Computer Sciences Commons^™

Context Matters: Auditing Gender Bias In T2i Generation Through Risk-Tiered Use-Case Profiles, Jose Luis Luna Campoverde, Yankun Wu, Xiaofei Xie, Noa Garcia

Research Collection School Of Computing and Information Systems

Text-to-image (T2I) generative models are increasingly used to produce content for education, media, and public-facing communication, and are starting to be integrated into higher-impact pipelines. Since generated images tend to reinforce stereotypes, producing representational erasure via “default” depictions and shaping perceptions of who belongs in certain roles, a growing body of work has proposed metrics to quantify gender bias in T2I outputs. Yet existing evaluations remain fragmented. Metrics are often reported without a shared view of what they measure, what assumptions they entail, or how their results should be interpreted under different deployment contexts. This limits the usefulness of gender …

Rc-Nf: Robot-Conditioned Normalizing Flow For Real-Time Anomaly Detection In Robotic Manipulation, Shijie Zhou, Bin Zhu, Jiarui Yang, Xiangyu Zhao, Jingjing Chen, Yu-Gang Jiang

Research Collection School Of Computing and Information Systems

Recent advances in Vision-Language-Action (VLA) models have enabled robots to execute increasingly complex tasks. However, VLA models trained through imitation learning struggle to operate reliably in dynamic environments and often fail under Out-of-Distribution (OOD) conditions. To address this issue, we propose Robot-Conditioned Normalizing Flow(RC-NF), a real-time monitoring model for robotic anomaly detection and intervention that ensures the robot's state and the object's motion trajectory align with the task. RC-NF decouples the processing of task-aware robot and object states within the normalizing flow. It requires only positive samples for unsupervised training and calculates accurate robotic anomaly scores during inference through the …

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 …

Crab: A Novel Clustering Score Using Clustering With Rivals And Buddies For Unsupervised Learning, Allen Choi

Master's Theses

Unsupervised clustering algorithms today are used across a wide variety of fields such as biology, engineering, and industry in order to classify observations into groups where labels are not provided. This can provide important latent information regarding the observations within groups, as well as insight regarding the groups themselves. In order to judge the optimal number of clusters for an unsupervised clustering algorithm, many methods exist such as the Elbow Method and Silhouette Score; however, these methods come with drawbacks and are not necessarily flexible across many unsupervised methods. We present a novel clustering score framework relying on a resampling-based …

Integer Quantization And Embedded Deployment Of Cnn Star Trackers For Cubesats, Meora R. Giusiano

Master's Theses

A star tracker determines spacecraft orientation by photographing the star field, detecting stars in the image, matching them against a catalog, and computing the rotation between observed and cataloged directions. Convolutional neural networks (CNNs) have been proposed as replacements for the detection and centroiding stage, offering improved sub-pixel accuracy and recovering faint stars that classical thresholds lose to stray light and sensor noise. The improvement comes at higher computational cost; the PolySat systemboard targeted in this work lacks the floating-point hardware these networks assume.

This thesis closes the gap between floating-point desktop evaluation and embedded integer deployment. Nine encoder-decoder CNN …

High-Resolution Queries, Low-Resolution Context: Scaling Vision Transformers With Asymmetric Spatial Reduction, James M. Dwyer

Master's Theses

Vision Transformers (ViTs) have demonstrated great performance on image classifica tion benchmarks, however, the quadratic complexity of the self-attention mechanism with respect to sequence length limits their scalability to higher resolution inputs. The attention score matrix grows as O(N2) in both compute and memory, where N is the number of patch tokens, making ViTs computationally expensive and memory intensive for applications that require real-time inference or operate under resource constraints.

This thesis investigates whether the key and value sequences of the self-attention mechanism can be compressed using the local spatial structure of the image — while keeping queries at full …

Scorespeak: An Agentic System For Natural Language Control Of Musical Scores, Nathan S. Lim

Master's Theses

With the recent popularization of large language models (LLMs), natural language has become one of the most accessible and powerful ways for people to interact with creative tools. Although they have become common in mainstream domains like image and audio editing, there is currently no robust AI-based system that can reliably turn free-form language into edits for symbolic musical scores. This gap represents a missed opportunity to improve human workflows for creating and editing sheet music, but it is also a fundamental limitation for other agentic music systems; without a robust mechanism for translating free-form language into structured scores, AI …

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 …

The Stars Align: Modeling User Rating Calibration With Sparse Semantic Review Features, Rodrigo Alves, Antoine Ledent

Research Collection School Of Computing and Information Systems

User ratings are often treated as comparable across users, although identical scores may reflect different experiences. We study whether ratings can be viewed as user-specific discretizations of a shared semantic continuum derived from review text. Our method maps reviews into sparse semantic features with a sparse autoencoder and learns user-specific filters for each rating level. On Amazon Electronics, the learned embeddings align along a shared low-dimensional rating axis. Users differ mainly in how they anchor and partition this continuum, while preserving its overall ordinal structure. These findings support a semantic view of calibration beyond scalar bias correction.

Sok: Understanding Zkvm: From Research To Practice, Guomin Yang, Yunbo Yang, Yuejia Cheng, Haibo Tang, Bingsheng Zhang, Kui Ren

Research Collection School Of Computing and Information Systems

Zero-knowledge virtual machine (zkVM) is a powerful infrastructure for proving the correctness of a program execution with a succinct proof, attracting significant interest from researchers, developers, and users. It has been widely used in applications such as blockchain rollups, privacy-preserving machine learning, and off-chain computation. As the field grows, a wide range of zkVMs have been proposed. However, they adopt different choices in instruction formats, trace layouts, and proving backends, which results in a highly heterogeneous design landscape and makes it difficult to understand the relations among these systems.To bridge this gap, we provide a comprehensive study of zkVMs that …

Ppg-Sport: A Dataset For Reliable Heart Rate Monitoring From Wrist Ppg Under Dynamic Sports Conditions, Changshuo Hu, Hung Manh Pham, Yiming Zhang, Guanru Yan, Xiao Ma, Yuezhong Wu, Thivya Kandappu, Archan Misra, Dong Ma

Research Collection School Of Computing and Information Systems

Photoplethysmography (PPG) has become a cornerstone of physiological sensing in wearable devices, enabling non-invasive monitoring of heart rate and related biomarkers. However, its reliability deteriorates sharply under dynamic, high-intensity, or non-periodic motions such as those in sports, where existing datasets fail to capture realistic wrist dynamics. To address this gap, we introduce PPG-Sport, the first large-scale dataset designed for heart rate monitoring from wrist-worn PPG under real sports conditions. The PPG-Sport dataset includes synchronized PPG, inertial measurement unit (IMU), and electrocardiography (ECG) recordings from both wrists of 30 participants across six representative activities: stationary, walking, running, badminton, table tennis, and …

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, …

“Alexa, Do Not Say That In Front Of My Boss!” A Cross-Cultural Comparison Of User And Ai Preferences For Privacy-Aware Smart Speaker Interactions Across Contexts, Lynne Warin, Emily Aurelia, Anthony Tang, Emily Aurelia, Delphine Reinhardt

Research Collection School Of Computing and Information Systems

Due to their limited ability to reason about the social context in which they are used, smart speakers pose significant privacy risks by responding in ways that may violate people's implicit social boundaries. We conducted a cross-cultural vignette study (N = 944) in Germany and Singapore to investigate how situational factors—specifically social context (bystander relationships and closeness), physical context (location), and interaction context (topic and deceptive intent)—regulate user preferences for smart speaker responses. Our results demonstrate that these factors are superior predictors of response preferences than dispositional user traits (i.e., intrinsic personal traits). We identify two distinct social dynamics: a …

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 …

Principles Of Privacy And Security In Artificial Intelligence And Applications, Khang Tran

Dissertations

Modern artificial intelligence (AI) systems have transformed critical domains such as healthcare, software engineering, finance, and the legal system. Despite their broad impact, concerns about trustworthiness, especially regarding privacy and security, remain major obstacles to wider adoption. Addressing these concerns requires both a systematic understanding of the privacy and security risks inherent in AI systems and the development of principled foundations for trustworthy AI that safeguard client privacy and security. This goal is particularly challenging because of the complexity of modern large-scale AI systems, the trade-offs between privacy and model utility, and the need to simultaneously ensure other important properties …

Differential-Geometric Methods For Neural Signed Distance Fields: Parameterized Surface Extraction And Curvature Regularization For Cad Models, Haotian Yin

Dissertations

Neural signed distance fields have emerged as a powerful framework for representing three-dimensional geometry through continuous and differentiable neural functions. Their flexibility, resolution independence, and compatibility with gradient-based optimization make them especially attractive for surface reconstruction and geometric learning. However, despite these advantages, two fundamental challenges remain for engineering-grade applications. First, higher-order geometric properties such as curvature are difficult to model reliably during training and often require computationally expensive second-order differentiation. Second, while neural signed distance fields provide implicit surface representations, they do not directly yield a globally consistent forward map or parameterization for downstream geometric processing.

This dissertation addresses …

Holistic Dram Enhancements: From Intrinsic In-Memory Operations To Robust Security Mechanisms, Ranyang Zhou

Dissertations

Dynamic Random-Access Memory (DRAM) is both the performance bottleneck and a critical security boundary of modern computing systems. Its physical properties make it an attractive substrate for near-data computation—yet those same properties expose it to disturbance-based hardware attacks. This dissertation argues that these two dimensions are not independent: the architectural choices that make DRAM efficient also reshape its threat landscape. Addressing both requires a unified approach to memory architecture and security co-design.

The first part of this dissertation attacks the memory wall through four processing-in-DRAM (PIM) frameworks. ReD-LUT and LT-PIM unify lookup-table arithmetic with charge-sharing logic, achieving up to 37.8x …

Disentangling Non-Thermal Electron Injection And Decay In Solar Flares Using Multi-Wavelength Radio Observations, Brian Eugene O’Donnell

Dissertations

The broadband microwave imaging spectroscopy capability provided by the Expanded Owens Valley Solar Array (EOVSA) allows new diagnostics of high-energy processes in solar flares, providing spatially and temporally resolved spectra rich in information about the acceleration and transport of energetic electrons.

In this work, injections and transport of energy and particles into the solar corona during flares are studied. This is accomplished through the development and use of the PIP_Decomp Fitter, an automated fitting tool made by the author to fit injection and precipitation/decay parameters using the spatially resolved radio spectra obtained by EOVSA. These tools are used to study …

A Generative Ai-Driven Computational Framework For Industry-Scale Discovery Of Novel Battery Materials, Joy Datta

Dissertations

The growing demand for sustainable, high-energy-density electrochemical storage has motivated the exploration of multivalent-ion batteries based on earth-abundant elements such as aluminum, calcium, magnesium, and zinc. While multivalent charge carriers offer higher theoretical energy density than lithium, their practical deployment is hindered by sluggish ion transport, strong ion-host interactions, and structural degradation of electrode materials. Identifying host materials that can reversibly accommodate multivalent ions while maintaining structural integrity remains a fundamental challenge. The dissertation develops a scalable, end-to-end computational framework that integrates density functional theory (DFT), machine learning (ML), and generative artificial intelligence (GenAI) to accelerate the discovery of next-generation …

Toward Learning-Based Reconstruction And Part Decomposition Of Man-Made 3d Geometry: Neural Implicit Representations And Scalable Supervision, Shen Fan

Dissertations

Digital three-dimensional (3D) models are central to engineering design, analysis, and manufacturing, but learning pipelines for man-made geometry often operate on sampled carriers that do not preserve all of the structure present in exact CAD representations. This dissertation studies learning-based reconstruction and part decomposition for structured man-made 3D geometry, from general object benchmarks to CAD-derived datasets, with a focus on neural implicit representations trained from signed-distance samples, point clouds, and tessellated meshes. The goal is to make these models more accurate, more part-aware, and more consistently supervised.

First, signed distance function (SDF) reconstruction with implicit neural representations is improved through …

Anonymity And Accountability In Secure Messaging, Erin Kenney

Dissertations

Encypted messaging has become more and more prevalent as time moves on, and its benefits in assuring privacy cannot be overstated, but it also brings along with it concerns on how to moderate platforms where all messages are hidden. Message Franking, followed by Traceback systems, addressed these concerns by allowing the sender of a message to be proven when reported, even for forwarded messages in the case of Traceback, however these systems damage the privacy guarantees that originally motivated encrypted messaging to begin with.

In practice, even without those concerns encrypted messaging alone is not enough to prevent the most …

Parameter Density Estimation For Cardiac Electrophysiology Models Using Data Consistent Deep Learning, Michael Luo

Dissertations

Mathematical models of biological rhythms and excitable systems can provide insights into mechanisms underlying cardiac electrical dynamics. However, estimating the parameters of these models from experimental observations is often difficult due to noise, heterogeneity, and unobserved variables. For example, in an electrocardiogram (ECG) recording, information about the electrical properties of different regions of the heart is compressed into a single voltage trace. Additionally, variability within these signals may contain important information about population heterogeneity, regional differences in electrophysiology, and time-dependent modulation.

This dissertation develops, explores, and evaluates methods that perform feature-based distributional inference for complex nonlinear dynamical systems. The objective …

Edge Co-Occurrence Regularization For Node Classification, Kadir Altunel

Theses

We propose a simple yet effective regularization technique for node classification on graphs that leverages edge-based label co-occurrence patterns. We first train an MLP on node features to produce class probability distributions, then compute a fixed penalty matrix from edge-based co-occurrence statistics of these predictions. This penalty matrix, which captures unlikely class combinations on connected nodes, is then used to regularize GNN training without further updates. We evaluate this approach across multiple homophilic datasets (Cora, CiteSeer, PubMed, ogbn-arxiv) and heterophilic benchmarks (Chameleon, Squirrel, Actor, Roman-Empire) using three GNN architectures: GCN, GraphSAGE, and H2GCN. Results show consistent improvements on homophilic graphs, …

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, …

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 …

Pypitfall: Dependency Chaos And Supply Chain Vulnerabilities In Python, Jacob Mahon

Theses

Python software development heavily relies on third-party packages. Direct and transitive dependencies create a labyrinth of software supply chains. While it is convenient to reuse code, vulnerabilities within these dependency chains can propagate through dependencies, potentially affecting downstream packages and applications. PyPI, the official Python package repository, hosts many packages and lacks a comprehensive analysis of the prevalence of vulnerable dependencies. PyPitfall, a quantitative analysis of vulnerable dependencies across the PyPI ecosystem, is introduced. The dependency metadata of 378,573 PyPI packages is analyzed. 4,655 packages that explicitly require a known vulnerable package version and 141,044 packages that permit a vulnerable …

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 …

Streamlined Biomedical Image Processing Pipelines, Jiehyun Kim

Graduate Doctoral Dissertations

This dissertation focuses on advancing carotid artery analysis through a series of visualizations and deep learning tools for calcified plaque assessment and related biomedical imaging tasks. Accurate plaque evaluation is essential, but current workflows depend on slow, clinician-dependent manual review. To address these limitations, this work introduces the CACTAS framework, a set of tools and methods that enable fast and reliable plaque segmentation for clinicians.

The first study, the CACTAS-Tool, provides a web-based labeling tool that enables clinicians to label plaque directly in three dimensions through a streamlined one-click interface. This tool significantly reduces the effort required to generate high-quality …

Static Data-Race Detection For Gpu Programs: Behavioral Types With Partial Completeness Guarantees, Zhen Rong Liew

Graduate Doctoral Dissertations

GPUs are essential to modern computing but notoriously difficult to program correctly. Static analysis tools can verify data-race freedom, but their over-approximations produce spurious reports of data races that do not occur, limiting their practical usefulness. This dissertation establishes when Memory Access Protocols (MAPs), a compositional abstraction modeling memory access behavior between synchronization barriers, can be simultaneously sound and complete. We first prove that MAP-based analysis is sound and complete for well-typed Jaminan programs---those without data-dependent array indexing. This result establishes the theoretical boundary: completeness is achievable when data-dependent control flow and indexing are absent. Jaminan extends MAPs with symbolic …

Mri-Based Deep Learning Radiomics Model For Automated Classification Of Disc Degeneration In The Lumbar Spine, Shiv Patil, Om Gandhi, Mert Karabacak, Matthew Carr, Konstantinos Margetis

Student Papers, Posters & Projects

Disc degeneration in the lumbar spine is a major cause of low back pain (LBP). The accurate grading of disc degeneration on magnetic resonance imaging (MRI) is critical for clinical management and patient selection for spine surgery. This study aims to develop and evaluate machine learning (ML) models that combine features from deep learning (DL) and radiomics for the automated prediction of Pfirrmann grade (PG), a measure of disc degeneration, using multi-parametric lumbar spine MRI. Sagittal T1, T2, and T2 SPACE MRIs of 218 patients with LBP were acquired from the SPIDER dataset. For each intervertebral disc and available sequence, …