Digital Commons Network^™

Benchmarking Gaslighting Negation Attacks Against Multimodal Large Language Models, Bin Zhu, Yinxuan Gui, Huiyan Qi, Jingjing Chen, Chong-Wah Ngo, Ee-Peng Lim

Research Collection School Of Computing and Information Systems

Multimodal Large Language Models (MLLMs) have exhibited remarkable advancements in integrating different modalities, excelling in complex understanding and generation tasks. Despite their success, MLLMs remain vulnerable to conversational adversarial inputs. In this paper, we systematically study gaslighting negation attacks—a phenomenon where models, despite initially providing correct answers, are persuaded by user-provided negations to reverse their outputs, often fabricating justifications. We conduct extensive evaluations of state-of-the-art MLLMs across diverse benchmarks and observe substantial performance drops when negation is introduced. Notably, we introduce the first benchmark GaslightingBench, specifically designed to evaluate the vulnerability of MLLMs to negation arguments. GaslightingBench consists of multiple-choice …

Not Too Early, Not All At Once: Design Tensions In Ai-Mediated Self-Disclosure In Online Dating, Pei-Hua Tsai, Tianyi Zhang, Emran Bin Elias Poh, Anthony Tang, Yung-Ju Chang

Research Collection School Of Computing and Information Systems

Online dating relies on self-disclosure, yet initial conversations are fragile: users must navigate uncertainty around timing, boundaries, and reciprocity with little shared context. While advances in AI raise the possibility of mediating disclosure, how such support might reshape the experience of early-stage relational disclosure remains underexplored. We conducted 29 semi-structured interviews to examine how daters envision AI-mediated self-disclosure in online dating. Our findings surface recurring design tensions rather than simple opportunities or risks. Participants welcomed guidance that could pace disclosure, support reflection, and reduce social awkwardness, but stressed preserving agency and authorship. They valued interpretive assistance for sense-making of ambiguous …

Hide-And-Sweep: Detecting Concealed Cameras Via Led Illumination Sweeps, Jonghyuk Yun, Jaeyoung Moon, Yunseo Park, Sean Rui Xiang Tan, Byunghyun Kim, Rajesh Krishna Balan, Jun Han

Research Collection School Of Computing and Information Systems

Hidden cameras have increasingly infiltrated hotel and Airbnb rooms, posing serious privacy risks. Detecting such cameras is challenging because they are visually inconspicuous and often embedded inside everyday objects. Even worse, existing handheld detectors are manual and also rely on single-angle illumination and hence suffer from high false-positive rates. We present SweepLED (pronounced "sweepled")1, a practical hidden camera detection system that operates on a commodity smartphone augmented with an unobtrusive LED-embedded case. SweepLED performs LED sweeping - a controlled sequence of multi-angle illumination - while the user simply holds the phone still by hand, enabling the camera to capture how …

Task Complexity Matters: An Empirical Study Of Reasoning In Llms For Sentiment Analysis, Donghao Huang, Zhaoxia Wang

Research Collection School Of Computing and Information Systems

Large language models (LLMs) with reasoning capabilities have fueled a compelling narrative that reasoning universally improves performance across language tasks. We test this claim through a comprehensive evaluation of 504 configurations across seven model families—including adaptive, conditional, and reinforcement learning-based reasoning architectures—on sentiment analysis datasets of varying granularity (binary, five-class, and 27-class emotion). Our findings reveal that reasoning effectiveness is strongly task-dependent, challenging prevailing assumptions: (1) Reasoning shows task-complexity dependence—binary classification degrades up to -19.9 F1% points (pp), while 27-class emotion recognition gains up to  +16.0 pp; (2) Distilled reasoning variants underperform base models by 3–18 pp on simpler tasks, …

How Do Machine Learning Models Change?, Joel Castaño, Rafael Cabañas, Antonio Salmerón, David Lo, Silverio Martínez-Fernández

Research Collection School Of Computing and Information Systems

The proliferation of Machine Learning (ML) models and their open source implementations has transformed AI research and applications. Platforms like Hugging Face (HF) enable this evolving ecosystem, yet a large-scale longitudinal study of how these models change is lacking. This study addresses this gap by analyzing over 680,000 commits from 100,000 models and 2,251 releases from 202 of these models on HF using repository mining and longitudinal methods. We apply an extended ML change taxonomy to classify commits and use Bayesian networks to model temporal patterns in commit and release activities. Our findings show that commit activities align with established …

Heat Transfer And Micropolar Characteristic Analysis For Fluid Flow Over A Contracting Permeable Infinite-Length Cylinder: A Fractional Derivative Study, Khalid Al Homran

Theses

This study presents a comprehensive analysis of heat transfer and micropolar fluid characteristics for flow over a contracting permeable infinite-length cylinder using fractional derivative modeling. The micropolar fluid model accounts for microstructure effects that are prevalent in many practical applications, such as polymeric fluids, colloidal suspensions, and biological fluids. The inclusion of fractional derivatives allows the analysis to incorporate memory effects, which provide a more accurate depiction of the physical phenomena compared to classical integer-order models. The governing partial differential equations, representing the momentum, micropolar momentum and energy transport processes, are transformed into a set of nonlinear ordinary differential equations …

Ai-Powered Student Support And Development Platform, Wadha Rashed Alyammahi

Theses

Over the past few years, academic stress and student mental well-being have become critical concerns in educational settings, significantly affecting academic performance and overall quality of life. Students increasingly rely on digital platforms for academic support; however, many existing solutions lack the ability to adapt to users’ emotional states or provide personalized, context-aware assistance. In response to this challenge, this study presents the design and development of an intelligent, stress-aware conversational support system for students based on advanced natural language processing techniques and large language models.

The main goal of this study is to explore the effectiveness of an LLM-based …

Draft Final Bpsou Unreclaimed Sites: Ur-46 Remedial Action Work Plan (Rawp), Pioneer Technical Services, Inc.

Silver Bow Creek/Butte Area Superfund Site

No abstract provided.

U.S. Men's Views About Women's And Men's Agency In The Abortion Decision-Making Process, Xiana Bueno, Lucretia Mena-Melendez, Brandon L. Crawford, Ronna C. Turner, Wen-Juo Lo, Kristen N. Jozkowski

Counseling, Leadership, and Research Methods Faculty Publications and Presentations

Research examining men's public opinion on women's and men's roles in abortion is limited but increasingly important as access to abortion has shifted dramatically in the U.S. In this study, we examined U.S. men's views of women's and men's agency in abortion decision-making. We explored 72 in-depth interviews on abortion attitudes conducted with men in 2021 and identified three main narratives. First, most men indicated that abortion is exclusively a women's decision. Several men noted wanting men's opinions to at least be heard in the decision-making process, while still maintaining that women should be the decision-makers. We identified subtle nuances, …

Conello Vendor Marketplace Capstone: A Review On The Capstone Process And Computer Science Degree, Levi Hauck

University Honors Theses

This Capstone Review Thesis discusses the current structure of the Computer Science Degree path and Computer Science Capstone at PSU. This thesis reviews the Conello Vendor Marketplace Capstone, a project designed to help address the issue of teambuilding in the workplace. It points out ways in which teamwork and organization is an underdeveloped skill in Computer Science. During the Capstone project, the project management method of Agile development was recommended and used. However, it became clear that the team, leader included, had a gap in knowledge and skills to be successful, as well as a lack of experience in team …

Shallow-Marine Crinoid Genera May Have Been More Resilient Across Permian Extinction Events In Terms Of Richness Than Deep-Marine Crinoid Genera, Amelia C. Kolstad

University Honors Theses

Throughout the Phanerozoic, one of the most successful phyla of marine invertebrates has been Echinodermata. Echinodermata was particularly successful throughout the Paleozoic, with class Crinoidea making up a majority of occurrences. The proportion of echinoderm occurrences that are crinoids declined massively at the end of the Permian and continued to decrease into modern day, never fully recovering. One of the factors thought to have been so important for crinoid success in the Paleozoic and their decreasing success throughout geologic time has been their propensity for success in shallow-marine environments, though crinoids also exist within deep-marine environments. This study seeks to …

Agile In The Age Of Ai: Considerations And Improvements For Streamlining Developer Workflows, Stephen Feng

University Honors Theses

This paper examines the integration of AI language models into Agile developer workflows during a six-month software development project. Using the development of SagacityWall, a mindfulness-based social media application built by a team of eight undergraduates, it identifies three key areas in Agile processes where AI provided meaningful leverage: translating business requirements into actionable developer work tickets, accelerating framework research and technology stack decisions, and reducing onboarding friction through AI-assisted code scaffolding and Behavior-Driven Development story formatting. The study finds that AI meaningfully boosted productivity across these stages – not by replacing developer judgment, but by reducing overhead at each …

Reexamining The Croton Aqueduct: A Ward-Level Microhistorical Analysis Of Infrastructure And Cholera Mortality In 1849 New York City, Alexandria R. Toothman

Dissertations, Theses, and Capstone Projects

This work reexamines the Croton Aqueduct’s role in public health by shifting analysis from the citywide level to the ward level, revealing that infrastructure expansion did not produce uniform benefits across New York City. When the Croton Aqueduct opened in 1842, it was celebrated as an achievement that would deliver pure water and eliminate disease. However, only seven years after its opening, in 1849, cholera returned with greater force in the Sixth Ward. This work shows that the uneven distribution of Croton infrastructure corresponded with uneven cholera mortality rates, challenging narratives that present Croton as a singular triumph of urban …

Hyper-Bishops, Hyper-Rooks, And Hyper-Queens: Percentage Of Safe Squares On Higher Dimensional Chess Boards, Caroline Cashman, Joseph Cooper, Raul Marquez, Steven J. Miller, Jenna Shuffelton

School of Mathematical & Statistical Sciences Faculty Publications

Chess has inspired an abundance of mathematical problems, especially in combinatorics and probability. One such problem, initially studied by Miller, Sheng, and Turek, considers the proportion of safe spaces when randomly placing n rooks on an 𝑛×𝑛 chess board. They show that as n approaches infinity, the proportion of safe spaces converges to 1/𝑒2. We first generalize their results to bishops and queens. This problem is significantly more interesting and difficult; while a rook attacks the same number of spaces regardless of its position, this is not so for bishops and queens. We prove that the proportion of safe spaces …

Mapping Public Engagement With Environmental Finance In The U.S: Spatiotemporal Insights From Google Trends On Public Engagement And Policy Equity In The U.S, Hilda Afeku-Amenyo, Charles Knoble, Pankaj Lal

Department of Earth and Environmental Studies Faculty Scholarship and Creative Works

This study analyzes U.S. public engagement with environmental finance from 2015–2023 using Google Trends data for four key terms: “green bonds,” “sustainable finance,” “climate finance,” and “ESG.” Our analysis reveals a notable rise in engagement with environmental finance topics, with Environmental, Social, and Governance (ESG) investing generating the highest levels of interest. Geographically, the District of Columbia exhibited disproportionate search activity relative to other regions. While public awareness of environmental finance has grown, we highlight disparities in knowledge and accessibility of information. These findings carry significant policy implications, particularly in ensuring equitable access to sustainable financial instruments. We underscore the …

Radiotheranostics For Gynecological Pathologies, Joni Sebastiano

Dissertations, Theses, and Capstone Projects

Molecular imaging, specifically positron emission tomography (PET), is vital for detecting disease and understanding its biological makeup. The exploitation of radiolabeled antibodies for PET imaging has proven to be indispensable to the field of molecular imaging over the last several decades. The development of radioimmunoconjugates for use in immunoPET has not only allowed for the sensitive, specific, and high-resolution detection of disease, but has also enabled a deeper understanding of the disease biology, ultimately guiding more personalized therapeutic strategies. Most typically, this technology is harnessed for the diagnosis and treatment of cancer, however more recently, the field has explored the …

Mediating Energy And Charge Transfer Dynamics Of Lead Halide Perovskite Nanocrystals Via Surface Chemical Approaches, Aaron S. Malinoski

Dissertations, Theses, and Capstone Projects

Energy and charge transfer are the fundamental mechanisms by which semiconductor nanocrystals interact with their external environment and underpin all light-conversion related applications. Lead halide perovskite nanocrystals (PNCs) are a relatively new class of semiconductor nanocrystals that possess a unique and dynamic surface chemical environment. Investigations into the influence of these surface chemical conditions on the electronic coupling between surface-bound molecular acceptors and the perovskite nanocrystals, and how the surface chemistry can be manipulated, in turn controlling the charge and energy transfer mechanisms, is the focus of this dissertation. Through ligand-shell engineering using commercially available, inexpensive small molecules, the surface …

Scaling Limits Of Critical Observables Through The High Dimensional Incipient Infinite Cluster, Pranav Chinmay

Dissertations, Theses, and Capstone Projects

We give a general construction of the incipient infinite cluster in high dimensional percolation, and use it as a decoupling tool to rigorize geometric heuristics for analyzing the asymptotics of observables at criticality. Examples include demonstrating the limiting distribution of the chemical distance and full-strength asymptotics for k-point functions, which constitute foundational inputs for scaling limit results associated to critical clusters.

Parkinson’S Disease Phenotype Stratification Using Multiple Correspondence Analysis, Kelly Astudillo

Dissertations, Theses, and Capstone Projects

Parkinson’s disease (PD) is the second most common neurodegenerative disorder, with over 12 million people projected to be affected by 2040 (Dorsey et al., 2018). Deep phenotyping and stratification can provide useful information regarding PD pathogenesis and can aid in the  development of disease modifying therapies that aim to delay the progression or prevent the onset of neurodegeneration (Blandini et al., 2019; Smith & Schapira, 2022). Utilizing multivariate methods such as multiple correspondence analysis (MCA) permits for the simultaneous analysis of distinct data modalities. To the best of our knowledge, MCA has not been previously used to explore phenotype patterns …

A Predictive Coding Account Of Spatial Working Memory Following Prophylactic Levetiracetam Administration Prior To Traumatic Brain Injury, Omeima Mutwali

Dissertations, Theses, and Capstone Projects

Traumatic brain injury (TBI) symptom prevention and remediation is an important area of research that would benefit vulnerable groups, including active-duty and veteran soldiers. These patients can sustain penetrative forces in fields of combat or in training, which result in focal lesions that trigger inflammatory and degenerative processes in the brain. Both primary and secondary injuries are associated with changes to cognition, behavior and affective state. This disease poses increased risk of epileptogenesis, as well. Given these outcomes, prior research has evaluated levetiracetam (LEV) as a prophylactic treatment for seizures, cognitive deficits and negative emotionality. LEV acts as a presynaptic …

Considering Numerical Relativity For Upgrades In Mcfacts Merger Dynamics, Shawn K. Ray

Dissertations, Theses, and Capstone Projects

Binary black hole mergers (BBH), some likely originating from active galactic nuclei (AGN), are observed by ground-based detectors such as LIGO-Virgo-KAGRA (LVK) and will be detected by future space-based detectors such as the Laser Interferometer Space Antenna (LISA) and the Laser Interferometer Lunar Antenna (LILA). The dynamics of AGN may be responsible for a significant fraction of BBH mergers soon to be observed by LISA and LILA, yet the dynamics of BBH mergers within AGN disks remains ambiguous despite their appearance in gravitational wave observations. Monte carlo For AGN Channel Testing and Simulation (McFACTS) is an open-source software used to …

Legal Ethics Of Ai Snake Oil: Navigating The Hype, Harm, And Hope Of Legal Ai, Drew Simshaw

Michigan Law Review

A review of AI Snake Oil.By Arvind Narayanan and Sayash Kapoor.

Integrating Multi-Scale And Multi-Filtration Topological Features For Medical Image Classification, Pengfei Gu, Huimin Li, Haoteng Tang, Dongkuan Xu, Erik Enriquez, Dongchul Kim, Bin Fu, Danny Z. Chen

Computer Science Faculty Publications

Modern deep neural networks have shown remarkable performance in medical image classification. However, such networks either emphasize pixel-intensity features instead of fundamental anatomical structures (e.g., those encoded by topological invariants), or they capture only simple topological features via single-parameter persistence. In this paper, we propose a new topology-guided classification framework that extracts multi-scale and multi-filtration persistent topological features and integrates them into vision classification backbones. For an input image, we first compute cubical persistence diagrams (PDs) across multiple image resolutions/scales. We then develop a "vineyard" algorithm that consolidates these PDs into a single, stable diagram capturing signatures at varying granularities, …

From 5g To 6g: A Survey On Security, Privacy, And Standardization Pathways, Mengmeng Yang, Youyang Qu, Thilina Ranbaduge, Chandra Thapa, Nazatul Haque Sultan, Ming Ding, Hajime Suzuki, Wei Ni, Sharif Abuadbba, David Smith, Paul Tyler, Josef Pieprzyk, Thierry Rakotoarivelo, Xinlong Guan, Sirine Mrabet

Research outputs 2022 to 2026

The vision for 6G aims to enhance network capabilities, supporting an intelligent digital ecosystem where artificial intelligence (AI) is a key. However, the expansion of 6G raises critical security and privacy concerns due to the increased integration of IoT devices, edge computing, and AI. This survey provides a comprehensive overview of 6G protocols with a focus on security and privacy, identifying risks that have not been experienced in preceding 5G systems, and presenting mitigation strategies. While many vulnerabilities from earlier generations persist, the introduction of AI/ML introduces novel risks like model inversion and malicious manipulation of AI. Vulnerabilities in emerging …

Ai In Healthcare: Regulatory Guidelines And Judge-Made Negligence Principles For Ai Implementers, Gary K. Y. Chan

Research Collection Yong Pung How School Of Law

The use of artificial intelligence (AI) in healthcare may, notwithstanding its potential benefits, result in harm to patients from allegedly negligent acts or omissions by hospitals and medical doctors. In such circumstances, how should the principles in the tort of negligence (duty of care, breach, causation, remoteness of damage, and defences) respond to AI innovations in healthcare? In particular, how may the standard of care expected of hospitals and medical doctors be informed by regulatory guidelines? We refer to case law precedents and regulatory guidelines on the roles and responsibilities of doctors and hospitals as AI implementers. Importantly, they prompt …

Municipality Of Bayamón V. Exxon Mobil Corp. (2025): Bringing Racketeering Charges For Climate Justice, Ilinca C. Johnson

Pace Environmental Law Review

Environmental justice communities should have their climate change damages heard in the courtroom. One means to increase environmental justice claims is through claims under the federal Racketeer Influence and Corrupt Organizations (RICO) Act. Using Municipality of Bayamón v. Exxon Mobil Corp. (2025) as a model, this Article suggests a novel RICO case strategy to pursue climate damages for marginalized communities. Fundamentally, environmental justice RICO claims involve careful case design based on the climate-related damages a marginalized community has faced because of the long-term deception by fossil fuel actors upon those communities. Pursuing such claims forwards the intersectional issues faced by …

Adaptive Multimodal Smart Home Control On A Raspberry Pi 5 Using Hand Gestures, Voice Cues, And User Feedback, Vaibhav Bora

Theses

A real time multimodal smart home control system deployed on a Raspberry Pi 5 is presented. The system combines hand gestures, short voice cues, and proximity aware interaction to execute household commands such as light brightness control, fan speed adjustment, and stop or kill switch actions. Lightweight gesture and keyword spotting voice classifiers were trained offline and exported to TensorFlow Lite for efficient on device inference. For more natural spoken phrases, the system additionally integrates a locally deployed pretrained offline ASR component rather than a speech recognizer trained from scratch. Using a USB camera and microphone, the system operates fully …

Robustness Of Ai-Driven Histopathology Under Real-World Adversarial Examples, Ruchik N. Yajnik

Theses

This robustness of histopathology classification models under adversarial and real-world perturbations resembling clinical artifacts is being investigated.

Using whole-slide images from the CAMELYON17 cohort, four representative architectures—ResNet-18, ResNet-50, HIPT-2MLP, and ViT-B/16 —are benchmarked across controlled pixel-level distortions and artifact-like transformations. Adversarial methods include iterative Fast Gradient Sign, Projected Gradient Descent, Salt-and-Pepper noise, and the Adversarial Watermark—Stain Shift (AWSS). Three defense strategies—Randomized Smoothing, Adversarial Training, and an Artifact Detector—are evaluated for their ability to preserve diagnostic accuracy and model reliability. Structured perturbations consistently degrade performance, with transformer-based models showing the greatest sensitivity. The benchmark developed here offers a reproducible framework for …

Sound Detection Of Memory Leaks In Llvm Ir Programs Using Accumulation Analysis, Robert Blacha

Theses

Resource leaks occur when a limited resource such as memory is allocated by a program and needlessly held past the point of use. Leaks can lead to a degradation of services which can be specifically triggered with malicious behavior, for example abusing a memory leak in a program to cause a server to slow down and crash for a denial-of-service attack.

Prior work has demonstrated that accumulation analysis provides a sound detection of resource leaks with a working implementation for programs written in Java. While useful, current implementations are limited to programs written in Java, which has a garbage collector, …

Perceptual And Geometric Advances In Crowd Simulation, Bilas Talukdar

Dissertations

Simulating realistic crowd motion remains a fundamental challenge in computer graphics and multi-agent systems, as it requires modeling both physically plausible interactions and perceptually natural behaviors. Existing crowd simulation methods typically employ simplified geometric abstractions, most commonly circular agent representations, and model navigation using either analytical interaction formulations (e.g., force, velocity, or constraint-based methods) or learned policies derived through reinforcement learning. Despite their effectiveness, these approaches often overlook detailed geometric structure and do not explicitly account for perceptual realism. This dissertation addresses these challenges by improving the realism of virtual crowd simulation through two key advancements: perceptual preference learning and …