Digital Commons Network^™

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 …

Advancing Imbalanced Classification Through Representation Learning, Morteza Mohammady Gharasuie

Computer Science Theses & Dissertations

Real-world datasets frequently exhibit severe class imbalance, where certain categories are significantly underrepresented relative to others, leading standard empirical risk minimization to bias learning toward majority classes and degrade performance on minority categories. This dissertation addresses imbalanced classification across both structured tabular data and long-tailed visual recognition by developing methods that improve representation learning and decision reliability under skewed distributions. For tabular data, it introduces Conditional Probability Representation (CPR), a target-based encoding framework that embeds feature–label relationships directly into the representation space, enhanced by a progressive feature upgrading mechanism for semi-supervised settings and a class-frequency–aware extension that improves robustness to …

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 …

Using Recommender Systems To Help Revitalize Local News, Payam Pourashraf

Theses and Dissertations from DePaul University

Local news outlets have experienced steep declines in readership as national and global online news sources have expanded and consolidated audience attention. Because many local media companies rely primarily on subscription-based business models, they must increase user engagement and provide clearer value to local subscribers. Personalization and recommender systems offer one promising path toward this goal. However, many recommender systems assume that a user’s past behavior can be summarized in a single unified profile. In practice, preferences are often context-dependent: the same person may prefer different content depending on geography, time, task, or situation. Local news offers a socially important …

Does Poor Regulation Lead To Pollution? Key Takeaways From A Systematic Review, Vincent Emenyonu, Kat O. Mara, Mark Lund

Research outputs 2022 to 2026

Formal recycling of used lead acid batteries (ULAB) is a multibillion-dollar industry, valued at US$12.12 billion in 2024, and expected to grow to more than $26 billion within eight years. Demand for recycled lead (Pb) for new batteries grows in developing countries where costs limit the uptake of lithium-ion batteries. ULAB recycling in these regions poses significant danger to human health and the environment. Stringent regulatory approaches for ULAB recycling in developing countries, intended to reduce pollution, have thus far been ineffective as they typically do not adequately cover small-scale informal recycling in rural and remote communities. We examine what …

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 …

Toward A Didactical Phenomenology For The Completeness Axiom, Sean Larsen, Tenchita Alzaga Elizondo, Kristen Vroom, Stephen Strand Ii

School of Mathematical & Statistical Sciences Faculty Publications

The study is part of an instructional design project focused on introductory real analysis. The goal of the project is to develop a theoretically grounded and empirically supported instructional approach that builds on students’ experiences in the calculus sequence to engage them in the reinvention of the rigorous foundations of the calculus. An essential aspect of this foundation is the completeness of the real numbers. Drawing on the didactical phenomenology heuristic from the theory of Realistic Mathematics Education (RME), we conducted an iterative instructional design study focused on the completeness axiom. The work proceeded in two phases. First, we conducted …

Quantization Dimension For A Generalized Inhomogeneous Bi-Lipschitz Iterated Function System, Shivam Dubey, Mrinal Kanti Roychowdhury, Saurabh Verma

School of Mathematical & Statistical Sciences Faculty Publications

For a given r∈(0,+∞), the quantization dimension of order r, if it exists, denoted by Dr(μ), of a Borel probability measure μ on R^d represents the speed how fast the nth quantization error of order r approaches to zero as the number of elements n in an optimal set of n-means for μ tends to infinity. If Dr(μ) does not exists, we call D̲r(μ) and D¯r(μ), the lower and upper quantization dimensions of μ of order r. In this paper, we estimate the quantization dimension of condensation measures associated with condensation systems ({fi}i=1^N,(pi)i=0^N,ν), where the …

Ai-Powered Synthetic Biology: Current Situation, Challenges, And Future Perspectives, Izem Olcay Sahin, Nuriye Gokce, Duygu T. Yildirim, A. Baki Yildirim, Hilal Akalın, Donald Martin, Tommaso Beccari, Oscar Vicente, Iza Radecka, Fideline Tchuenbou-Magaia, Robert S. Marks, Ratnesh Lal, Satya Prakash, Adam Mechler, Mario Petrov Milkov, Svetlana Fotkova Georgieva, Ilia Iliev, Kosi Gramatikoff, Ed Judge, Milica Markovic, Radka Kaneva, Galina Aleksieva Yaneva, Nadya Vasileva Agova, Nikoleta Dobromirova Ivanova, Mariya Kiryakova Tsvetkova, Ivelin Rosenov Iliev, Michel Salzet, Kisung Ko, Michele Maffia, Chiara Coppola, Matteo Bertelli

Research Outputs: 2025-Present

Synthetic biology has evolved from a set of engineering aspirations to an operationally sophisticated discipline, and artificial intelligence (AI) is its fastest-growing accelerant. This review traces that convergence across six interlocking domains: systems-level biological modeling, de novo protein engineering, metabolic and microbial programming, multi-omics data integration, regulatory element design, and clinical translation. For each domain, we survey established results, integrate findings from 2010–2026 literature, and articulate the trajectories that will define the next decade. Emerging themes include physics-informed neural networks for mechanistically constrained biological modeling, drug design, and federated learning architectures that allow global omics collaboration without centralizing sensitive data, …

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 …

Variational Speculative Decoding: Rethinking Draft Training From Token Likelihood To Sequence Acceptance, Xiandong Zou, Jianshu Li, Jing Huang, Pan Zhou

Research Collection School Of Computing and Information Systems

Speculative decoding accelerates inference for (M)LLMs, yet a training-decoding discrepancy persists: while existing methods optimize single greedy trajectories, decoding involves verifying and ranking multiple sampled draft paths. We propose Variational Speculative Decoding (VSD), formulating draft training as variational inference over latent proposals (draft paths). VSD maximizes the marginal probability of target-model acceptance, yielding an ELBO that promotes high-quality latent proposals while minimizing divergence from the target distribution. To enhance quality and reduce variance, we incorporate a path-level utility and optimize via an Expectation-Maximization procedure. The E-step draws MCMC samples from an oracle-filtered posterior, while the M-step maximizes weighted likelihood using …

Train In Vain: Functionality-Preserving Poisoning To Prevent Unauthorized Use Of Code Datasets, Yuan Xiao, Yuchen Chen, Jiaming Wang, Wei Song, Jun Sun, Shiqing Ma, Yanzhou Mu, Juan Zhai, Chunrong Fang, Jin Song Dong, Zhenyu Chen

Research Collection School Of Computing and Information Systems

The widespread availability of large-scale code datasets has accelerated the development of code large language models (CodeLLMs), raising concerns about unauthorized dataset usage. Dataset poisoning offers a proactive defense by reducing the utility of such unauthorized training. However, existing poisoning methods often require full-dataset poisoning and introduce transformations that break code compilability. In this paper, we introduce FunPoison, a functionality-preserving poisoning approach that injects short, compilable weak-use fragments into executed code paths. FunPoison leverages reusable statement-level templates with automatic repair and conservative safety checking to ensure side-effect freedom, while a type-aware synthesis module preserves type correctness, suppresses static-analysis warnings, and …

Itimo: An Llm-Empowered Synthesis Dataset For Travel Itinerary Modification, Zhuoxuan Huang, Yunshan Ma, Hongyu Zhang, Hua Ma, Zhu Sun

Research Collection School Of Computing and Information Systems

Addressing itinerary modification is crucial for enhancing the travel experience as it is a frequent requirement during traveling. However, existing research mainly focuses on fixed itinerary planning, leaving modification underexplored due to the scarcity of shape need-to-modify itinerary data. To bridge this gap, we formally define the itinerary modification task and propose a general pipeline to construct the corresponding dataset, namely iTIMO. This pipeline frames the generation of shape need-to-modify itinerary data as an intent-driven perturbation task. It instructs large language models to perturb real-world itineraries using three operations: REPLACE, ADD, and DELETE. Each perturbation is grounded in three intents: …

Bridging Llm Embeddings And Vae Parameters For Disentangled Recommendation, Nhu-Thuat Tran, Hady Wirawan Lauw

Research Collection School Of Computing and Information Systems

Disentangled recommendation within the Variational Autoencoder (VAE) framework aims to capture multiple user interests. While effective, these VAEs are fundamentally constrained by their reliance on interaction data alone, lacking the rich external semantic knowledge needed to properly structure and separate latent interests. Meanwhile, Large Language Models (LLMs) excel at deriving profound user preference signals from textual data. Prevailing methods for integrating LLMs into recommendation, however, either focus on single-interest modeling or perform a shallow fusion by aligning LLM and VAE representation spaces. Thus, they fail to fundamentally shape the VAE's latent space for multi-interest learning, hindering recommendation performance. To bridge …

Neural Network Technologies In The Automatical Control Systems Of Absorption Process For Pureficating Natural Gas, Abdishukurov Maqsudovich Shavkat Mr, Xuecheng Li Li Xuecheng Mr

Technical science and innovation

Analysis of methods and algorithms for synthesizing adaptive control systems for technological processes based on the neural network approach is carried out in this search. The stages of mathematical modeling of complex technological processes using neural network technology were considered. Additionally, an algorithm for solving the interpolation and extrapolation problem that arises in the training process a neural network to control system was proposed. At the final stage of this article, algorithms based on neural network technology are synthesized for the control system for the parameters of the technological process of natural gas purification by absorption

Semantic Context Improvisational Retrieval-Augmented Generation For Empathic Conversational Ai, Sharjeel Tahir, Judith Johnson, Jumana Abu-Khalaf, Syed Afaq Ali Shah

Research outputs 2022 to 2026

A fundamental limitation of modern conversational AI is its limited capacity to demonstrate sustained empathy in long-form interactions. We propose SCIRAG (Semantic Context Improvisational Retrieval-Augmented Generation), a feedback-driven retrieval framework for adaptive empathic dialogue. It employs a dual-loop retrieval framework, iteratively optimizing a static counseling dataset through user metadata and feedback memory refinement. To enhance contextual alignment, we deploy retrieval adaptation, enabling the model to retain and leverage past conversational cues based on user preferences. When integrated with Mixtral-8x7B, SCIRAG improves human-rated empathic understanding by +1.26 points and empathic response by +1.00 point on the RoPE scale, while increasing acceptability …

Farm Workers And The Structural Basis Of Climate Change Perceptions, Cody Mortell, Anna Erwin, Zhao Ma, Dongkyu Kim, Elizabeth Rodriquez, Marla Perez-Lugo

School of Earth, Environmental, & Marine Sciences Faculty Publications

Farm workers are particularly vulnerable to climate change, especially from extreme heat; however, little is known about how they perceive this risk. This study examines climate change perceptions (CCP) among 404 farm workers surveyed in person in the Lower Rio Grande Valley, Texas from April to May 2024. Guided by Pierre Bourdieu’s theory of practice, we conducted a latent class analysis of economic, cultural, social, and symbolic capital indicators to identify distinct class profiles. We then assessed CCP by latent profile, using two primary outcome variables: (1) perceived temperature trends in the region (assessed alongside NOAA data) and (2) perceived …

An Emergent Seafood Nationalism Takes Hold In The United States, Owen Temby

School of Earth, Environmental, & Marine Sciences Faculty Publications

This paper introduces the concept of ‘seafood nationalism’ to describe how, since 2024–2025, the United States increasingly uses tariffs, trade remedies, food safety enforcement and origin-based rules to favour American seafood over foreign products. Federal actions include Executive Order 14276, Restoring American Seafood Competitiveness, which directs agencies to address unfair trade practices, unsafe imports and to develop a coordinated seafood trade strategy. Additional measures include a tariff package imposing a 10% duty on nearly all seafood imports and 30% on seafood from China, as well as new anti-dumping and countervailing duty orders on warm-water shrimp. This paper presents Gulf …

Hsv-1 Us3 Hijacks Conserved Actin Regulatory Complexes To Drive F-Actin Remodeling, Md Imran Hossain, Md Arifuzzaman, Md Mehedi Hasan, Seung Jong Park, Leila Rahimian, Ojasvi Dutta, Vladimir Chouljenko, Harikrishnan Mohan, Reza Ghavimi, Konstantin G. Kousoulas

Computer Science Faculty Research & Creative Works

The herpes simplex virus 1 (HSV-1) US3 is a multifunctional serine/threonine kinase that promotes HSV-1 replication and spread. But its role and the mechanisms by which US3 regulates actin cytoskeletal remodeling remain poorly defined. We combined flow cytometry, confocal microscopy, immunoprecipitation-mass spectrometry (IP-MS), protein complex mapping, and machine learning to characterize US3-mediated F-actin dynamics. Flow cytometry and confocal microscopy showed that wild-type HSV-1 induces significant F-actin remodeling, while the ΔUS3 mutant displays F-actin levels comparable to uninfected cells, identifying US3 as a key regulator. IP-MS identified 47 high-confidence US3 interactors enriched in conserved actin regulatory complexes, including Arp2/3 nucleation machinery, …

Multi-Angle Dynamic Light Scattering Experiments With Real-Time Autocorrelation Using Field-Programmable Gate Arrays, Spiros Simotas

Theses and Dissertations from DePaul University

Dynamic light scattering (DLS) determines particle size from temporal fluctuations in coherently scattered light. Here a field-programmable gate array (FPGA) computes intensity autocorrelations in real time, and an automated diffractometer with custom 3D-printed cuvette, laser, homing, and GT2-gear mounts enables multi-angle measurements. Square-wave timing tests at 1 Hz, 100 Hz, and 10 kHz verified the correlator response. Calibrated latex spheres were then measured at seven scattering angles from 32.5◦ to 44.5◦. Each autocorrelation function was fit to the standard single-exponential form g(τ ) = B + βe−2Γτ with weights from analytic per-lag Poisson uncertainties σg = pg/N based on coincidence …

Towards Understanding Users' Theory Of Recommender Systems, Mohammed Muheeb Faizan Ghori

Theses and Dissertations from DePaul University

Recommender systems have become deeply embedded in digital platforms such as streaming services, social media, and e-commerce applications, shaping how users discover information, consume content, and interact with online environments. Despite their widespread adoption, these systems are frequently perceived as opaque ‘black boxes’, leaving users uncertain about how recommendations are generated, why certain content is prioritized, and whether their interactions meaningfully influence recommendation outcomes. Existing research at the intersection of recommender systems and human–computer interaction has largely focused on improving transparency, explainability, and recommendation quality from a system-centered perspective. However, limited research has examined how users conceptualize recommender systems and …

Prototyping Of A Spectropolarimetry Module For Lenslet-Based Integral Field Spectrographs, Manasa Lakshmi Narasimhan

Theses

Integral-field spectropolarimetry enables simultaneous spatial, spectral, and polarization measurements. Observations of polarized light allow researchers to map magnetic fields, distinguish scattering from intrinsic emission, and characterize the geometry of sources illuminating the large-scale dust distribution within and around galaxies. In tandem with spectral information, polarized emission can be used to detail the kinematics of dust expelled from galaxies and to probe the underlying spectrum of scattered light—measurements that help constrain key open questions such as the missing baryon problem and the role of feedback in galaxy evolution. However, spectropolarimetry is a rare capability among ground-based observatories. This thesis showcases a …

Comparing Deposition-Dependent Variations In Tungsten Oxide Thin-Film Composition Using X-Ray Photoelectron Spectroscopy, Tevfik Ozkaynak

Theses

In this work, tungsten oxide thin films were fabricated using magnetron sputter deposition and electron beam evaporation under varying oxygen conditions to produce a range of stoichiometries, including substoichiometric compositions near WO₂.₇. The primary goal was to evaluate how reliably X-ray photoelectron spectroscopy (XPS) can quantify the stoichiometry of WOₓ thin films, determine how deposition conditions influence measured composition and oxidation states, and assess how argon ion sputtering, used for surface cleaning, alters tungsten oxidation states and affects XPS based stoichiometry. The films were analyzed using XPS under ultra-high vacuum conditions. Survey spectra were used to identify elemental composition, while …

Project Prioritization For Near-Optimal Functionality Risk Reduction In Bridge Networks, David Y. Yang, Anteneh Deriba

Civil and Environmental Engineering Faculty Publications and Presentations

Bridge retrofit prioritization is commonly guided by structural vulnerability, traffic conditions, or indirect consequences following a single bridge failure. However, these metrics often fail to identify retrofit projects that maximally reduce system-level functionality risk, because they do not account for the consequences of joint failures among multiple bridges. This presentation develops a mathematically rigorous, gradient-based approach to bridge retrofit prioritization for achieving a near-optimal reduction in functionality risk. This risk is formulated as the expected increase in total travel time given probabilistic bridge survival or failure under external stressors such as earthquakes. By reinterpreting risk as the evidence term in …

Reports Of Legal Writing’S Death Are Greatly Exaggerated: Why Generative Ai Will Not Kill The Craft, Jayne T. Woods

Journal of Dispute Resolution

In preparing for this essay, I typed the phrase “legal writing is” into Google to see how common searches finished the thought. The only completed thought to appear was “legal writing is hard.” Amen. Legal writing is an amalgamation of logical reasoning, cognitive psychology, classical rhetoric, legal terminology, legal citation, and proper grammar and style, all of which must result in a product accessible to readers of all education levels. When generative AI arrived, producing cogent, plausible legal prose in seconds, it led many to believe that writing, generally, and legal writing, specifically, had reached the end of days.

Glacier Elevation Difference Data For The Conterminous U.S., Andrew G. Fountain, Bryce Allen Glenn, Christopher Mcneil, Brian Menounos

Geology Faculty Datasets

Historic and recent elevation data for the glaciers in the conterminous US were differenced to estimate glacier mass and volume change.

Historic glacier elevations were obtained from the USGS National Elevation Dataset (https://doi.org/10.15760/geology-data.04), a digitized product from the USGS 1:24,000 paper topographic maps that included the glacier extent (Gesch, 2007). Recent glacier elevations were derived from aerial lidar and from satellite radar. The lidar data were largely obtained from the USGS 3D Elevation Program (Stoker and Miller, 2022; US Geological Survey, 2024), and filtered by acquisition date using only those data collected between mid-summer and early autumn. The …