Digital Commons Network^™

Multilingual Cyber Threat Intelligence Feeds Preprocessing For Threat Intelligence Event Extraction: A Systematic Literature Review, Jamal H. Al-Yasiri, Mohamad Fadli Bin Zolkipli, Nik Fatinah N. Mohd Farid

Karbala International Journal of Modern Science

In cyber threat intelligence (CTI), information security specialists face overwhelming data flows from multiple sources, including hacker forums, dark web markets, and social media. These diverse and multilingual information streams require extensive analysis and processing. However, the current state of CTI faces several challenges, such as reliance on manual annotation and evaluation, as well as limited support for non-English languages, which hinders advanced threat detection and comprehensive analysis. This systematic literature review proposes a conceptual framework designed to overcome existing state-of-the-art limitations. It evaluates recent advancements in CTI methodologies by following PRISMA guidelines and analyzing selected studies from reputable sources, …

Coli@Fire2024: Findings Of Word-Level Code-Mixed Language Identification In Dravidian Languages, Asha Hegde, Fazlourrahman Balouchzahi, Sabur Butt, Sharal Coelho, Kavya G, Harshitha S. Kumar, Sonith D, Shashirekha H. L., Ameeta Agrawal

Computer Science Faculty Publications and Presentations

Code-mixing, a linguistic phenomenon where multiple languages are blended within a single text, has become increasingly prevalent in multilingual societies, particularly in digital communication. The CoLI-Dravidian shared task, organized as part of Forum for Information Retrieval and Evaluation (FIRE) 2024, aimed to address these challenges by inviting researchers to develop models capable of classifying words in code-mixed texts involving Dravidian languages — Tamil, Kannada, Malayalam, and Tulu - interwoven with English. The task presents significant challenges due to the complexity of linguistic structures, mixed-language tokens, and dialectal variations, especially in low-resource languages like those in the Dravidian family. The participating …

Assessing The Adversarial Robustness Of Multimodal Medical Ai Systems: Insights Into Vulnerabilities And Modality Interactions, Ekaterina Mozhegova, Asad Masood Khattak, Adil Khan, Roman Garaev, Bader Rasheed, Muhammad Shahid Anwar

All Works

The emergence of both task-specific single-modality models and general-purpose multimodal large models presents new opportunities, but also introduces challenges, particularly regarding adversarial attacks. In high-stakes domains like healthcare, these attacks can severely undermine model reliability and their applicability in real-world scenarios, highlighting the critical need for research focused on adversarial robustness. This study investigates the behavior of multimodal models under various adversarial attack scenarios. We conducted experiments involving two modalities: images and texts. Our findings indicate that multimodal models exhibit enhanced resilience against adversarial attacks compared to their single-modality counterparts. This supports our hypothesis that the integration of multiple modalities …

What Drives Weight Status Among Female University Students? A Machine Learning Analysis Of Sociodemographic, Dietary, And Lifestyle Determinants, Radwan Qasrawi, Abir Ajab, Leila Cheikh Ismail, Ayesha Al Dhaheri, Sharifa Alblooshi, Razan Abu Ghoush, Stephanny Vicuna Polo, Malak Amro, Suliman Thwib, Ghada Issa, Haleama Al Sabbah

All Works

Background: Obesity and underweight are increasingly common among young adult women, often resulting from complex interactions between diet, lifestyle, and socioeconomic factors. This study addresses that gap by applying machine learning to a wide range of behavioral, dietary, and demographic data. The main research question asks: What are the key factors influencing weight status among female university students, and how accurately can machine learning models identify them? We hypothesize that different factors are significantly associated with underweight, overweight, and obesity, and that machine learning can reliably detect these patterns. The aim is to identify the strongest predictors and support more …

Using Confidence Scores To Improve Eyes-Free Detection Of Speech Recognition Errors, Sadia Nowrin, Keith Vertanen

Michigan Tech Publications

Conversational systems rely heavily on speech recognition to interpret and respond to user commands and queries. Despite progress on speech recognition accuracy, errors may still sometimes occur and can significantly affect the end-user utility of such systems. While visual feedback can help detect errors, it may not always be practical, especially for people who are blind or low-vision. In this study, we investigate ways to improve error detection by manipulating the audio output of the transcribed text based on the recognizer's confidence level in its result. Our findings show that selectively slowing down the audio when the recognizer exhibited uncertainty …

"Chatgpt Told Me To Say It": Ai Chatbots And Class Participation Apprehension In University Students, Daisuke Akiba

Publications and Research

The growing prevalence of AI chatbots in everyday life has prompted educators to explore their potential applications in promoting student success, including support for classroom engagement and communication. This exploratory study emerged from semester-long observations of class participation apprehensions in an introductory educational psychology course, examining how chatbots might scaffold students toward active and independent classroom contribution. Four students experiencing situational participation anxiety voluntarily participated in a pilot intervention using AI chatbots as virtual peer partners. Following comprehensive training in AI use and prompt design given to the entire class, participants employed systematic consultation frameworks for managing classroom discourse trepidations. …

Qder: Query-Specific Document And Entity Representations For Multi-Vector Document Re-Ranking, Shubham Chatterjee, Jeff Dalton

Computer Science Faculty Research & Creative Works

Neural IR has advanced through two distinct paths: entity-oriented approaches leveraging knowledge graphs and multi-vector models capturing fine-grained semantics. We introduce QDER, a neural re-ranking model that unifies these approaches by integrating knowledge graph semantics into a multi-vector model. QDER's key innovation lies in its modeling of query-document relationships: rather than computing similarity scores on aggregated embeddings, we maintain individual token and entity representations throughout the ranking process, performing aggregation only at the final scoring stage-an approach we call "late aggregation." We first transform these fine-grained representations through learned attention patterns, then apply carefully chosen mathematical operations for precise matches. …

Type-2 Neutrosophic Set With Mcdm Methodology For Fire Safety Estimation In Healthcare Services, Eman Sayed

Neutrosophic Systems with Applications

Fire safety represents a critical priority in healthcare facilities, where complex infrastructures and the vulnerability of patients present significant challenges to evacuation and emergency response. Traditional fire risk assessment methods often fall short in addressing the linguistic variability, uncertainty, inconsistency, and indeterminacy inherent in expert evaluations. While fuzzy and Neutrosophic approaches have been applied in broader healthcare decision-making contexts, no existing study has utilized Type-2 Neutrosophic Numbers Sets (T2NNs) for prioritizing hospital departments based on fire risk. To address this gap, this study introduces a novel multi-criteria decision-making (MCDM) framework that integrates T2NNs for expert modeling, the Entropy method for …

Bridging Classical Rhetoric And Ai: A Systematic Framework For Developing Authorial Voice Through Large Language Models, Daniel Plate

Theses

This project addresses critical gaps in AI-assisted writing by developing the first systematic framework that integrates classical rhetorical principles with modern large language model capabilities for authorial voice development. The primary focus is on creating reliable methods for stylistic control through strategic AI collaboration rather than ad hoc prompting approaches. The project develops a comprehensive coding system for analyzing prose style, creates ten distinct authorial personas, and establishes a dual curation methodology that structures both stylistic analysis and content preparation. Implementation through the AI Writing Guide website provides practical tools including prompt templates, annotated examples, and instructional materials that demonstrate …

Du Undergraduate Showcase Abstracts: Research, Scholarship, And Creative Works, Kabe Aberle, Nadia Kako, Kateri Mcrae, Brooke Agulnek, Sky Palmon, Yasmine Ramirez, Francisca Aguirre Beltran, Ashley Juarez, Bridget Kim, Tessa Appel, Sterling Kerr, Spencer Ingley, Gabe Meyer, Robin Tinghitella, Dale Broder, Lily Baeza, Chloe Beers, Julia Coakley, Whitney Kelsey, Sydney Gainforth, Gabi Wing, Audrey Martin, Aaliyah Amore Berry, Brooke Watley, Kiruthika Venkatesan, Rachel Bienstock, Annabella Brotherston, Madison Bryant, Mia Burgener, Emma P. Lieb, Rachel A. Johnson, Jennifer L. Hoffman, Kania Campbell, Kiena Campbell, Courtney Cassidy, Sage Krzyzkowski, Maddox Jones, Skylar Abookire, Luke Hawkins, Sunnah Yoon, Andrea Chu, Yan Qin, Nyah Cubbison, Brian Gearity, Daniel Mcintosh, Mariely Cruz, Edward Garrido, Grady Dionne, Nicole Doris, Lyndsie Salvagio, Ann-Charlotte Granholm-Bentley, Anna Dymov, Hannah Eckert, Gabrielle Welsh, Erica Larson, Charlie Ernst, Anna Zhou, Sarah Watamura, Larissa Fedorovich-Klein, Georgie Fields, Kimberly A. Guevara, Aven Mccall, Ben Peltier, Feruz Yahia, Patrick Flores, Jadyn Floyd, Sophia Forcier, J. Von R. Monteza, Peter Sokol-Hessner, Gwendolyn Geiger, Scott Nichols, Camryn Gunter, Kendal Hengst, Charlie Bednarz, Issy Garside, Addison Baker, Rachel Mina, Brooke Hermanson, Amanda Klingler, William Highfill, Sydney Jaques, Kerstin Lewey, Allison Grossery, Daniel Linseman, Ethan Lim, Jagger Livengood, Owen Mantelli, Gabby Pappas, Abby Mcdonald, Madeleine Dierking, Eve Miller, Emma Loeber, Anna Marlow, Michael Kerwin, Ella Mathews, Hillary Hamann, Khadija Mohamed, Vivian Nguyen, Gabri Notov, Ifunayachi Ogbonna-Ukuku, Sunil Kumar, Charles Baysah, Sarah Olson, Don Sullivan, Anna Paradiso, Jay Parrish, Mira Pronobis, Alisha Pravasi, Kerstin Haring, Diego Ramirez, Christopher Reardon, Juliana Ramirez, Casey Doherty, Ella Kestner, Teagan Weindel, Cate Billings, Pablo Torre-Walter, Lucy Rand, Samantha Reynolds, Mark Siemens, Khadeeja Rashid, Laine Satterlee, Piper Heilbronner, Lily Pound, Ben Whitehurst, Anna Respet, Lizzie Lesoing, Sydney Hertel, Aya Saad-Masri, Brooke Ballenger, Max Proske, Hannah Rosenberg, Ellia Nakahara, Sophia Espinoza, Ivan Woolhouse, Simon Ruland, Gorkem Er, Timothy Sweeny, Melaku Saketa, Michela Schenk, Maren Lynch, Madi Hamm, Grace Schroeder, Michelle Rozenman, Rana Seif, Jackson Hall, Marisela Simental, Daniel Paredes, Aaron Mena, Preston Spaan, Evelyn Stovin, David Andrew Swartz, Anh Tran, Daniel Pittman, Luke Farchione, Emily Boyer, Ukari Verner, Lacey Conrad, Jonathan Velotta, James Weiner, Jagger Gossett, Noah Sherry, Sam Proud, Ben Block, Avi Narayana, Zoey Weiss, Alyssa Wilson, Gabrielle Walsh, David Zonana, Keely Wright, Kena Riveria, Lillybelle Deer, Jena Doom, Elysia Davis, Isabelle Yaremenko, Caitlyn Young

DU Undergraduate Research Journal Archive

Abstracts from the DU Undergraduate Research Showcase.

An Action Research Study On Ai Video Vs. Written Feedback: Enhancing Undergraduate Academic Writing And Critical Thinking, Sandra Baroudi, Nida Mubeen, Suha Karaki

All Works

This action research explores how AI-assisted feedback formats influence students’ academic writing and critical thinking, a gap particularly relevant in the digital learning landscape. This study involved 40 undergraduate Emirati students divided into two groups, receiving either AI-generated video or written feedback across three assignments. Using a quantitative design, data were gathered using standardized critical thinking and academic writing skills rubrics. Additionally, students completed post-surveys to capture their perceptions, engagement levels and use of AI tools in the learning process. Results showed written feedback significantly improvemed critical thinking (M = 2.80, SD = 0.75) compared to video feedback (M = …

Syntax-Enhanced Boundary-Aware Named Entity Recognition Model, Chuanming Yu, Bin Deng, Zhengang Zhang

Journal of Scientific Information Research

[Purpose/significance] This study addresses the issue of inadequate perception of entity boundaries in traditional character-level modeling-based named entity recognition models by integrating syntax information containing entity boundary features into the task using a multi-head graph attention network with dense connections. This integration enhances the effectiveness of named entity recognition.

[Method/process] This study proposes a Syntax-enhanced Boundary-aware Named Entity Recognition Model (SynBNER), which utilizes BERT for text semantic representation and integrates syntax information using a dense-connected graph attention network. This integration incorporates implicit entity boundary information from syntax information into word representations, thereby enhancing the model's entity boundary perception capability.

[Result/conclusion] …

From Palimpsest To Prompt: Rewriting Shakespeare, Creative Authorship, And The Generative Logics Of Large Language Models In Contemporary Theatre, Michael Harding, James Hutson

Faculty Scholarship

This article examines the convergence of creative authorship, adaptation, and generative artificial intelligence within contemporary theatre, taking Michael Harding‘s Awake, Young King as a central case study. Through the rewriting of Shakespearean drama, Harding‘s creative process demonstrates how theatrical meaning emerges through ongoing negotiation among playwright, performer, and audience, with scripts historically subject to revision, improvisation, and reinterpretation. Concerns regarding copyright, intellectual property, and the role of AI in the performing arts are reframed as extensions of enduring debates over originality and authorship, rather than novel threats. Tracing the evolution from The Rise of James VI to Awake, Young King, …

Prompttutor: Effects Of An Llm-Based Chatbot On Learning Outcomes And Motivation In Flipped Classrooms, Yuhao Zhang, Eng Lieh Ouh, Chong Jee Adam Ho, Siaw Ling Lo, Kar Way Tan, Feng Lin

Research Collection School Of Computing and Information Systems

This study explores the integration of a Large Language Model (LLM) based chatbot, PromptTutor, into flipped classrooms (FC) for undergraduate Computer Science (CS) education. PromptTutor is designed to provide personalized, immediate feedback to support student learning in FC by incorporating reflective learning and scaffolding strategies. The traditional FC typically lacks this immediate feedback during the pre-class learning phase, risking decreased student motivation according to existing literature. This study examines if students improve in learning outcomes and motivation after using PromptTutor. Through a controlled crossover experiment with 50 students, the study demonstrates statistically significant improvements in students' quiz performance and motivation …

Finir: The 2nd Workshop On Financial Information Retrieval In The Era Of Generative Ai, Fengbin Zhu, Yunshan Ma, Fuli Feng, Chao Wang, Huanbo Luan, Guangnan Ye, Shuo Zhang, Dhagash Mehta, Pingping Chen, Bing Xiang, Tat‑Seng Chua

Research Collection School Of Computing and Information Systems

Recent advancements in Generative AI, such as Large Language Models (LLMs), have demonstrated remarkable success across various general tasks. Extensive studies have explored leveraging generative models in finance, but significant challenges persist. This half-day workshop explores potential approaches and research directions to address these challenges by equipping generative models with advanced Information Retrieval (IR) models. Specifically, this workshop seeks to provide a platform for discussing innovative ideas that facilitate the advancement of IR technology to enrich generative models in finance from four key perspectives: (i) financial IR techniques (ii) financial IR benchmarking and evaluation (iii) financial systems and agents/assistants (iv) …

Serving Others Using Generative Ai, Kenneth C. Arnold

University Faculty Publications and Creative Works

Ken Arnold, computer science professor at Calvin University, explores the idea of use Generative AI as a tool to help us serve others.

Llmscan: Causal Scan For Llm Misbehavior Detection, Mengdi Zhang, Kai Kiat Goh, Peixin Zhang, Jun Sun, Lin Xin Rose, Hongyu Zhang

Research Collection School Of Computing and Information Systems

Despite the success of Large Language Models (LLMs) across various fields, their potential to generate untruthful and harmful responses poses significant risks, particularly in critical applications. This highlights the urgent need for systematic methods to detect and prevent such misbehavior. While existing approaches target specific issues such as harmful responses, this work introduces LLMSCAN, an innovative LLM monitoring technique based on causality analysis, offering a comprehensive solution. LLMSCAN systematically monitors the inner workings of an LLM through the lens of causal inference, operating on the premise that the LLM’s ‘brain’ behaves differently when generating harmful or untruthful responses. By analyzing …

Hps: Hard Preference Sampling For Human Preference Alignment, Xiandong Zou, Wanyu Lin, Yuchen Li, Pan Zhou

Research Collection School Of Computing and Information Systems

Aligning Large Language Model (LLM) responses with human preferences is vital for building safe and controllable AI systems. While preference optimization methods based on PlackettLuce (PL) and Bradley-Terry (BT) models have shown promise, they face challenges such as poor handling of harmful content, inefficient use of dispreferred responses, and, specifically for PL, high computational costs. To address these issues, we propose Hard Preference Sampling (HPS), a novel framework for robust and efficient human preference alignment. HPS introduces a training loss that prioritizes the most preferred response while rejecting all dispreferred and harmful ones. It emphasizes “hard” dispreferred responses — those …

Repairing Adversarial Texts Through Perturbation, Guoliang Dong, Jingyi Wang, Jun Sun, Sudipta Chattopadhyay, Xinyu Wang, Ting Dai, Jie Shi, Jin Song Dong

Research Collection School Of Computing and Information Systems

It is known that neural networks are subject to attacks through adversarial perturbations. Worse yet, such attacks are impossible to eliminate, i.e., the adversarial perturbation is still possible after applying mitigation methods such as adversarial training. Multiple approaches have been developed to detect and reject such adversarial inputs. Rejecting suspicious inputs however may not be always feasible or ideal. First, normal inputs may be rejected due to false alarms generated by the detection algorithm. Second, denial-of-service attacks may be conducted by feeding such systems with adversarial inputs. To address this, in this work, we focus on the text domain and …

Sepsis: I Can Catch Your Lies – A New Paradigm For Deception Detection, Anku Rani, Dwip Dalal, Shreya Gautam, Pankaj Gupta, Vinija Jain, Aman Chadha, Amitava Das, Amit P. Sheth

Publications

Deception is the intentional practice of twisting information. It is a nuanced societal practice deeply intertwined with human societal evolution, characterized by a multitude of facets. This research explores the problem of deception through the lens of psychology, employing a framework that categorizes deception into three forms: lies of omission, lies of commission, and lies of influence. The primary focus of this study is specifically on investigating only lies of omission. We propose a novel framework for deception detection leveraging NLP techniques. We curated an annotated dataset of 876,784 samples by amalgamating a popular large-scale fake news dataset and scraped …

Mtu-Llm: Llm-Based Multi-Robot Task Allocation And Path Planning For Heterogeneous Robots In Search And Rescue Operations, Kaushik Kannan, Jungyun Bae

Michigan Tech Publications

Urban Search and Rescue operations after natural disasters involve locating and assisting victims in hazardous environments, which is challenging. Classical Multi-Robot Task Allocation (MRTA) and path planning approaches have been used to deploy heterogeneous robot teams in unsafe areas. However, existing methods often lack focus on workload balance and requirement fulfillment and struggle to generalize across different scenarios. To address these challenges, we propose Multi-robot Task allocation Utilizing LLMs (MTU-LLM), a framework designed to reduce the development time for task allocation and path planning approaches, enabling faster robot deployment. The framework uses an LLM-based “prompt engineering” approach that generates task …

Breaking The Reasoning Barrier: A Survey On Llm Complex Reasoning Through The Lens Of Self-Evolution, Tao He, Hao Li, Jingchang Chen, Runxuan Liu, Yixin Cao, Lizi Liao, Zihao Zheng, Zheng Chu, Jiafeng Liang, Ming Liu, Bing Qin

Research Collection School Of Computing and Information Systems

The release of OpenAI’s O1 and subsequent projects like DeepSeek R1 has significantly advanced research on complex reasoning in LLMs. This paper systematically analyzes existing reasoning studies from the perspective of self-evolution, structured into three components: data evolution, model evolution, and self-evolution. Data evolution explores methods to generate higher-quality reasoning training data. Model evolution focuses on training strategies to boost reasoning capabilities. Self-evolution research autonomous system evolution via iterating cycles of data and model evolution. We further discuss the scaling law of self-evolution and analyze representative O1-like works through this lens. By summarizing advanced methods and outlining future directions, this …

Simulating Before Planning: Constructing Intrinsic User World Model For User-Tailored Dialogue Policy Planning, Tao He, Lizi Liao, Ming Liu, Bing Qin

Research Collection School Of Computing and Information Systems

Recent advancements in dialogue policy planning have focused on optimizing system agent policies to achieve predefined goals, emphasizing strategy design, trajectory acquisition, and training efficiency. However, these approaches often overlook the critical role of user characteristics, which are essential in real-world scenarios like conversational search and recommendation, where interactions must adapt to individual user traits such as personality, preferences, and goals. To address this gap, we conduct a comprehensive study using task-specific user personas to evaluate dialogue policy planning under diverse user behaviors. Our analysis, based on these user profiles, reveals significant shortcomings in existing approaches, underscoring the necessity for …

Cracking Aegis: An Adversarial Llm-Based Game For Raising Awareness Of Vulnerabilities In Privacy Protection, Jiaying Fu, Yiyang Lu, Zehua Yang, Fiona Fui-Hoon Nah, Ray Lc

Research Collection School Of Computing and Information Systems

Traditional methods for raising awareness of privacy protection often fail to engage users or provide hands-on insights into how privacy vulnerabilities are exploited. To address this, we incorporate an adversarial mechanic in the design of the dialogue-based serious game Cracking Aegis. Leveraging LLMs to simulate natural interactions, the game challenges players to impersonate characters and extract sensitive information from an AI agent, Aegis. A user study (n=22) revealed that players employed diverse deceptive linguistic strategies, including storytelling and emotional rapport, to manipulate Aegis. After playing, players reported connecting in-game scenarios with real-world privacy vulnerabilities, such as phishing and impersonation, and …

Do Supplier Ceo's National Cultural Origins Affect Supplier-Customer Relationships?, Peng Liang, Hasan Cavusoglu, Nan Hu

Research Collection School Of Computing and Information Systems

This study investigates how the national cultural origins of supplier chief executive officers (CEOs), as characterized by Hofstede’s cross-cultural dimensions, influence the duration of supplier–customer relationships. By analyzing the cultural origins of supplier CEOs from 20 countries over a 25-year period, we find that supplier CEOs with high long-term orientation (LTO) and high uncertainty avoidance (UNA) are associated with longer lasting supplier–customer relationships, while those with high individualism (IND) are associated with shorter relationship durations. These findings are robust to several alternative explanations of customer and supplier CEO variables. To address potential endogeneity—specifically, the concern that CEOs with certain cultural …

Position: Trustworthy Ai Agents Require The Integration Of Large Language Models And Formal Methods, Yedi Zhang, Yufan Cai, Xinyue Zuo, Xiaokun Luan, Kailong Wang, Zhe Hou, Yifan Zhang, Zhiyuan Wei, Meng Sun, Jun Sun, Jing Sun, Jin Song Dong

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) have emerged as a transformative AI paradigm, profoundly influencing broad aspects of daily life. Despite their remarkable performance, LLMs exhibit a fundamental limitation: hallucination—the tendency to produce misleading outputs that appear plausible. This inherent unreliability poses significant risks, particularly in high-stakes domains where trustworthiness is essential. On the other hand, Formal Methods (FMs), which share foundations with symbolic AI, provide mathematically rigorous techniques for modeling, specifying, reasoning, and verifying the correctness of systems. These methods have been widely employed in mission-critical domains such as aerospace, defense, and cybersecurity. However, the broader adoption of FMs remains constrained …

Query Understanding In Llm-Based Conversational Information Seeking, Yifei Yuan, Zahra Abbasiantaeb, Mohammad Aliannejadi, Yang Deng

Research Collection School Of Computing and Information Systems

Query understanding in CIS involves accurately interpreting user intent through context-aware interactions. This includes resolving ambiguities, refining queries, and adapting to evolving information needs. LLM enhance this process by interpreting nuanced language and adapting dynamically, improving the relevance and precision of search results in real-time. In this tutorial, we explore advanced techniques to enhance query understanding in LLM-based CIS systems. We delve into LLM-driven methods for developing robust evaluation metrics to assess query understanding quality in multi-turn interactions, strategies for building more interactive systems, and applications like proactive query management and query reformulation. We also discuss key challenges in integrating …

A Mixed-Curvature Based Pre-Training Paradigm For Multi-Task Vehicle Routing Solver, Suyu Liu, Zhiguang Cao, Shanshan Feng, Yew-Soon Ong

Research Collection School Of Computing and Information Systems

Solving various types of vehicle routing problems (VRPs) using a unified neural solver has garnered significant attentions in recent years. Despite their effectiveness, existing neural multi-task solvers often fail to account for the geometric structures inherent in different tasks, which may result in suboptimal performance. To address this limitation, we propose a curvature-aware pre-training framework. Specifically, we leverage mixed-curvature spaces during the feature fusion stage, encouraging the model to capture the underlying geometric properties of each instance. Through extensive experiments, we evaluate the proposed pre-training strategy on existing neural multi-task solvers across a variety of testing scenarios. The results demonstrate …

Sparse-To-Dense: A Free Lunch For Lossless Acceleration Of Video Understanding In Llms, Xuan Zhang, Cunxiao Du, Sicheng Yu, Jiawei Wu, Fengzhuo Zhang, Wei Gao, Qian Liu

Research Collection School Of Computing and Information Systems

Due to the auto-regressive nature of current video large language models (Video-LLMs), the inference latency increases as the input sequence length grows, posing challenges for the efficient processing of video sequences that are usually very long. We observe that during decoding, the attention scores of most tokens in Video-LLMs tend to be sparse and concentrated, with only certain tokens requiring comprehensive full attention. Based on this insight, we introduce Sparse-to-Dense (StD), a novel decoding strategy that integrates two distinct modules: one leveraging sparse top-K attention and the other employing dense full attention. These modules collaborate to accelerate Video-LLMs without loss. …

Retrieval Augmented Generation For Dynamic Graph Modeling, Yuxia Wu, Lizi Liao, Yuan Fang

Research Collection School Of Computing and Information Systems

Modeling dynamic graphs, such as those found in social networks, recommendation systems, and e-commerce platforms, is crucial for capturing evolving relationships and delivering relevant insights over time. Traditional approaches primarily rely on graph neural networks with temporal components or sequence generation models, which often focus narrowly on the historical context of target nodes. This limitation restricts the ability to adapt to new and emerging patterns in dynamic graphs. To address this challenge, we propose a novel framework, Retrieval-Augmented Generation for Dy namic Graph modeling (RAG4DyG ), which enhances dynamic graph predictions by incorporating contextually and temporally relevant examples from broader …