Physical Sciences and Mathematics Commons^™

Enhancing Visual Grounding In Vision-Language Pre-Training With Position-Guided Text Prompts, Alex Jinpeng Wang, Pan Zhou, Mike Zheng Shou, Shuicheng Yan

Research Collection School Of Computing and Information Systems

Vision-Language Pre-Training (VLP) has demonstrated remarkable potential in aligning image and text pairs, paving the way for a wide range of cross-modal learning tasks. Nevertheless, we have observed that VLP models often fall short in terms of visual grounding and localization capabilities, which are crucial for many downstream tasks, such as visual reasoning. In response, we introduce a novel Position-guided Text Prompt ( PTP ) paradigm to bolster the visual grounding abilities of cross-modal models trained with VLP. In the VLP phase, PTP divides an image into N x N blocks and employs a widely-used object detector to identify objects …

Vaid: Indexing View Designs In Visual Analytics System, Lu Ying, Aoyu Wu, Haotian Li, Zikun Deng, Ji Lan, Jiang Wu, Yong Wang, Huamin Qu, Dazhen Deng, Yingcai Wu

Research Collection School Of Computing and Information Systems

Visual analytics (VA) systems have been widely used in various application domains. However, VA systems are complex in design, which imposes a serious problem: although the academic community constantly designs and implements new designs, the designs are difficult to query, understand, and refer to by subsequent designers. To mark a major step forward in tackling this problem, we index VA designs in an expressive and accessible way, transforming the designs into a structured format. We first conducted a workshop study with VA designers to learn user requirements for understanding and retrieving professional designs in VA systems. Thereafter, we came up …

Exploring Diffusion Time-Steps For Unsupervised Representation Learning, Zhongqi Yue, Jiankun Wang, Qianru Sun, Lei Ji, Eric I-Chao Chang, Hanwang Zhang

Research Collection School Of Computing and Information Systems

Representation learning is all about discovering the hidden modular attributes that generate the data faithfully. We explore the potential of Denoising Diffusion Probabilistic Model (DM) in unsupervised learning of the modular attributes. We build a theoretical framework that connects the diffusion time-steps and the hidden attributes, which serves as an effective inductive bias for unsupervised learning. Specifically, the forward diffusion process incrementally adds Gaussian noise to samples at each time-step, which essentially collapses different samples into similar ones by losing attributes, e.g., fine-grained attributes such as texture are lost with less noise added (i.e., early time-steps), while coarse-grained ones such …

Regret-Based Defense In Adversarial Reinforcement Learning, Roman Belaire, Pradeep Varakantham, Thanh Hong Nguyen, David Lo

Research Collection School Of Computing and Information Systems

Deep Reinforcement Learning (DRL) policies are vulnerable to adversarial noise in observations, which can have disastrous consequences in safety-critical environments. For instance, a self-driving car receiving adversarially perturbed sensory observations about traffic signs (e.g., a stop sign physically altered to be perceived as a speed limit sign) can be fatal. Leading existing approaches for making RL algorithms robust to an observation-perturbing adversary have focused on (a) regularization approaches that make expected value objectives robust by adding adversarial loss terms; or (b) employing "maximin'' (i.e., maximizing the minimum value) notions of robustness. While regularization approaches are adept at reducing the probability …

Optimizing Adult Learner Success: Applying Random Forest Classifier In Higher Education Predictive Analytics, Emily Barnes, James Hutson, Karriem Perry

Faculty Scholarship

This study examines the application of the Random Forest Classifier (RF) model in predicting academic success among adult learners in higher education. It focuses on evaluating the model's effectiveness using key statistical measures like accuracy, precision, recall, and F1 score across a comprehensive dataset from 2013–14 to 2021–22, which includes variables such as age, ethnicity, gender, Pell Grant eligibility, and academic performance metrics. The research highlights the RF model's capability to handle large datasets with varying data types and demonstrates its superiority over traditional regression models in predictive accuracy. Through an iterative process, the study refines the RF model to …

Local Existence Of Solutions To A Nonlinear Autonomous Pde Model For Population Dynamics With Nonlocal Transport And Competition, Michael R. Lindstrom

School of Mathematical and Statistical Sciences Faculty Publications and Presentations

Highlights

• Partial differential equation models are ubiquitous in applied sciences.

• A partial differential equation based in ecology is studied for solution existence.

• Energy methods and convergence analysis lead to local classical solutions.

Abstract

In this paper, we prove that a particular nondegenerate, nonlinear, autonomous parabolic partial differential equation with nonlocal mass transfer admits the local existence of classical solutions. The equation was developed to qualitatively describe temporal changes in population densities over space through accounting for location desirability and fast, long-range travel. Beginning with sufficiently regular initial conditions, through smoothing the PDE and employing energy arguments, we obtain a sequence …

Determining The Viability Of Marine Sensor Construction, Roman Sequeira

Honors College

I intend to determine the viability of building a sensor that can be deployed in marine environments. Viability is defined as a summary of economic affordability, practicality of construction and deployment, knowledge required, and effectiveness. In order to get the best results, our sensor must be able to be modified to suit individual circumstances, constructed out of easy to obtain materials, be robust enough to withstand less than lab grade environments, and most importantly function well enough to be worth the effort of building it. I have built a sensor using the cheapest and most widely available options available. It …

Ghost Connect-Net: A Connectivity-Based Companion Network To Enhance Pruning Methods, Mary Isabelle Wisell

Honors College

Deep neural network (DNN) approaches excel in various real-world applications like robotics and computer vision, yet their computational demands and memory requirements hinder usability on advanced devices. Also, larger models heighten overparameterization risks, making networks more vulnerable to input disturbances. Recent studies aim to boost DNN efficiency by trimming redundant neurons or filters based on task relevance. Instead of introducing a new pruning method, this project aims to enhance existing techniques by introducing a companion network, Ghost Connect-Net (GC-Net), to monitor the connections in the original network. The initial weights of GC- Net are equal to the connectivity measurements of …

The Conservation Project: An Exploration Of Multimedia In Ocean Conservation, Ilaria Bardini

Honors College

The Conservation Project ties in many elements of multimedia and its possible applications in marine conservation. The purpose of this thesis was to develop new skills in videography, photography, podcast production, and website development through which to deepen my understanding of the multimedia as a tool in science, through the development of a website.

Understanding The Connection Between Water, Fish, And Pfas Concentrations: Implications Of Fish Diet And Species-Specific Variability, Ece Yeldan

Honors College

Perfluoroalkyl and polyfluoroalkyl substances, commonly known as PFAS or “forever chemicals,” have emerged as a significant concern for both human health and the environment. These persistent compounds permeate various aspects of commerce, leading to widespread exposure and the cycling of these compounds through environmental feedback routes. Among the sources of exposure to humans, fish consumption stands out prominently. Recreational and sustenance fishing, cherished by Maine residents and indigenous communities alike, underscores the need to comprehend PFAS dynamics in fish and water. To explore the variability of PFAS accumulation in fish, water, and sediment, I conducted a comprehensive literature review. The …

Assessing The Influence Of Interseeded Cover Crops On Beneficial Arthropod Abundance In A Northeastern Agroecosystem, Charles Cooper

Honors College

Conservation agriculture approaches are gaining traction as the planet’s food system grapples with climate change, oil depletion, and rampant environmental degradation (Palm et al., 2014). Cover cropping is an integral practice of conservation agriculture. Ground dwelling arthropods play an important role in agroecosystems, providing ecosystem services including seed predation and nutrient cycling. Because the relationship between cover crops and arthropod abundance are likely influenced by management conditions, I investigated arthropod abundance in a field interseeded with cover crops on a research farm in Maine, United States. Interseeding is an emerging practice in the northeastern United States, with potential to address …

Using Cdom Optical Properties To Estimate Its Source, Distribution, And Doc Conentration In Maine Estuaries, Camille Michaud

Honors College

Colored dissolved organic matter (CDOM) contributes largely to ocean color, and therefore, influences processes and applications that interact with light in the ocean such as remote sensing and biogeochemistry. It is a component part of dissolved organic carbon (DOC) and there is a strong relationship in coastal waters between CDOM absorption and DOC concentration. CDOM optical properties also vary with salinity and source. These important aspects of CDOM absorption led us to collect CDOM samples across the salinity gradients of 3 estuaries in Maine: the Penobscot, Damariscotta, and Sheepscot. We sampled stations along the salinity gradient in each estuary from …

Ecological Effects And Ethics Of A Species Range Shift To High Elevation Ponds, Ava Ardito

Honors College

As temperatures warm, the climate crisis is having a significant influence on species across the world and altering their distributions. Colorado's high elevation ponds have recently observed a species range shift, traveling to higher altitudes than they have previously inhabited. The Mexican Cut Nature Preserve near the Rocky Mountain Biological Laboratory (RMBL) in Gothic, CO, is home to a prominent caddisfly Limnephilus externus, but recently a new range-shifting species, Nemotaulius hostilis, has appeared. Caddisflies play an essential role in nutrient cycling and putting energy back into the food web. The range-shifting species is known to excrete high rates of phosphorus …

Evaluating The Accuracy Of Different Remote Sensing Methods Used To Derive Absorption From Chromophoric Dissolved Organic Matter In Maine Estuaries, Andrea Rudai

Honors College

Current knowledge of dissolved organic carbon (DOC) in Maine estuaries is based on limited field research. By using chromophoric dissolved organic matter (CDOM) as a proxy, it is possible to use satellite images to estimate the concentration of DOC in an entire body of water. This study aims to evaluate different methods that calculate absorption by CDOM from ocean color remote sensing using images taken by Landsat 9 of the Sheepscot, Penobscot, and Damariscotta rivers. The study revealed that although there were some similarities in along-estuary trends between the remotely- sensed absorption and absorption measured in situ, the algorithms that …

Analysis Of Nonsmooth Neural Mass Models, Cadi Howell

Honors College

Neural activity in the brain involves a series of action potentials that represent “all or nothing” impulses. This implies the action potential will only “fire” if the mem- brane potential is at or above a specific threshold. The Wilson-Cowan neural mass model [6, 28] is a popular mathematical model in neuroscience that groups excita- tory and inhibitory neural populations and models their communication. Within the model, the on/off behavior of the firing rate is typically modeled by a smooth sigmoid curve. However, a piecewise-linear (PWL) firing rate function has been considered in the Wilson-Cowan model in the literature (e.g., see …

Physical Forcing Of Productivity, Biomass, And Demography Of Ascophyllum Nodosum, Henry Guy

Honors College

The intertidal coastlines of Maine are dominated by a near monoculture of Ascophyllum nodosum. This species of macroalgae is foundational to the ecosystems in which it inhabits, so what controls the Ascophyllum nodosum controls the coasts. A. nodosum has been documented as a steady-state species in terms of biomass, canopy height, and primary productivity. Algal biomass results from algal recruitment, primary productivity (growth rate), and environmental disturbance (rate of biomass loss). My study estimated the abiotic disturbance of the environment on biotic factors ofAscophyllum nodosum. Physical forcings of the environment, ice, water motion, and light, drive Ascophyllum nodosum abundance, distribution, …

Microplastics: The Relationship Between Wastewater And Amount Of Microplastics In The Environment, Julia Muth

Honors College

In this paper, the contribution of wastewater to the cycle of microplastics in the environment is addressed. Several articles, journal papers and reliable websites were consulted for a literature review on this topic. The sources suggest that approximately 10% of the plastics in the influent exit in the effluent, and further analysis showed that the effluent wastewater and sludge produced in the USA contributes approximately 89,100 tons of plastic yearly (Basic Information about Biosolids 2023; Carr, Liu, and Tesoro 2016; Center for Sustainable Systems 2023; EPA 2023; Okoffo, Tscharke, and Thomas 2023). This outflow of plastic into the environment causes …

An Investigation Into Problem Solving In The Calculus Iii Classroom, Joseph Godinez

Honors College

The importance of tertiary education has grown to new heights, especially in the United States. A critical component of successful modern professionals remains the ability to employ problem-solving strategies and techniques. This study seeks to investigate initial problem-solving strategies employed by post-secondary students enrolled in Calculus II when presented with problems common to integral calculus. In- person pair-wise interviews were conducted asking six participants to sort integrals into categories based on the technique they would use to solve it. Participant responses were analyzed using a concept image composed of general and topic-specific symbolic forms, related conceptual images and concept definitions, …

Characterization Of Carbonaceous Fault Rocks, Pioneer Fault Zone, South-Central Idaho, Genna Baldassarre

All Graduate Reports and Creative Projects, Fall 2023 to Present

Differing crystallinities of carbonaceous material are common in fault rocks across a range of geologic settings and spatial scales and may provide constraints on strain rate, the nature of fault slip, fluid-rock interactions, and temperature variations over the earthquake cycle. The Pioneer fault at Little Fall Creek in south-central Idaho provides an excellent opportunity to study nanostructure changes of carbonaceous matter as a function of fault deformation. At this location, the Pioneer Fault exhibits a well-defined principal slip zone (PSZ) composed of multi-layered white siliceous mineralization and black carbon to graphite with a continuously exposed adjacent damage zone that includes …

Convergence Estimate Of Minimal Residual Methods And Random Sketching Of Krylov Subspace Methods, Peter Westerbaan

All Dissertations

This study concerns two main issues in numerical linear algebra: convergence estimate of minimal residual methods based on explicit construction of approximate min-max polynomials for in- definite matrices, and development and analysis of Krylov subspace methods using non-orthonormal basis vectors based on random sketching. For a matrix A with spectrum Λ(A), it is well known that the min-max polynomial problem min max |pk (z)| pk ∈Pk, pk (0)=1, z∈Λ(A) is used to bound the relative error of Krylov subspace minimum residual methods or similar methods. For a symmetric positive definite matrix A, the min-max polynomial for the Conjugate Gradient (CG) …

Models Of Functional Redundancy In Ecological Communities, Sandra Annie Tsiorintsoa

All Dissertations

Functional redundancy is the number of taxa that perform a given function within a given community. In most systems, high levels of functional redundancy are important, because they contribute to ecosystem stability. However, we currently have very little understanding of why functional redundancy varies among communities. One possible factor that could affect functional redundancy is environmental complexity. Many studies show that simplified ecosystems harbor communities with lower taxon diversity. What is less clear is if this simplicity and lower taxon diversity also affects functional redundancy. To answer this question, we use metacommunity models to explore the connection between environmental complexity …

Time Resolved X-Ray Spectroscopy Of Highly Charged Ar, Nd, And Pr, Timothy Burke

All Dissertations

Highly Charged Ions (HCIs) may be considered ideal mini-laboratory in which one can study the physics of matter and light in an environment of high internal electric field that can not be recreated with standard lab equipment. The remaining electron(s) exist in the extremely large electric field of the nucleus and therefore measurements of electronic transitions in these systems provide stringent tests of our understanding of physics in extreme conditions. Quantum electrodynamics (QED) despite being a powerful theory exhibits large discrepancies for systems under extreme conditions. The work here investigates the atomic properties within non-Maxwellian plasmas. The HCI plasmas studied …

Applications Of Conic Programming Reformulations, Sarah Kelly

All Dissertations

In general, convex programs have nicer properties than nonconvex programs. Notably, in a convex program, every locally optimal solution is also globally optimal. For this reason, there is interest in finding convex reformulations of nonconvex programs. These reformulation often come in the form of a conic program. For example, nonconvex quadratically-constrained quadratic programs (QCQPs) are often relaxed to semidefinite programs (SDPs) and then tightened with valid inequalities. This dissertation gives a few different problems of interest and shows how conic reformulations can be usefully applied.

In one chapter, we consider two variants of the trust-region subproblem. For each of these …

Design, Fabrication, And Characterization Of Advanced High-Power Single-Mode 9xxnm Semiconductor Lasers, Xiaolei Zhao

All Dissertations

This thesis presents the comprehensive design, fabrication, and demonstration of advanced high-power, high-efficiency single-mode semiconductor lasers operating at a wavelength of 9xxnm. We begin with the design of the laser epitaxial structure, serving as the cornerstone for achieving high-power high-efficiency lasers. Our methodology integrates a semi-analytical calculation model, which accounts for Longitudinal Spatial Hole Burning (LSHB) and Two-Photon Absorption (TPA) effects, facilitating a thorough exploration of how design parameters influence output power and conversion efficiency. This approach offers an effective and time-efficient epitaxial structure optimization strategy compared to conventional full 3D simulation models.

Subsequently, we demonstrate high-power, high-efficiency ridge waveguide …

Engineering Multifunctional Silicon Nanostructures From Biorenewable Cellulose Nanocrystals, Nancy Chen

All Dissertations

The imperative search for alternative materials to address the pressing demand for advance energy storage is underscored by the escalating environmental predicaments. Lithium-ion batteries (LIBs) with graphite anodes have become the benchmark in energy storage; however, they are approaching a saturation point in terms of energy density. Silicon emerges as a promising contender to supplant graphite, owing to its profuse availability, cost-effectiveness, and impressive specific capacity of 4200 mAh g^-1. By integrating silicon anodes, LIBs stand to undergo a radical transformation, markedly diminishing in weight and size, thus heralding a novel wave of compact, lightweight energy storage systems. …

Final Butte Priority Soils Operable Unit Repository Screening Report, Pioneer Technical Services, Inc.

Silver Bow Creek/Butte Area Superfund Site

No abstract provided.

The Impact Of Ai On Ux: Challenges And Opportunities, Susan Stephanie Wells

Theses

Integrating artificial intelligence (AI) in user experience (UX) design is reshaping the field of UX, offering new opportunities and challenges for designers. This thesis project explores the multifaceted relationship between AI and UX design, focusing on the challenges, opportunities, and skills demanded of UX designers in the age of AI. Through a review of academic research and real-world experiences, this project studies the impact of AI on web design processes, UX testing, and data analysis. Key findings highlight the transformative potential of AI in enhancing user experiences, from suggesting website structures to facilitating UX testing and data analysis.

Comparative analysis …

Ethical Imperatives In Ai-Driven Educational Assessment: Framework And Implications, Ming Soon Tristan Lim

Dissertations and Theses Collection (Open Access)

This dissertation embarks on an extensive exploration of the ethical challenges emerging from the integration of AI in educational assessments. It uncovers the complex interplay between AI and the ethical imperatives these technologies pose within educational assessments.

Amidst the rapid development of AI-enabled educational technologies, such as Ubiquitous, Adaptive, and Immersive technologies, this research identifies a notable gap in literature specifically concerning the ethical imperatives and implications of AI in educational assessments. Addressing this gap, the dissertation has three primary objectives: to comprehend and analyze the underpinning educational technologies driving assessments, to elucidate the intricate relationship between AI, ethics, and …

Using Pre-Trained Models For Vision-Language Understanding Tasks, Rui Cao

Dissertations and Theses Collection (Open Access)

In recent years, remarkable progress has been made in Artificial Intelligence (AI), with an increasing focus on integrating AI systems into people’s daily lives. In the context of our diverse world, research attention has shifted towards applying AI to multimodal understanding tasks. This thesis specifically addresses two key modalities, namely, vision and language, and explores Vision-Language Understanding (VLU).

In the past, addressing VLU tasks involved training distinct models from scratch using task-specific data. However, limited by the amount of training data, models may easily overfit the training data and fail to generalize. A recent breakthrough is the development of Pre-trained …

Enabling And Optimizing Multi-Modal Sense-Making For Human-Ai Interaction Tasks, Dulanga Kaveesha Weerakoon Weerakoon Mudiyanselage

Dissertations and Theses Collection (Open Access)

The rapid pace of adoption of mixed-reality in tandem with advances in NLP and computer vision have opened up unprecedented opportunities for more naturalistic interaction interfaces which underpin Human-AI collaborative applications such as spatial computing and interactive conversational agents. One notable example is the emergence of interactive virtual assistants, which facilitate more natural communication of instructions and queries through modalities like voice and text. This trend is driving the development of innovative ubiquitous, mixed-reality computing applications. Such interactive, natural communication is also critical to support advances in human-robot interactive co-working, across a variety of industrial, commercial and home environments. Conventional …