or more participants to solve a linear system to compute the secret. Any group of less than k participants does not have enough to determine anything about the …

Multiple Cut Systems Of Cannabis Sativa L. For Micropropagation Without The Use Of Cytokinins, Molly Mckay

All Theses

Micropropagation is a technique used to propagate clean, high-quality clonal stock plants using plant growth regulators (PGR) in a single harvest batch system. This two-part study uses an in vitro multiple harvest system “hedging” combined with the fed- batch media process, to observe (1) the effects of modified physical state and (2) LED light quality to improve shoot production and quality. In Experiment 1, four genotypes of Cannabis sativa (‘Cherry 1’,‘BaOx’,‘T1’,‘Peach’) were observed in three different physical states: stationary agar (A), stationary Oasis® infused with liquid (OIL_s) and agitated Oasis® infused with liquid (OIL_a). Fifteen …

Bounding The Convex Hull Relaxation Of The Unit Commitment Problem With The Shapley-Folkman Theorem, Lauren Henderson

All Theses

The Unit Commitment (UC) problem finds an optimal schedule for a set of generators by minimizing the total operation cost subject to demand and operational constraints. The UC problem is often modeled with a mixed-integer linear program (MILP). We employ the Shapley-Folkman Theorem to provide a bound on the size of fractional solutions of its convex hull relaxation. This result is used to obtain a bound on the optimality gap between the MILP and the convex hull relaxation, which is further tightened using several problem-specific properties of UC. We conduct extensive numerical experiments to study the tightness of this threshold, …

Unraveling Phytosulfokine Trafficking In Arabidopsis Thaliana Using Fiber-Optic Fluorescence Microscopy, Issaka Obuaba

Electronic Theses and Dissertations

As sessile organisms, plants manage stress through complex signaling networks involving phytohormones such as phytosulfokine (PSK). PSK, a disulfated pentapeptide, regulates plant growth, development, and stress responses by interacting with specific PSK receptors (PSKRs). In this study, we explored the trafficking dynamics of PSK, its post-application fate, and the synthesis of an analog. We administered both native PSK and a fluorescent version tagged with TAMRA (5(6)-carboxytetramethylrhodamine) to various Arabidopsis thaliana genotypes, including wild type, a PSKR-deficient mutant, and a strain overexpressing PSKR1 tagged with green fluorescent protein (GFP) over the wild-type background. Fiber-optic fluorescence microscopy revealed that receptor presence influences …

Mapping Amis Mill In Rogersville, Tennessee In The 1780s Using Geophysical Methods And Remote Sensing, Amy Collins

Electronic Theses and Dissertations

Geophysics and remote sensing were used at the Amis Mill and Homesite to determine the location of structures during the 1780s. Areas of interest were surveyed using shallow geophysical and remote sensing techniques including the orchard, horse pasture, original kitchen on the east side of the house, the west side of the house, plowed field, store site, and cemetery. After analyzing the results of the geophysical surveys and remote sensing, 22 sites were chosen to test inside and outside potential features. A large rectangular feature, possibly a tavern with a cellar, was discovered. Four other features were identified, three of …

Landslide Inventory And Unstable Slope Monitoring Along Highways In Eastern Tennessee, Robert Mcsweeney

Electronic Theses and Dissertations

This research introduces an unstable slope management program (USMP) for Tennessee based on federal slope management standards, along with improved methods for landslide monitoring with unmanned aerial systems (UAS) lidar and photogrammetry. In mountainous regions, monitoring slope hazards is a critical function of transportation management. A mobile field assessment form created with Survey123 was used to collect 22 unstable slope ratings in eastern Tennessee. Location points were appended with photographs, notes, and site information. Landslide scores ranged from 325 (Fair) to 1005 (Poor). UAS monitoring of a slow-moving soil landslide along I-40 near Rockwood, TN, produced high-resolution lidar and photogrammetry …

Investigating Phytosulfokine Trafficking: Insights Into The Role Of Phytohormones In Plant Signaling, Martin Tindi

Electronic Theses and Dissertations

Phytosulfokine is a critical signaling peptide involved in plant growth, development, and stress responses. This thesis investigates the trafficking mechanisms of Phytosulfokine (PSK) in Arabidopsis thaliana, employing fluorescently labelled PSK for non-destructive imaging using a custom-built fiber-optic fluorescence microscope. The study aims to determine the mobility and fate of PSK in plant tissues. Experimental results reveal the short-range and long-range movement of PSK across leaves. Additionally, the project highlights the optimization of non-destructive imaging techniques and the synthesis of caged dexamethasone to facilitate optogenetic studies. These findings provide new insights into PSK trafficking mechanisms in plants, enhancing our understanding …

Neural Network Architecture Search Enabled Wide-Deep Learning (Nas-Wd) For Spatially Heterogenous Property Awared Chicken Woody Breast Classification And Hardness Regression, Chaitanya Pallerla, Yihong Feng, Casey M. Owens, Ramesh Bahadur Bist, Siavash Mahmoudi, Pouya Sohrabipour, Amirreza Davar, Dongyi Wang

Poultry Science Faculty Publications and Presentations

Due to intensive genetic selection for rapid growth rates and high broiler yields in recent years, the global poultry industry has faced a challenging problem in the form of woody breast (WB) conditions. This condition has caused significant economic losses as high as $200 million annually, and the root cause of WB has yet to be identified. Human palpation is the most common method of distinguishing a WB from others. However, this method is time-consuming and subjective. Hyperspectral imaging (HSI) combined with machine learning algorithms can evaluate the WB conditions of fillets in a non-invasive, objective, and high-throughput manner. In …

Triadic Temporal-Semantic Alignment For Weakly-Supervised Video Moment Retrieval, Jin Liu, Jialong Xie, Fengyu Zhou, Shengfeng He

Research Collection School Of Computing and Information Systems

Video Moment Retrieval (VMR) aims to identify specific event moments within untrimmed videos based on natural language queries. Existing VMR methods have been criticized for relying heavily on moment annotation bias rather than true multi-modal alignment reasoning. Weakly supervised VMR approaches inherently overcome this issue by training without precise temporal location information. However, they struggle with fine-grained semantic alignment and often yield multiple speculative predictions with prolonged video spans. In this paper, we take a step forward in the context of weakly supervised VMR by proposing a triadic temporalsemantic alignment model. Our proposed approach augments weak supervision by comprehensively addressing …

Harnessing Collective Structure Knowledge In Data Augmentation For Graph Neural Networks, Rongrong Ma, Guansong Pang, Ling Chen

Research Collection School Of Computing and Information Systems

Graph neural networks (GNNs) have achieved state-of-the-art performance in graph representation learning. Message passing neural networks, which learn representations through recursively aggregating information from each node and its neighbors, are among the most commonly-used GNNs. However, a wealth of structural information of individual nodes and full graphs is often ignored in such process, which restricts the expressive power of GNNs. Various graph data augmentation methods that enable the message passing with richer structure knowledge have been introduced as one main way to tackle this issue, but they are often focused on individual structure features and difficult to scale up with …

An Aggregate Matching And Pick-Up Model For Mobility-On-Demand Services, Xinwei Li, Jintao Ke, Hai Yang, Hai Wang, Yaqian Zhou

Research Collection School Of Computing and Information Systems

This paper presents an Aggregate Matching and Pick-up (AMP) model to delineate the matching and pick-up processes in mobility-on-demand (MoD) service markets by explicitly considering the matching mechanisms in terms of matching intervals and matching radii. With passenger demand rate, vehicle fleet size and matching strategies as inputs, the AMP model can well approximate drivers’ idle time and passengers’ waiting time for matching and pick-up by considering batch matching in a stationary state. Properties of the AMP model are then analyzed, including the relationship between passengers’ waiting time and drivers’ idle time, and their changes with market thickness, which is …

Towards Unified Multimodal Editing With Enhanced Knowledge Collaboration, Kaihang Pan, Zhaoyu Fan, Juncheng Li, Qifan Yu, Hao Fei, Siliang Tang, Richang Hong, Hanwang Zhang, Qianru Sun

Research Collection School Of Computing and Information Systems

The swift advancement in Multimodal LLMs (MLLMs) also presents significant challenges for effective knowledge editing. Current methods, including intrinsic knowledge editing and external knowledge resorting, each possess strengths and weaknesses, struggling to balance the desired properties of reliability, generality, and locality when applied to MLLMs. In this paper, we propose UniKE, a novel multimodal editing method that establishes a unified perspective and paradigm for intrinsic knowledge editing and external knowledge resorting. Both types of knowledge are conceptualized as vectorized key-value memories, with the corresponding editing processes resembling the assimilation and accommodation phases of human cognition, conducted at the same semantic …

Reinforcement Learning Based Online Request Scheduling Framework For Workload-Adaptive Edge Deep Learning Inference, Xinrui Tan, Hongjia Li, Xiaofei Xie, Lu Guo, Nirwan Ansari, Xueqing Huang, Liming Wang, Zhen Xu, Yang Liu

Research Collection School Of Computing and Information Systems

The recent advances of deep learning in various mobile and Internet-of-Things applications, coupled with the emergence of edge computing, have led to a strong trend of performing deep learning inference on the edge servers located physically close to the end devices. This trend presents the challenge of how to meet the quality-of-service requirements of inference tasks at the resource-constrained network edge, especially under variable or even bursty inference workloads. Solutions to this challenge have not yet been reported in the related literature. In the present paper, we tackle this challenge by means of workload-adaptive inference request scheduling: in different workload …

Mvgamba : Unify 3d Content Generation As State Space Sequence Modeling, Xuanyu Yi, Zike Wu, Qiuhong Shen, Qingshan Xu, Pan Zhou, Joo-Hwee Lim, Shuicheng Yan, Xinchao Wang, Hanwang Zhang

Research Collection School Of Computing and Information Systems

Recent 3D large reconstruction models (LRMs) can generate high-quality 3D content in sub-seconds by integrating multi-view diffusion models with scalable multi-view reconstructors. Current works further leverage 3D Gaussian Splatting as 3D representation for improved visual quality and rendering efficiency. However, we observe that existing Gaussian reconstruction models often suffer from multi-view inconsistency and blurred textures. We attribute this to the compromise of multi-view information propagation in favor of adopting powerful yet computationally intensive architectures (e.g., Transformers). To address this issue, we introduce MVGamba, a general and lightweight Gaussian reconstruction model featuring a multi-view Gaussian reconstructor based on the RNN-like State …

A Data-Driven Approach For Automated Multi-Site Competitive Facility Location, Ming Hui Tan, Kar Way Tan, Hoong Chuin Lau

Research Collection School Of Computing and Information Systems

This paper addresses the challenge of optimal retail expansion in competitive urban environments through a novel approach to the Competitive Facility Location (CFL) problem. Traditional methods for solving CFL problems often struggle with large-scale scenarios, relying on manual pre-selection of candidate sites and imposing limitations on the number of new locations. Our approach leverages Adaptive Large Neighborhood Search (ALNS) enhanced with data enrichment techniques, including community detection on road networks and population weighting based on mobility data. We developed two ALNS variants: Community Geometric Centroid (CGC-ALNS) and Population Weighted Centroid (PWC-ALNS). These methods automate site selection, eliminating manual pre-selection while …

4-Bit Shampoo For Memory-Efficient Network Training, Sike Wang, Pan Zhou, Jia Li, Hua Huang

Research Collection School Of Computing and Information Systems

Second-order optimizers, maintaining a matrix termed a preconditioner, are superior to first-order optimizers in both theory and practice. The states forming the preconditioner and its inverse root restrict the maximum size of models trained by second-order optimizers. To address this, compressing 32-bit optimizer states to lower bitwidths has shown promise in reducing memory usage. However, current approaches only pertain to first-order optimizers. In this paper, we propose the first 4-bit second-order optimizers, exemplified by 4-bit Shampoo, maintaining performance similar to that of 32-bit ones. We show that quantizing the eigenvector matrix of the preconditioner in 4-bit Shampoo is remarkably better …

Generative Semi-Supervised Graph Anomaly Detection, Hezhe Qiao, Qingsong Wen, Xiaoli Li, Ee-Peng Lim, Guansong Pang

Research Collection School Of Computing and Information Systems

This work considers a practical semi-supervised graph anomaly detection (GAD) scenario, where part of the nodes in a graph are known to be normal, contrasting to the extensively explored unsupervised setting with a fully unlabeled graph. We reveal that having access to the normal nodes, even just a small percentage of normal nodes, helps enhance the detection performance of existing unsupervised GAD methods when they are adapted to the semi-supervised setting. However, their utilization of these normal nodes is limited. In this paper we propose a novel Generative GAD approach (namely GGAD) for the semi-supervised scenario to better exploit the …

Sampdetox : Black-Box Backdoor Defense Via Perturbation-Based Sample Detoxification, Yanxin Yang, Chentao Jia, Dengke Yan, Ming Hu, Tianlin Li, Xiaofei Xie, Xian Wei, Mingsong Chen

Research Collection School Of Computing and Information Systems

The advancement of Machine Learning has enabled the widespread deployment of Machine Learning as a Service (MLaaS) applications. However, the untrustworthy nature of third-party ML services poses backdoor threats. Existing defenses in MLaaS are limited by their reliance on training samples or white-box model analysis, highlighting the need for a black-box backdoor purification method. In our paper, we attempt to use diffusion models for purification by introducing noise in a forward diffusion process to destroy backdoors and recover clean samples through a reverse generative process. However, since a higher noise also destroys the semantics of the original samples, it still …

Reevo: Large Language Models As Hyper-Heuristics With Reflective Evolution, Haoran Ye, Jiarui Wang, Zhiguang Cao, Federico Berto, Chuanbo Hua, Haeyeon Kim, Jinkyoo Park, Guojie Song

Research Collection School Of Computing and Information Systems

The omnipresence of NP-hard combinatorial optimization problems (COPs) compels domain experts to engage in trial-and-error heuristic design process. The long-standing endeavor of design automation has gained new momentum with the rise of large language models (LLMs). This paper introduces Language Hyper-Heuristics (LHHs), an emerging variant of Hyper-Heuristics that leverages LLMs for heuristic generation, featuring minimal manual intervention and open-ended heuristic spaces. To empower LHHs, we present Reflective Evolution (ReEvo), a generic searching framework that emulates the reflective design approach of human experts while far surpassing human capabilities with its scalable LLM inference, Internet-scale domain knowledge, and powerful evolutionary search. Evaluations …

Flexfl: Heterogeneous Federated Learning Via Apoz-Guided Flexible Pruning In Uncertain Scenarios, Zekai Chen, Chentao Jia, Ming Hu, Xiaofei Xie, Anran Li, Mingsong Chen

Research Collection School Of Computing and Information Systems

Along with the increasing popularity of Deep Learning (DL) techniques, more and more Artificial Intelligence of Things (AIoT) systems are adopting federated learning (FL) to enable privacy-aware collaborative learning among AIoT devices. However, due to the inherent data and device heterogeneity issues, existing FL-based AIoT systems suffer from the model selection problem. Although various heterogeneous FL methods have been investigated to enable collaborative training among heterogeneous models, there is still a lack of i) wise heterogeneous model generation methods for devices, ii) consideration of uncertain factors, and iii) performance guarantee for large models, thus strongly limiting the overall FL performance. …