Digital Commons Network^™

Learning By Writing: Exploring Authentic Legal Learning Through Case Summaries, Ee-Ing Ong, Wei Yang Quek, Duan Ning, Magdeleine Lew

Research Collection Yong Pung How School Of Law

We use authentic learning as a pedagogical framework in a collaboration between our law school and the national Supreme Court of a Southeast Asian country, which facilitates law students’ development of their legal analytical and writing skills, and helps them better bridge the gap between existing legal curricula and the needs of legal practice. Akin to a writing apprenticeship, students write summaries on selected Supreme Court judgments, with their output reviewed by faculty as well as judicial law clerks from the court. The results are published on the court’s website and circulated to other stakeholders. In the post-exercise survey, participating …

Robust Generator Maintenance Schedule For Frequency-Secure Power Systems, Yang Yang, Qiuzhuang Sun, Jimmy Chih-Hsien Peng, Loon Ching Tang, Zhisheng Ye

Research Collection College of Integrative Studies

Problem definition: Normal operations of a power system require that alternating current frequency be maintained at a nominal value, for example, 50 Hz, whereas severe deviation from this value due to power deficiencies can cause cascading generator trips. Maintaining the frequency requires adequate inertia and frequency regulation reserve, which are primarily provided by online generators. In daily operations, generators due for preventive maintenance must be taken offline, and thus an improper maintenance schedule could jeopardize frequency security, as exemplified by the recent Texas power blackout. However, this natural nexus between frequency security and maintenance has been over-looked largely in the …

Online Planning Of Power Flows For Power Systems Against Bushfires Using Spatial Context, Jianyu Xu, Qiuzhuang Sun, Yang Yang, Huadong Mo, Daoyi Dong

Research Collection College of Integrative Studies

A power station or transmission line can be affected due to bushfires, increasing operation costs. We study a fundamental but challenging problem of planning the optimal power flow (OPF) for power systems under bushfires. We develop a model to capture the stochastic nature of bushfire spread based on Moore’s neighborhood model and propose an online optimization modeling framework to sequentially plan power flows in the electricity network. Our framework assumes that bushfire spread is non-stationary over time and that the spread and containment probabilities are unknown. To address these challenges, we develop a contextual online learning algorithm that treats the …

Thiol-Based Neuroprotective Copolymers Acutely Restore Redox Metabolism And Mediate Vasogenic Edema In A Mouse Model Of Traumatic Brain Injury, Evan T. Curtis, Brandon Z. Mcdonald, Aria W. Tarudji, Aaron M. Priester, Anthony J. Convertine, Forrest M. Kievit

Materials Science and Engineering Faculty Research & Creative Works

Effective pharmaceutical interventions for treating the secondary damage associated with traumatic brain injury (TBI) are limited due to poor delivery into the brain, insufficient target engagement, and an incomplete understanding of the pathophysiological changes that occur post-impact. Thus, nanoparticles (NP), which have an enhanced permeation and retention-like effect within the perturbed blood-brain barrier, have grown as a potential candidate for treating TBI. We have investigated the antioxidant capacity of thiol-based NP, termed neuroprotective copolymers (NPC3), and their ability to neutralize reactive oxygen species (ROS) and lipid peroxidation products (LPOx). Here, we assessed the efficacy of NPC3 for alleviating the secondary …

The Narrative Dynamics Of Special Books By Special Kids Youtube Videos And Comments: A Study Of Disability Rhetoric And Public Narrative Engagement, Jessica Louise Allee

Graduate Theses and Dissertations

Digital media has expanded disability advocacy and community building by enabling accessible, visual storytelling and connection. Special Books by Special Kids (SBSK), a YouTube channel using interview-based disability storytelling, foregrounds people’s experiences with disabilities in the videos, while inviting viewers to interpret, identify with, and respond in the comments. As a participatory storytelling space, SBSK offers a site for examining how disability meanings are constructed, negotiated, and transformed through narratives, multimodal practices, and audience uptake, while revealing how dominant disability rhetoric or ableist discourse that positions people with disabilities as burdens, inherently deficient, or socially devalued, can be reproduced, complicated, …

Fugue State: The Body Politic Of Identity Theatre, Sean Patrick Ryan

Graduate Theses and Dissertations

This paper will cover my approach as an actor and a creator in two separate productions during my time at the University of Arkansas. The first is a solo theatre piece that I devised under the tutelage of Michael Landman. This work, Half Brother, is a piece of theatre that takes an abstract approach to exploring the familial and psychological ripples that can occur in the life of a queer person when a parental unit takes on a political and religious identity that is oppositional to the LGBTQ community. It exposes the effect this transformation has on the child, their …

Purifai: Detecting And Fixing Search-Induced Distortions In Web-Augmented Llms, Guoqing Wang, Zhao Zhang, Zeyu Sun, Xiaofei Xie, Yizhou Chen, Yanchao Tan, Dan Hao

Research Collection School Of Computing and Information Systems

As Large Language Models (LLMs) increasingly serve as interfaces for proprietary data (e.g., enterprise knowledge bases, legal statutes), ensuring their fidelity to trusted internal information is paramount. While integrating real-time web search can enhance model utility, it introduces a critical vulnerability: the ingestion of conflicting, misleading, or hallucinated content from the open web can override the model's adherence to its verified internal knowledge. We define this failure mode as search-induced distortion, a significant risk in high-stakes domains where the internal knowledge base serves as the absolute ground truth.To address this challenge, we present PurifAI, a proactive, model-agnostic, cache-level purification system …

Verbalizing Lightgcn: Direct Learning Of Textual Representations From User-Item Interaction Graph Via Llms, Manh-Khanh Ngo Huu, Hady Wirawan Lauw

Research Collection School Of Computing and Information Systems

In this work, we propose VerbaLightGCN, a novel LLM-based recommendation framework that integrates the semantic understanding of LLMs with user-item interaction modeling. Traditional collaborative filtering (CF) models typically embed user and item IDs into a latent space to capture interaction signals. However, pretrained LLMs cannot natively interpret these learned embeddings. To bridge this gap, VerbaLightGCN adopts a CF-as-text paradigm, in which collaborative signals are encoded in textual form and directly learned from the user–item interaction graph, and are then combined with semantic information to construct user and item profiles that function as latent embeddings. Inspired by LightGCN, our method retains …

Generation-Augmented Video Corpus Moment Retrieval, Mingjin Kuai, Qianyin Xiao, Juncheng Li, Jin Peng, Lizi Liao, Wei Ji

Research Collection School Of Computing and Information Systems

Video Corpus Moment Retrieval (VCMR) requires models to efficiently retrieve and precisely locate specific moments relevant to natural language queries within a massive, untrimmed video corpus. However, existing discriminative approaches typically rely on shallow visual-textual feature matching mechanisms, which often struggle to capture fine-grained semantic differences. To address this limitation, we propose Video-GAR, a novel framework that reframes the conventional retrieval task from superficial matching to generative understanding, positing that the capability for query reconstruction evidences deep semantic comprehension. Specifically, Video-GAR orchestrates three synergistic components: To overcome the computational efficiency bottleneck, we construct a Bi-Mamba backbone that leverages the linear …

Understanding Chronic Bodily Pain Across Multisystemic Illnesses: A Comparative Analysis Using Exploratory Factor Analysis And Invariance Testing, Suvetha Ravichandran

Theses and Dissertations from DePaul University

Chronic pain is a debilitating symptom that often remains under-researched and poorly treated, particularly for those with multisystemic conditions. This study explored the shared and distinct experiences of chronic bodily pain across four multisystemic illnesses: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS), Multiple Sclerosis (MS), Post-Polio Syndrome (PPS), and Long COVID. Comparative exploratory factor analysis (EFA) and invariance testing were used to examine commonalities and differences in underlying pain experiences for those with the above diagnoses as well as healthy controls based on their responses to pain items from the DePaul Symptom Questionnaire-1 (DSQ-1). A total of 3,724 participants were included: 2,403 …

Analysis Of Clinical Health Worker Notes With Explainable Artificial Intelligence For A Better Perspective In Preventing Readmission, Christopher Scott Klimp

Theses and Dissertations from DePaul University

Emergency Department (ED) readmissions remain a major challenge for healthcare systems, affecting both patient care quality and financial costs. Most prediction models depend largely on structured data from clinical tools that assign points to a small set of predefined factors – such as comorbidities, previous hospital visits, and behaviors such as smoking and drinking – and then sum those points to produce an overall risk score. These point-based tools leave out important information from Social Determinants of Health and a wealth of information from Community Health Workers (Community health workers (CHWs), trusted members of a community who help connect people …

Breaking Point: Exploring Side Bets, Support, And Pension Investment As Predictors Of Turnover Intention In Urban Policing, Kenneth M. Quick, Kevin T. Wolff

Publications and Research

Guided by Becker’s side-bet theory, this study examines turnover intentions among officers in the New York City Police Department by assessing pension investment and affective commitment as accumulated organizational “side bets,” alongside job satisfaction, perceived organizational support, and workplace conditions. Data was drawn from an online survey of 1,454 officers. Population weights based on gender, race/ethnicity, and rank were applied to improve representativeness. Weighted multinomial logistic regression models examined predictors of turnover intention, while second-difference tests of average marginal effects (AMEs) assessed whether workplace factors operated differently across tenure groups. Results indicate that perceived organizational attitudes and workplace conditions significantly …

Advancing Dimensional Models Of Psychopathology In Cancer: Insights From Applying The Hierarchical Taxonomy Of Psychopathology (Hitop), Darren Haywood, Jai Carmichael, Miriam K. Forbes, Roman Kotov, Robert F. Krueger, Aidan G.C. Wright, Susan L. Rossell, David Castle, Wendy Lam, Maryam Lustberg, Colin E. Vize, Anna Ugalde, Phyllis Butow, Nicolas H. Hart

Research outputs 2022 to 2026

Objective: Cancer survivors, defined as those living with or beyond a cancer diagnosis, experience more than double the prevalence of psychopathology when compared to the general population. Categorical diagnostic systems, such as the Diagnostic and Statistical Manual of Mental Disorders (DSM), remain dominant in psycho-oncology despite concerns about reliability, validity, and clinical utility for this population. Dimensional frameworks, such as the Hierarchical Taxonomy of Psychopathology (HiTOP), offer a more precise alternative; however, they have not yet been widely applied in cancer survivors. Accordingly, the objective of this research was to examine the applicability of HiTOP to cancer survivors. Methods: Data …

Sliding Mode Control In Grid-Tied Inverters: Techniques, Applications, And Future Directions, Taimoor Muzaffar Gondal, Asma Aziz, Daryoush Habibi, Iftekhar Ahmad

Research outputs 2022 to 2026

Sliding mode control (SMC) has emerged as a robust and adaptive strategy for grid-tied inverters. It has unique attributes which make it an appropriate control choice under increasingly complex power systems. However, there are a limited number of review articles which comprehensively explore the true potential of SMC for grid-tied inverter applications. This systematic review has been structured to explore SMC techniques with respect to certain challenges, i.e., stability challenges, unbalanced conditions, low inertia scenarios, and harmonics distortion. The classification of various SMC techniques is presented with respect to their control law and sliding surfaces design. The comparative analysis indicates …

Co-Design And Pilot Testing Of A Nurse-Led Gp-Supported Self-Management Intervention For Breast Cancer Survivors With Cardiovascular Diseases: A Study Protocol, Anu Correya, Jing Yu Tan, Xian Liang Liu, Daniel Bressington, Leah East, Tao Wang

Research outputs 2022 to 2026

Background/Objectives: Self-management intervention strategies are recommended to address the needs of breast cancer survivors (BCSs) who have cardiovascular diseases (CVDs). Despite the benefits, these strategies are often suboptimally implemented, resulting in inadequate cardiovascular management among BCS with CVDs. The project aims to develop a nurse-led and GP-supported self-management (NGPS) intervention to reduce cardiovascular risks in BCS with CVDs and evaluate its feasibility and potential effects in primary care settings using the double diamond co-design process and the Medical Research Council (MRC) Framework for Developing and Evaluating Complex Interventions.

Methods: The study will be conducted in two phases. In phase I, …

Relationship Of Early Rapid Weight Loss To Efficacy And Safety Of Tirzepatide And Semaglutide For Obesity: Surmount-5 Post Hoc Analysis, Louis J Aronne, Deborah B Horn, Alexander D Kokkinos, Beverly L Falcon, Hui Wang, Hunter T Hoffmann, Clare J Lee, Julia P Dunn

Faculty, Staff and Student Publications

Background: A knowledge gap remains in how the rate of body weight reduction impacts efficacy and safety of obesity management medications. This SURMOUNT-5 post hoc analysis aimed to define rapid responders and evaluated efficacy and safety of tirzepatide versus. semaglutide in rapid responders versus non-rapid responders.

Methods: Rapid responders and non-rapid responders were defined as participants that obtained ≥ 15% and < 15% body weight reduction by Week 24, respectively. Baseline characteristics, and proportion of participants achieving body weight reduction thresholds by Week 72 were assessed. End-of-study safety and gastrointestinal adverse events measures were summarized.

Results: Overall, 32.3% were rapid responders (tirzepatide: 44%, semaglutide: 21%). The proportions of participants achieving body weight reduction thresholds by Week 72 were higher for rapid responders versus non-rapid responders. Safety trends for tirzepatide and semaglutide were similar between …

Do I Fit In Here? How Immigrants’ Endorsement Of Social Markers Is Related To Their Wellbeing In Three Migrant Receiving Societies, Justine Dandy, Chan Hoong Leong, Adam Komisarof, Travis Lim, Krystal Perkins

Research outputs 2022 to 2026

Social markers of acceptance are the attributes seen as important for immigrants to achieve or display in order to be accepted similarly to a local-born citizen. In research on constructions of national identity, factor analysis reveals a culture-specific, two-factor structure akin to the ethnic–civic distinction observed. In this paper, we extend this work by examining the relationship between social markers of acceptance and immigrants’ well-being in three national contexts: Australia, Japan, and the U.S. Specifically, in study 1 we examined the factor structure of social markers in non-immigrant samples in each context (Australia, n = 1357; Japan, n = 2000; …

Using Μ-Xrf To Quantify Redox Features In Bottomland Hardwood Forest Soils, Jaslyn Ilene Herold

Electronic Theses and Dissertations Archive

Secondary soil iron-manganese concentrations (FMC) form because of soil redox reactions and influence the cycling and availability of key nutrients and toxins in wetland ecosystems. However, the effects of landscape position and soil age on FMC formation are not well understood. This study examines the µ-XRF spectroscopy of FMC from soils sampled along terrace, floodplain, and channel landscapes in a bottomland hardwood forest in the Mississippi Embayment of western Kentucky. We found that FMC area is linked to epi- and endosaturation. The 25 µm resolution µ-XRF chemical maps show decreasing area with depth in episaturated conditions along the terrace and …

Mental Health Stigma In Latin American Youth With Obsessive-Compulsive Disorder, Ryan J. Mccarty, Matias Jensen, Elissa Paz Jensen Santa Cruz, Melisa N. Sagarnaga, Pamela Claisse Quiroz, Isabella A. Cruz, Laura D. Seligman, Joseph D. Hovey, Monica E. Ochoa, Nadeen Salhadar

Psychological Science Faculty Publications

Obsessive-compulsive disorder (OCD) is an impairing mental health condition that affects individuals across the lifespan. Some literature has evaluated mental health stigma in adults with OCD but limited attention has been given to youth, as well as those of Latino ancestry. The present study reported the extent of stigma experiences among Latin American youth with OCD, as well as examined sociodemographic, clinical and psychological correlates. 235 youth with clinician confirmed diagnoses of OCD from 10 North and South American countries completed questionnaires assessing stigma experiences, symptomatology, and functioning. Sociodemographic and clinical variables were also collected. The average OCD stigma response …

Antibiotic Prophylaxis And Surgical Site Infection In Mohs Surgery: A Systematic Review, Jordan Shelestak, Olivia Sheppard, Jared Nichols

Student Publications

Introduction
Mohs micrographic surgery is a highly effective treatment for non-melanoma skin cancers, with low complication rates. Despite a baseline surgical site infection rate of approximately 2%, prophylactic antibiotic use remains common, though its benefit is uncertain. The primary objective of this systematic review is to evaluate whether the benefits of prophylactic antibiotic use in Mohs micrographic surgery outweigh the potential risks, particularly with regard to preventing surgical site infection.

Methods
This systematic review was conducted in accordance with PRISMA guidelines. A comprehensive PubMed search identified studies published between 2010 and 2025 that investigated antibiotic use before, during, or after …

Weathering In-Beyond Hydro(S)Cenes, Bronwyn Bailey-Charteris, Kathryn Mcmahon, Mary Christina Rapp, Natalie Davey

Research outputs 2022 to 2026

This paper examines the cultural, ecological and political dimensions of the contemporary water crisis through a curatorial and environmental humanities lens. Through the watery conceptual epoch of the Hydrocene, it highlights the work of artists and interdisciplinary practitioners whose hydro-Artistic methods propose ways to realign human-water relations. Central to this analysis is the role of curatorial care, here understood as a practice of convening, holding and enabling reciprocal encounters between bodies of water, artists and communities. Bringing these diverse practices into dialogue, the paper proposes water not only as subject matter but as a potential collaborator and leaky agent. It …

Nb–Zr Interdiffusion In The (Hf,Ta,Ti,Nb)B2–(Hf,Ta,Ti,Zr)B2 High-Entropy Boride System, Ana C. Feltrin, Yue Zhou, Stefano Curtarolo, Gregory E. Hilmas, William G. Fahrenholtz

Materials Science and Engineering Faculty Research & Creative Works

Niobium (Nb) presents a challenge for achieving homogeneous solid solutions in high-entropy boride and carbide ceramics. In this work, we synthesized (Hf,Ta,Ti,Zr)B₂ (Zr-HEB) and (Hf,Ta,Ti,Nb)B₂ (Nb-HEB) ceramics by boro/carbothermal reduction, followed by densification via spark plasma sintering (SPS). Diffusion couples were prepared to investigate the interdiffusion behavior of Nb and zirconium (Zr) in these complex systems at temperatures between 2000°C and 2200°C. Aside from Nb and Zr, the other elements exhibited no significant concentration gradient across the diffusion interface. Nb exhibited lower diffusion coefficients ((Formula presented.) Nb = 1.7 x 10⁻¹⁷–1.2 x 10⁻¹⁶ m^{2 …}

Vehicle Autonomy To Ecosystem Intelligence: A Systematic Review Of Dynamic Vision Architectures In Surface Mining Operations, Nana Yaa Damtewaa Anti, Samuel Frimpong, Muhammad Azeem Raza

Mining Engineering Faculty Research & Creative Works

Autonomous Haulage Systems (AHS) have significantly transformed surface mining operations by improving safety, productivity, and operational consistency. Currently, AHS predominantly rely on vehicle-centric perception architectures. Onboard LiDAR, radar, cameras, and Global Navigation Satellite Systems (GNSS) perform sensing, interpretation, and decision-making within individual systems. These processes enable collision avoidance and path tracking. However, they are limited in their ability to consider the broader, dynamic mining environment characterized by dust, terrain degradation, geotechnical instability, heterogeneous traffic, and rapidly evolving operational conditions. This paper presents a systematic review of dynamic vision systems of AHS in surface mining. It critically analyzes the transition from …

Finite-Temperature Grain-Boundary Segregation In High-Entropy Carbides, Marium M. Mou, Tarek Md Anamul Haque, Samuel E. Daigle, Josiah Roberts, William G. Fahrenholtz, Jon Paul Maria, Douglas E. Wolfe, Eva Zurek, Stefano Curtarolo, Donald W. Brenner

Materials Science and Engineering Faculty Research & Creative Works

High-entropy carbides are promising candidates for extreme-temperature environments, but their grain-boundary chemistry remains difficult to resolve because segregation involves both chemical disorder and finite-temperature configurational sampling. Here, we quantify temperature-dependent grain-boundary segregation in high-entropy carbides using a universal message-passing atomic cluster expansion (MACE) machine learning interatomic potential combined with a hybrid Monte Carlo–molecular dynamics workflow. A 53.1 (Formula presented.) (Formula presented.) symmetric tilt grain boundary was sampled for six representative high-entropy carbide compositions containing group IV, V, and VI transition metals at 300 and 2000 K. Element-resolved metal-sublattice composition profiles reveal composition-dependent segregation modes. Several chemistries exhibit selective near-boundary enrichment …

A Comparative Evaluation Of Experimentally Validated Finite Element Modeling Strategies To Simulate Compression-After-Impact Behavior Of Multidirectional Cfrp Laminates With Barely Visible Impact Damage, Niildiip Chandraa, Vinh Tung Le, Arief Yudhanto, Abhendra K. Singh, Douglas E. Smith

Mechanical and Aerospace Engineering Faculty Research & Creative Works

Abstract The finite element-based approaches to predict compression-after-impact (CAI) performance of carbon fiber reinforced polymer (CFRP) subject to a low-velocity impact rely on assumptions about compressive failure mechanisms within the barely visible impact damage (BVID). Comprehensive evaluations of finite element (FE) models concerning accuracy, efficiency, and validity with respect to experimental tests are limited. This study explores several finite element-based approaches developed in ABAQUS Explicit to predict the residual strength and related compressive failure mechanisms of multidirectional CFRP laminates. Drop-weight impact experiments followed by CAI tests employing 3D Digital Image Correlation (DIC) were used to validate our FE models by …

Larger Is Not Always Better: Exploring Small Open-Source Language Models In Logging Statement Generation, Renyi Zhong, Yichen Li, Guangba Yu, Wenwei Gu, Jinxi Kuang, Yintong Huo, Michael R. Lyu

Research Collection School Of Computing and Information Systems

Developers use logging statements to create logs that document system behavior and aid in software maintenance. As such, high-quality logging is essential for effective maintenance; however, manual logging often leads to errors and inconsistency. Recent methods emphasize using large language models (LLMs) for automated logging statement generation, but these present privacy and resource issues, hindering their suitability for enterprise use. This paper presents the first large-scale empirical study evaluating small open-source language models (SOLMs) for automated logging statement generation. We evaluate four prominent SOLMs using various prompt strategies and parameter-efficient fine-tuning techniques, such as Low-Rank Adaptation (LoRA) and Retrieval-Augmented Generation …

Multicbr: Multi‑View Contrastive Learning For Bundle Recommendation, Yunshan Ma, Yingzhi He, Xiang Wang, Yinwei Wei, Xiaoyu Du, Yuyangzi Fu, Tat‑Seng Chua

Research Collection School Of Computing and Information Systems

Bundle recommendation seeks to recommend a bundle of related items to users to improve both userexperience and the profits of platform. Existing bundle recommendation models have progressed from capturing only user-bundle interactions to the modeling of multiple relations among users, bundles, and items.CrossCBR, in particular, incorporates cross-view contrastive learning into a two-view preference learningframework, significantly improving SOTA performance. It does, however, have two limitations: (1) the twoview formulation does not fully exploit all the heterogeneous relations among users, bundles, and items; and(2) the “early contrast and late fusion” framework is less effective in capturing user preference and difficultto generalize to …

Deep Learning For Video Anomaly Detection: A Review, Peng Wu, Chengyu Pan, Yuting Yan, Guansong Pang, Qingsen Yan, Peng Wang, Yanning Zhang

Research Collection School Of Computing and Information Systems

Video anomaly detection (VAD) aims to discover behaviors or events deviating from the normality in videos. As a long-standing task in the field of computer vision, VAD has witnessed much good progress. In the era of deep learning, with the explosion of architectures of continuously growing capability and capacity, a great variety of deep learning-based methods are constantly emerging for the VAD task, greatly improving the generalization ability of detection algorithms and broadening the application scenarios. Therefore, such a multitude of methods and a large body of literature make a comprehensive survey a pressing necessity. In this article, we present …

Anomaly Management In Unmanned Aerial Vehicles: A Systematic Literature Review, Ivan Tan Wei Han, Christopher M. Poskitt, Lingxiao Jiang, Lwin Khin Shar

Research Collection School Of Computing and Information Systems

Unmanned Aerial Vehicles (UAVs) are increasingly deployed in safety-critical applications such as logistics, surveillance, disaster response, and urban air mobility. While their autonomy enables powerful capabilities, it also introduces vulnerabilities due to hardware faults, software defects, communication failures, and adversarial interference. This survey presents a comprehensive review of research studies closely related to UAV anomalies published between 2015 and 2025, covering 111 papers from academic and industrial sources. We introduce a unified five-pillar taxonomy—anomaly generation, prevention, detection, recovery, and analysis—that organizes existing work across the full anomaly management lifecycle. In contrast to prior surveys that focus primarily on detection algorithms, …

Towards Uniformity And Alignment For Multimodal Representation Learning, Wenzhe Yin, Pan Zhou, Zehao Xiao, Jie Liu, Shujian Yu, Jan-Jakob Sonke, Efstratios Gavves

Research Collection School Of Computing and Information Systems

Multimodal representation learning aims to construct a shared embedding space in which heterogeneous modalities are semantically aligned. Despite strong empirical results, InfoNCE-based objectives introduce inherent conflicts that yield distribution gaps across modalities. In this work, we identify two conflicts in the multimodal regime, both exacerbated as the number of modalities increases: (i) an alignment–uniformity conflict, whereby the repulsion of uniformity undermines pairwise alignment, and (ii) an intra-alignment conflict, where aligning multiple modalities induces competing alignment directions. To address these issues, we propose a principled decoupling of alignment and uniformity for multimodal representations, providing a conflict-free recipe for multimodal learning that …