Physical Sciences and Mathematics Commons^™

Self-Supervised Pseudo Multi-Class Pre-Training For Unsupervised Anomaly Detection And Segmentation In Medical Images, Yu Tian, Fengbei Liu, Guansong Pang, Yuanhong Chen, Yuyuan Liu, Johan W. Verjans, Rajvinder Singh, Gustavo Carneiro

Research Collection School Of Computing and Information Systems

Unsupervised anomaly detection (UAD) methods are trained with normal (or healthy) images only, but during testing, they are able to classify normal and abnormal (or disease) images. UAD is an important medical image analysis (MIA) method to be applied in disease screening problems because the training sets available for those problems usually contain only normal images. However, the exclusive reliance on normal images may result in the learning of ineffective low-dimensional image representations that are not sensitive enough to detect and segment unseen abnormal lesions of varying size, appearance, and shape. Pre-training UAD methods with self-supervised learning, based on computer …

Llm-Adapters: An Adapter Family For Parameter-Efficient Fine-Tuning Of Large Language Models, Zhiqiang Hu, Lei Wang, Yihuai Lan, Wanyu Xu, Ee-Peng Lim, Lidong Bing, Xing Xu, Soujanya Poria, Roy Ka-Wei Lee

Research Collection School Of Computing and Information Systems

The success of large language models (LLMs), like GPT-4 and ChatGPT, has led to the development of numerous cost-effective and accessible alternatives that are created by finetuning open-access LLMs with task-specific data (e.g., ChatDoctor) or instruction data (e.g., Alpaca). Among the various fine-tuning methods, adapter-based parameter-efficient fine-tuning (PEFT) is undoubtedly one of the most attractive topics, as it only requires fine-tuning a few external parameters instead of the entire LLMs while achieving comparable or even better performance. To enable further research on PEFT methods of LLMs, this paper presents LLMAdapters, an easy-to-use framework that integrates various adapters into LLMs and …

Graph Contrastive Learning With Stable And Scalable Spectral Encoding, Deyu Bo, Yuan Fang, Yang Liu, Chuan Shi

Research Collection School Of Computing and Information Systems

Graph contrastive learning (GCL) aims to learn representations by capturing the agreements between different graph views. Traditional GCL methods generate views in the spatial domain, but it has been recently discovered that the spectral domain also plays a vital role in complementing spatial views. However, existing spectral-based graph views either ignore the eigenvectors that encode valuable positional information, or suffer from high complexity when trying to address the instability of spectral features. To tackle these challenges, we first design an informative, stable, and scalable spectral encoder, termed EigenMLP, to learn effective representations from the spectral features. Theoretically, EigenMLP is invariant …

Memory Network-Based Interpreter Of User Preferences In Content-Aware Recommender Systems, Nhu Thuat Tran, Hady W. Lauw

Research Collection School Of Computing and Information Systems

This article introduces a novel architecture for two objectives recommendation and interpretability in a unified model. We leverage textual content as a source of interpretability in content-aware recommender systems. The goal is to characterize user preferences with a set of human-understandable attributes, each is described by a single word, enabling comprehension of user interests behind item adoptions. This is achieved via a dedicated architecture, which is interpretable by design, involving two components for recommendation and interpretation. In particular, we seek an interpreter, which accepts holistic user’s representation from a recommender to output a set of activated attributes describing user preferences. …

Robust Test Selection For Deep Neural Networks, Weifeng Sun, Meng Yan, Zhongxin Liu, David Lo

Research Collection School Of Computing and Information Systems

Deep Neural Networks (DNNs) have been widely used in various domains, such as computer vision and software engineering. Although many DNNs have been deployed to assist various tasks in the real world, similar to traditional software, they also suffer from defects that may lead to severe outcomes. DNN testing is one of the most widely used methods to ensure the quality of DNNs. Such method needs rich test inputs with oracle information (expected output) to reveal the incorrect behaviors of a DNN model. However, manually labeling all the collected test inputs is a labor-intensive task, which delays the quality assurance …

Molca: Molecular Graph-Language Modeling With Cross-Modal Projector And Uni-Modal Adapter, Zhiyuan Liu, Sihang Li, Yanchen Luo, Hao Fei, Yixin Cao, Kenji Kawaguchi, Xiang Wang, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Language Models (LMs) have demonstrated impressive molecule understanding ability on various 1D text-related tasks. However, they inherently lack 2D graph perception — a critical ability of human professionals in comprehending molecules’ topological structures. To bridge this gap, we propose MolCA: Molecular Graph-Language Modeling with Cross-Modal Projector and Uni-Modal Adapter. MolCA enables an LM (i.e., Galactica) to understand both text- and graph-based molecular contents via the cross-modal projector. Specifically, the cross-modal projector is implemented as a QFormer to connect a graph encoder’s representation space and an LM’s text space. Further, MolCA employs a uni-modal adapter (i.e., LoRA) for the LM’s efficient …

Efficient Meta Neural Heuristic For Multi-Objective Combinatorial Optimization, Jinbiao Chen, Zizhen Zhang, Te Ye, Zhiguang Cao, Siyuan Chen, Jiahai Wang

Research Collection School Of Computing and Information Systems

Recently, neural heuristics based on deep reinforcement learning have exhibited promise in solving multi-objective combinatorial optimization problems (MOCOPs). However, they are still struggling to achieve high learning efficiency and solution quality. To tackle this issue, we propose an efficient meta neural heuristic (EMNH), in which a meta model is first trained and then fine-tuned with a few steps to solve corresponding single-objective subproblems. Specifically, for the training process, a (partial) architecture-shared multi-task model is leveraged to achieve parallel learning for the meta model, so as to speed up the training; meanwhile, a scaled symmetric sampling method with respect to the …

Metabox: A Benchmark Platform For Meta-Black-Box Optimization With Reinforcement Learning, Zeyuan Ma, Hongshu Guo, Jiacheng Chen, Zhenrui Li, Guojun Peng, Yue-Jiao Gong, Yining Ma, Zhiguang Cao

Research Collection School Of Computing and Information Systems

Recently, Meta-Black-Box Optimization with Reinforcement Learning (MetaBBO-RL) has showcased the power of leveraging RL at the meta-level to mitigate manual fine-tuning of lower-level black-box optimizers. However, this field is hindered by the lack of a unified benchmark. To fill this gap, we introduce MetaBox, the first benchmark platform expressly tailored for developing and evaluating MetaBBO-RL methods. MetaBox offers a flexible algorithmic template that allows users to effortlessly implement their unique designs within the platform. Moreover, it provides a broad spectrum of over 300 problem instances, collected from synthetic to realistic scenarios, and an extensive library of 19 baseline methods, including …

The Persuasive Effect Of Ai-Synthesized Voices, Hannah H. Chang, Anirban Mukherjee

Research Collection Lee Kong Chian School Of Business

Artificial intelligence (AI) technology seeks to emulate humans. One aspect is AI-synthesized voices, used in voice assistants (such as Amazon Alexa, Apple Siri, and Google Assistant) to assistive technologies (such as voiceover narration in product videos). For example, there are currently more than 3.25 billion voice assistants; a number that is expected to touch about 8 billion by next year (i.e., 2023) (Statista 2022). With the extensive availability and enhanced accuracy of AI-synthesized voices, consumer research is starting to examine the impact of AI-synthesized voices on consumer information processing and decision making. The extant literature, however, is relatively limited because …

Predictive Model For Cfpb Consumer Complaints, Vyshnavi Nalluri

Electronic Theses, Projects, and Dissertations

Within the dynamic and highly competitive financial industry, the timely and efficient resolution of customer complaints stands as a central challenge, particularly in the intricate domain of mortgage services. The traditional processes for handling these complaints have long been recognized as laborious and resource-intensive, a situation that financial institutions, including the esteemed Wells Fargo, are keen to improve.

Currently, the industry largely relies on basic data analytics for identifying trends in customer complaints. However, this approach has its limitations, especially when dealing with complaints within the mortgage services domain. In response to this challenge, this research advocates the adoption of …

Accelerated Evolution Of Sars-Cov-2 In Free-Ranging White-Tailed Deer, Dillon S. Mcbride, Sofya K. Garushyants, John Franks, Andrew F. Magee, Steven H. Overend, Devra Huey, Amanda M. Williams, Seth A. Faith, Ahmed Kandeil, Sanja Trifkovic, Lance Miller, Trushar Jeevan, Anami Patel, Jacqueline M. Nolting, Michael J. Tonkovich, J. Tyler Genders, Andrew J. Montoney, Kevin Kasnyik, Timothy J. Linder, Sarah N. Bevins, Julianna B. Lenoch, Jeffrey C. Chandler, Thomas J. Deliberto, Eugene V. Koonin, Marc A. Suchard, Philippe Lemey, Richard J. Webby, Martha I. Nelson, Andrew S. Bowman

United States Department of Agriculture Wildlife Services: Staff Publications

The zoonotic origin of the COVID-19 pandemic virus highlights the need to fill the vast gaps in our knowledge of SARS-CoV-2 ecology and evolution in non-human hosts. Here, we detected that SARS-CoV-2 was introduced from humans into white-tailed deer more than 30 times in Ohio, USA during November 2021-March 2022. Subsequently, deer-to-deer transmission persisted for 2–8 months, disseminating across hundreds of kilometers. Newly developed Bayesian phylogenetic methods quantified how SARS-CoV-2 evolution is not only three-times faster in white-tailed deer compared to the rate observed in humans but also driven by different mutational biases and selection pressures. The long-term effect of …

Twitter Policing, Hemanth Kumar Medisetty

Electronic Theses, Projects, and Dissertations

Police departments are frequently utilizing social media platforms to actively interact with the public. Social media offers an opportunity to share information, facilitate communication, and foster stronger connections between police departments and the communities they serve. In this context sentiment analysis of social media data has become a tool, for identifying sentiments and tracking emerging trends.

This project utilizes sentiment analysis to examine the social media interactions with particular data obtained from the Twitter (X). Initially, the project gathers social media data, from twitter mentioned accounts on Twitter utilizing web scraping techniques. Afterwards, we perform a thorough sentiment analysis using …

Effect Of Self-Interaction Correction On Molecular Polarizabilities And Core Ionization Energies, Sharmin Akter

Open Access Theses & Dissertations

Density Functional Theory (DFT) is one of the most successful and popular computational Quantum Mechanical approaches to understanding materials. DFT allows the prediction of material properties from the electron density. Although in principle, density functional theory is exact, it, however, relies on approximate functional for exchange-correlation energy. Due to the approximate nature of the exchange-correlation functional, the self-Coulomb energy of the electrons is not exactly canceled out by the self-exchange, leading to the spurious self-interaction error (SIE). Due to this error, the potential shows incorrect behavior which leads to errors in calculated properties such as ionization energies, electron affinities, polarizabilities, …

Upcyclying Of Polyethylene Terephtalate By Addition Of Thermoplastic Elastomer, Diego Francisco Bermudez

Open Access Theses & Dissertations

Continual overconsumption of single-use plastics has generated challenges of solid waste management across the United States. Common plastic waste management solutions, such as landfill, have caused the migration of contaminants into the environment consequently affecting not only the health of wildlife, but also that of human beings. Alternative strategies for the handling of single-use plastic such as polyethylene terephthalate (PET), used in the food packaging industry, can ultimately help mitigate the noxious consequences of single-use plastics affecting entire ecosystems. This study demonstrates a potential avenue of materials upcycling by studying the effects of coupling PET with the thermoplastic elastomer styrene-ethylene-butylene-styrene …

Investigating Seismicity And Structure Of The Pecos, Texas Region Of The Delaware Basin Using A Temporary Nodal Network, Jenna Lynn Faith

Open Access Theses & Dissertations

With increasing earthquakes in the Delaware Basin since 2009, earthquake studies, including accurate hypocenters, are critically needed in the Delaware Basin to identify the structures producing earthquakes, and to determine if they are related to unconventional petroleum development and production. In 2018, with funding from the Texas Seismological Network (TexNet), the University of Texas at El Paso deployed and maintained a nodal network of 25 Magseis Fairfield Z-Land Generation 2 5-Hz seismic nodes in the Pecos, Texas region of the Delaware Basin, known as The Pecos Array. The network was deployed from November 2018 until the beginning of January 2020, …

Integrating Machine Learning Methods For Medical Diagnosis, Jazmin Quezada

Open Access Theses & Dissertations

Abstract:The rapid advancement of machine learning techniques has revolutionized the field of medical diagnosis by offering powerful tools to analyze complex data sets and make accurate predictions. In this proposed method, we present a novel approach that integrates machine learning and optimization models to enhance the accuracy of medical diagnoses. Our method focuses on fine-tuning and optimizing the parameters of machine learning algorithms commonly used in medical diagnosis, such as logistic regression, support vector machines, and neural networks. By employing optimization techniques, we systematically explore the parameter space of these algorithms to discover the most optimal configurations. Moreover, by representing …

A Mathematical App For The Conceptual Understanding Of Area And Perimeter, Jumela F. Sarmiento, Debbie Marie B. Verzosa, Maria Alva Q. Aberin, Ma. Louise Antonette De Las Penas, Agnes D. Garciano, Juan Carlo F. Mallari, Mark Anthony C. Tolentino

Mathematics Faculty Publications

This paper discusses an app that was developed to build a strong understanding of the concepts of area and perimeter in students. An important feature of the app is the three-component feature which highlights progressive learning: Explore, designed for the learning of the conceptual understanding of area and perimeter; Apply, where area and perimeter concepts are applied; and Create intended for constructing representations to develop higher order thinking skills. The pedagogical basis for the creation of the app, the game design elements employed in the app as well as the integration of the app in the classroom will be presented.

A Visualization App On Proving Geometric Concepts, Ma. Louise Antonette N. De Las Peñas, Debbie Marie B. Verzosa, Maria Alva Q. Aberin, Agnes D. Garciano, Jumela F. Sarmiento, Mark Anthony C. Tolentino, Juan Carlo F. Mallari

Mathematics Faculty Publications

This paper discusses the description, design and pedagogical basis of a mathematical app called Two Column Proof which provides students a framework for writing proofs of geometric statements. The app focuses on proving concepts on: properties of parallelograms, conditions that determine when a quadrilateral is a parallelogram, properties on trapezoids and kites for high school mathematics. It employs visual representations for students to understand the statement or reason of each given line in the proof, strengthening their logical and mathematical knowledge needed in the proof construction.

Design Of A Mobile App To Promote Understanding And Fluency In Finding The Equation Of A Line, Agnes D. Garciano, Maria Alva Q. Aberin, Ma Louise Antonette N. Delaspeñas, Juan Carlo F. Mallari, Jumela F. Sarmiento, Mark Anthony C. Tolentino, Debbie Marie B. Verzosa

Mathematics Faculty Publications

This paper focuses on the design of a mobile app called Pick or Fish that fosters comprehension and mastery of the concepts of slopes, y-intercepts and equations of lines. The app's pedagogical value lies in its potential to help students understand and become proficient in these concepts. The app is suitable for use on low-cost mobile devices. It functions within an engaging game-like setting featuring visual elements that enable students to see the effect of parameter changes on the direction of a line. The beginner and advanced levels of the app have scaffolding features that gradually introduce the students to …

An Optimized Search For Dark Matter In The Galactic Halo With Hawc, A. Albert, R. Alfaro, C. Alvarez, J. C. Arteaga-Velázquez, D. Avila Rojas, H. A. Ayala Solares, E. Belmont-Moreno, K. S. Caballero-Mora, T. Capistrán, A. Carramiñana, S. Casanova, O. Chaparro-Amaro, U. Cotti, J. Cotzomi, E. De La Fuente, R. Diaz Hernandez, B. L. Dingus, M. A. Duvernois, M. Durocher, J. C. Diaz-Velez, C. Espinoza, M. M. Gonzalez, J. A. Goodman, J. P. Harding, D. Huang, F. Hueyotl-Zahuantitla, A. Iriarte, V. Joshi, G. J. Kunde, J. Lee, H. Leon Vargas, J. T. Linneman, A. L. Longinotti, G. Luis-Raya, J. Lundeen, K. Malone, O. Martinez, J. Martinez-Castro, J. A. Matthews, E. Moreno, M. Mostafa, A. Nayerhoda, L. Nellen, A. Peisker, E. G. Perez-Perez, C. D. Rho, D. Rosa-Gonzalez, H. Salazar, D. Salazar-Gallegos, A. Sandoval, J. Serna-Franco, R. W. Springer, O. Tibolla, K. Tollefson, I. Torres, R. Torres-Escobedo, R. Turner, F. Urena-Mena, L. Villasenor, Xiaojie Wang, H. Zhou, C. De Leon

Physics Faculty Research & Creative Works

The Galactic Halo is a key target for indirect dark matter detection. The High-Altitude Water Cherenkov (HAWC) observatory is a high-energy (∼300 GeV to >100 TeV) gamma-ray detector located in central Mexico. HAWC operates via the water Cherenkov technique and has both a wide field of view of ∼ 2 sr and a >95% duty cycle, making it ideal for analyses of highly extended sources. We made use of these properties of HAWC, and a new background-estimation technique optimized for extended sources to probe a large region of the Galactic Halo for dark matter signals. With this approach, we set …

Exploring Non-Linear Dynamical Structure For Knee Kinematics Using Machine Learning, Liora Mayats-Alpay, Rahul Soangra

Physical Therapy Faculty Articles and Research

Human movement involves complex coordination between multiple limbs during execution. Human gait is cyclic, and the knee's movement inherently follows nonlinear dynamic behavior that linear models cannot adequately capture. In this study, advanced Machine Learning (ML) techniques were employed to combine the Sparse Identification of Nonlinear Dynamics (SINDy) algorithm using Python to reveal governing equations of knee movement during walking. We gathered a single subject's knee motion data using infrared markers during normal walking. We utilized the PySINDy library to determine the governing equations and calculated the coefficient of dynamical systems associated with knee kinematics. Our results emphasize governing equations …

Stochastic Optimal Control Of Conditional Mckean-Vlasov Equations With Jump And Markovian Switching, Charles Samuel Conly Sharp

Theses and Dissertations

This thesis obtains a number of results in stochastic optimal control for conditional McKean-Vlasov equations with jump and Markovian switching. First, we prove the uniqueness of the solutions and derive a relevant version of Itô's formula. We provide the dynamic programming principle and prove the associated verification theorem. A stochastic maximum principle is established. Further, we derive the relationship between dynamic programming and the stochastic maximum principle. Additionally, we utilize our stochastic maximum principle result for a mean-variance portfolio selection problem.

Analyzing The Effects Of Ultrafast Laser Processing On Mechanical Properties Of 3d-Printed Pla Parts, Darshan Pramodbhai Yadav

Theses and Dissertations

Recent advances in additive manufacturing technologies have already led to wide-scale adoption of 3D-printed parts in various industries. The expansion in choice of materials that can be processed, particularly using Fused Deposition Modeling (FDM), and the steady advancements in dimensional accuracy control have extended the range of applications far beyond rapid prototyping. However, additive manufacturing still has considerable limitations compared to traditional and subtractive manufacturing processes. This work addresses limitations associated with the as-deposited surface roughness of 3D-printed parts. The effects of roughness-induced stress concentrations were studied on ultimate tensile strength and fatigue life. The samples were manufactured using a …

A Multi-Channel Search For Signatures Of Ev2-Scale Sterile Neutrinos In Icecube, Benjamin Randolph Smithers

Physics Dissertations - Archive

The IceCube Neutrino Observatory is a cubic-kilometer-scale neutrino observatory built using a gigaton of instrumented ice near the Amundsen-Scott South Pole Station in Antarctica; it measures the Cherenkov light emitted by the daughter particles of deep inelastic scattering interactions occurring inside and around the detector from neutrinos of both astrophysical and atmospheric origin. For eV2-scale sterile neutrino models, matter-effects can yield resonant muon antineutrino to sterile oscillations for TeV-scale up-going neutrinos. For certain combinations of mixing angles, this is expected to yield resonant muon antineutrino to tau antineutrino oscillations at the same energies and angles as the muon neutrino disappearance. …

Measurement, Modeling, And Analysis Of The Energy Spectra Of Electrons And Annihilation Gammas Resulting From Low Energy Positron Irradiation Of Clean And Adsorbate Covered Surfaces, Sima Lotfimarangloo

Physics Dissertations - Archive

This dissertation presents measurements and theoretical modeling demonstrating the capability of Doppler Broadened annihilation gamma Spectroscopy (DBS) to provide element-specific information from the topmost atomic layer of surfaces that are either clean or covered with adsorbates or thin films. Our measurements show that the energy spectra of Doppler-shifted annihilation gamma photons emitted following the annihilation of positrons from the topmost atomic layers of clean gold (Au) and copper (Cu) differ significantly. With the aid of the positron annihilation-induced Auger electron spectroscopy (PAES) performed simultaneously with DBS, we show that measurable differences between the Doppler broadened gamma spectra from Au and …

Exposure To Climate Change Information Predicts Public Support For Solar Geoengineering In Singapore And The United States, Sonny Rosenthal, Peter J. Irvine, Christopher L. Cummings, Shirley S. Ho

Research Collection College of Integrative Studies

Solar geoengineering is a controversial climate policy measure that could lower global temperature by increasing the amount of light reflected by the Earth. As scientists and policymakers increasingly consider this idea, an understanding of the level and drivers of public support for its research and potential deployment will be key. This study focuses on the role of climate change information in public support for research and deployment of stratospheric aerosol injection (SAI) in Singapore (n = 503) and the United States (n = 505). Findings were consistent with the idea that exposure to information underlies support for research and deployment. …

Draft Final Butte Treatment Lagoons (Btl) Groundwater Treatment System Routine Operations, Maintenance, And Monitoring (Om&M) Plan, Pioneer Technical Services, Inc.

Silver Bow Creek/Butte Area Superfund Site

No abstract provided.

Thin Film Piezoelectric Energy Harvesting Nanostructured Materials: Tailoring Size, Porosity, Morphology Of Zinc Stannate Perovskites (Abx3), Christopher Munoz

Theses and Dissertations

Three dimensional (3D) piezoelectric zinc stannate (ZnSnO3) nanoweb arrays are synthesized using a combination of treatment methods deposited in PDMS thin films for electrochemical analysis of its piezoelectric response. Advantages of hydrothermal, molten salt, and solvothermal synthesis methods were leveraged to facilitate several chemical and surface engineering techniques. The combination of these treatments reduce the size of zinc stannate to approximately ~40nm-80nm weblike networks, much smaller than previously reported ZnSnO3 sub-micro cubes. Scanning electron microscopy (SEM) and X-Ray Diffraction (XRD) analysis reveal a mesoporous protonated tristannate (H2Sn3O7) nanoweb template with connecting wirelike strands having diameters ranging from 12-27nm across and …

Electrolysis Degradation Of Cyanuric Acid In Wastewater At Cu/Go Cathodes, Chioma Chinwe Nwakanobi

Theses and Dissertations

This work evaluates the development of electrochemical processes for the oxidative degradation of persistent organic chemicals in waste waters, specifically the oxidation of cyanuric acid in the presence of NaCl at Cu/GO cathodes and Ti anodes. The experiment used a graphene oxide doped copper mesh cathode to examine different concentrations of cyanuric acid (30 mg/L, 40 mg/L, and 50 mg/L) and various concentrations of NaCl electrolyte (0.007mol/L, 0.03mol/L, 0.07mol/L, 0.1mol/L, and 0.2mol/L). In addition, the effect of pH (2, 4, 6, and 8), and applied current (0.200A, 0.150A and 0.100A) were studied. The Cu/GO electrode played a central role in …

Enhancing Time Series Hashing Performance Via Deep Orthogonal Hashing, Mahmudul Hasan Robin

Theses and Dissertations

Deep hashing has been widely used for efficient retrieval and classification of high-dimensional data like images and text. However, its application to time series data is still challenging due to the data’s temporal nature. To tackle this issue, a new deep hashing method has been proposed that generates efficient hash codes and enhances the time series hashing performance using a ResNet model with Orthohash (Cosine Similarity Loss). The proposed method uses one loss architecture while using ResNet model for efficient hashing. It uses the Character Trajectories dataset to extract discriminative features from the time series data. These features are then …