Computer Sciences Commons^™

Evaluating And Mitigating Linguistic Discrimination In Large Language Models: Perspectives On Safety Equity And Knowledge Equity, Guoliang Dong, Haoyu Wang, Jun Sun, Xinyu Wang

Research Collection School Of Computing and Information Systems

By training on text in various languages, large language models (LLMs) typically possess multilingual support and demonstrate remarkable capabilities in solving tasks described in different languages. However, LLMs can exhibit linguistic discrimination due to the uneven distribution of training data across languages. That is, LLMs are hard to keep the consistency of responses when faced with the same task but depicted in different languages. In this study, we first explore the consistency in the LLMs’ outputs responding to queries in various languages from two aspects: safety and quality. We conduct this analysis with two datasets (AdvBench and NQ) based on …

Mpo: Multilingual Safety Alignment Via Reward Gap Optimization, Weixiang Zhao, Yulin Hu, Yang Deng, Tongtong Wu, Wenxuan Zhang, Jiahe Guo, An Zhang, Yanyan Zhao, Bing Qin, Tat-Seng Chua, Ting Liu

Research Collection School Of Computing and Information Systems

Large language models (LLMs) have become increasingly central to AI applications worldwide, necessitating robust multilingual safety alignment to ensure secure deployment across diverse linguistic contexts. Existing preference learning methods for safety alignment, such as RLHF and DPO, are primarily monolingual and struggle with noisy multilingual data. To address these limitations, we introduce Multilingual reward gaP Optimization (MPO), a novel approach that leverages the well-aligned safety capabilities of the dominant language (e.g., English) to improve safety alignment across multiple languages. MPO directly minimizes the reward gap difference between the dominant language and target languages, effectively transferring safety capabilities while preserving the …

Leveraging Reviewer Experience In Code Review Comment Generation, Hong Yi Lin, Patanamon Thongtanunam, Christoph Treude, Michael W. Godfrey, Chunhua Liu, Wachiraphan Charoenwet

Research Collection School Of Computing and Information Systems

Modern code review is a ubiquitous software quality assurance process aimed at identifying and resolving potential issues (e.g., functional, evolvability) within newly written code. Despite its effectiveness, the process demands large amounts of effort from the human reviewers involved. To help alleviate this workload, researchers have trained various deep learning based language models to imitate human reviewers in providing natural language code reviews for submitted code. Formally, this automation task is known as code review comment generation. Prior work has demonstrated improvements in code review comment generation by leveraging machine learning techniques and neural models, such as transfer learning and …

Quantizing Text-Attributed Graphs For Semantic-Structural Integration, Jianyuan Bo, Hao Wu, Yuan Fang

Research Collection School Of Computing and Information Systems

Text-attributed graphs (TAGs) have emerged as a powerful representation for modeling complex relationships across diverse domains. With the rise of large language models (LLMs), there is growing interest in leveraging their capabilities for graph learning. However, current approaches face significant challenges in embedding structural information into LLM-compatible formats, requiring either computationally expensive alignment mechanisms or manual graph verbalization techniques that often lose critical structural details. Moreover, these methods typically require labeled data from source domains for effective transfer learning, significantly constraining their adaptability. We propose STAG, a novel self-supervised framework that directly quantizes graph structural information into discrete tokens using …

Causalabstain: Enhancing Multilingual Llms With Causal Reasoning For Trustworthy Abstention, Yuxi Sun, Aoqi Zuo, Wei Gao, Jing Ma

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) often exhibit knowledge disparities across languages. Encouraging LLMs to abstain when faced with knowledge gaps is a promising strategy to reduce hallucinations in multilingual settings. Current abstention strategies for multilingual scenarios primarily rely on generating feedback in various languages using LLMs and performing self-reflection. However, these methods can be adversely impacted by inaccuracies and biases in the generated feedback. To address this, from a causal perspective, we introduce CausalAbstain, a method that helps LLMs determine whether to utilize multiple generated feedback responses and how to identify the most useful ones. Extensive experiments demonstrate that CausalAbstain effectively …

Computational Fact-Checking With Limited Resources, Fengzhu Zeng

Dissertations and Theses Collection (Open Access)

The rapid dissemination of information through online platforms has sparked widespread concern about the propagation of misinformation. Manual fact-checking by pro- fessional fact-checkers is time-consuming and lacks scalability to address the vast volume of daily information. Consequently, computational fact-checking, driven by automated techniques in natural language processing (NLP), has garnered interest as
a potential solution. However, computational fact-checking faces critical challenges limited resources, particularly due to the issues of data scarcity and computing resource constraints. One key challenge is data scarcity, which arises from the constant generation of new information and emerging events on social media. This scarcity manifests in …

Evometric: An Interactive Framework For Scalable Visual Analytics Of Time Series Data With Dynamic Changes., Jiahang Huang

Electronic Theses and Dissertations

In today's data-intensive landscape, rapid advances in digital sensing and recording technologies have enabled the acquisition of high-resolution multimodal time series data, capturing intricate real-world dynamics across various domains such as healthcare, behavioral science, and environmental monitoring. However, the complexity and scale of these datasets present significant analytical challenges, particularly in understanding dynamic changes at both individual and cohort levels. This dissertation introduces EvoMetric, a novel visual analytics framework designed to support scalable exploration and analysis of large-scale multimodal time series data with dynamic changes. EvoMetric seamlessly integrates individual-level temporal dynamics with population-level comparative insights, enabling users to visually …

Ai-Enhanced Structured Literacy Intervention For Secondary Students: A Case Study Of Science Of Reading, Jennifer Bird

Teaching & Learning Faculty Publications

This study examines the effectiveness of Lexia PowerUp, an AI-powered literacy program, for sixth-grade students requiring Tier 3 reading intervention. Seven sixth-grade students (six boys, one girl; five African American, two Caucasian; all qualifying for free/reduced lunch) participated in a six-month intervention combining 50 minutes of daily small-group instruction with individualized Lexia PowerUp usage. Researchers measured progress through Achieve 3000 Lexile assessments and Lexia PowerUp performance data across three skill strands: Word Study, Grammar, and Comprehension. All participants demonstrated Lexile level improvements from beginning-of-year to mid-year assessments, though students remained below sixth-grade benchmarks (925-1070L). Analysis of Lexia PowerUp progression showed …

Training Robot Swarms For Adaptive Foraging In Environments With Obstacles, Pigar Biteng, Tameem Uz Zaman, Qi Lu

Computer Science Faculty Publications

In this work, we train adaptive and efficient foraging strategies for robot swarms in a large, unmapped search space with multiple randomly distributed box obstacles using the penalty-reward based NeuroEvolution of Augmented Topologies (NEAT), P-NeatFA. This model enables efficient multi-robot foraging behavior and obstacle avoidance by rewarding effective actions and penalizing inefficient ones, thereby minimizing redundant exploration and outperforming traditional stochastic foraging algorithms. We optimize foraging strategies and search patterns in robot swarms by training models that maximize cumulative rewards in three types of resource distribution environments. The evaluation focuses on the number of resources collected within a fixed time …

Autoregressive Temporal Modeling For Advanced Tracking-By-Diffusion, Pha Nguyen, Rishi Madhok, Bhiksha Raj, Khoa Luu

Electrical Engineering and Computer Science Faculty Publications and Presentations

Object tracking is a widely studied computer vision task with video and instance analysis applications. While paradigms such as tracking-by-regression,-detection,-attention have advanced the field, generative modeling offers new potential. Although some studies explore the generative process in instance-based understanding tasks, they rely on prediction refinement in the coordinate space rather than the visual domain. Instead, this paper presents Tracking-by-Diffusion, a novel paradigm for object tracking in video, leveraging visual generative models via the perspective of autoregressive models. This paradigm demonstrates broad applicability across point, box, and mask modalities while uniquely enabling textual guidance. We present DIFTracker, a framework that utilizes …

Revolutionizing Digital Privacy Education For Older Adults: Enhanced Interventions And Ai-Assisted Learning Strategies, Heba Aly

All Dissertations

As older adults increasingly engage with digital platforms, they face unique privacy risks stemming from limited digital literacy, reduced trust in AI technologies, and constrained access—especially in rural or underserved communities. While digital tools offer benefits like social connection and information access, current privacy education efforts often neglect the needs of older adults. This dissertation addresses this gap by developing, testing, and refining digital privacy education interventions tailored for older adults, with a focus on trust, personalization, and AI-assisted learning.

Study 1 evaluates multiple instructional modalities across age groups, revealing older adults prefer structured videos and interactive tutorials, while younger …

Towards Securing Ai Systems: Investigating Threats In Multimodal Autonomous Driving & Rag Systems, Saket Sanjeev Chaturvedi

All Dissertations

Artificial Intelligence (AI) systems have become central to high-stakes applications such as autonomous driving and language-based decision support. As their deployment accelerates, ensuring the security and trustworthiness of these systems becomes paramount. Among the most stealthy and potent threats are backdoor attacks, where models behave as expected under normal conditions but exhibit malicious behavior when triggered by specific inputs, either digital or physical.

This thesis investigates novel backdoor and adversarial vulnerabilities across two emerging classes of AI architectures: (1) multimodal 3D object detection systems that fuse LiDAR and camera data, and (2) Retrieval-Augmented Generation (RAG) systems that pair large language …

Three Essays In The Economics Of Disagreements: Incentives, Measurement And Persistence, Mustafa W. Alam

All Dissertations

This dissertation presents three chapters that contribute to the study of economic forces shaping disagreements in society.

In Chapter 1, I demonstrate that political polarization can intensify due to innovations in the information market even if a population's ideological distribution is fixed. Viewership-maximizing news firms cater to a diverse audience who assess source accuracy using noisy private signals that vary in precision and ideological bias. If better-informed consumers disproportionately migrate to newer platforms for news (e.g., the Internet), traditional media firms increase news slant to appeal more to less-informed partisans on both sides of the ideological spectrum. This leads to …

L2m2: A Hierarchical Framework Integrating Large Language Model And Multi‑Agent Reinforcement Learning, Minghong Geng, Shubham Pateria, Budhitama Subagdja, Lin Li, Xin Zhao, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Multi-agent reinforcement learning (MARL) has demonstrated remarkable success in collaborative tasks, yet faces significant challenges in scaling to complex scenarios requiring sustained planning and coordination across long horizons. While hierarchical approaches help decompose these tasks, they typically rely on hand-crafted subtasks and domain-specific knowledge, limiting their generalizability. We present L2M2, a novel hierarchical framework that leverages large language models (LLMs) for high-level strategic planning and MARL for low-level execution. L2M2 enables zero-shot planning that supports both end-to-end training and direct integration with pre-trained MARL models. Experiments in the VMAS environment demonstrate that L2M2's LLM-guided MARL achieves superior performance while requiring …

Collisionrepair: First‑Aid And Automated Patching For Storage Collision Vulnerabilities In Smart Contracts, Yu Pan, Wanjing Han, Yue Duan, Mu Zhang

Research Collection School Of Computing and Information Systems

Storage collision vulnerabilities, a significant security risk in upgradeable smart contracts, often arise when a user-facing proxy contract and a backend logic contract share storage space. While static analysis techniques can detect such issues, they often over-approximate program states, leading to false positives and requiring developers to manually verify each issue, giving attackers time to exploit any overlooked vulnerabilities. To address this, we propose COLLISIONREPAIR, an automated patching technique for mitigating storage collision risks. COLLISIONREPAIR monitors storage access sequences between proxy and logic contracts by defining an "ownership" property for storage locations. It then replays historical transactions to recover existing …

L3net: Localized And Layered Reparameterization For Incremental Learning, Xuandi Luo, Huaidong Zhang, Yi Xie, Hongrui Zhang, Xuemiao Xu, Shengfeng He

Research Collection School Of Computing and Information Systems

Model-based class incremental learning (CIL) methods aim to address the challenge of catastrophic forgetting by retaining certain parameters and expanding the model architecture. However, retaining too many parameters can lead to an overly complex model, increasing inference overhead. Additionally, compressing these parameters to reduce the model size can result in performance degradation. To tackle these challenges, we propose a novel three-stage CIL framework called Localized and Layered Reparameterization for Incremental Learning (L3Net). The rationale behind our approach is to balance model complexity and performance by selectively expanding and optimizing critical components. Specifically, the framework introduces a Localized Dual-path Expansion structure, …

Consistent Client Simulation For Motivational Interviewing-Based Counseling, Yizhe Yang, Palakorn Achananuparp, Heyan Huang, Jing Jiang, Nicholas Gabriel Lim, Cameron Shi Ern Tan, Phey Ling Kit, Jenny Xiuhui Giam, John Pinto, Ee-Peng Lim

Research Collection School Of Computing and Information Systems

Simulating human clients in mental health counseling is crucial for training and evaluating counselors (both human or simulated) in a scalable manner. Nevertheless, past research on client simulation did not focus on complex conversation tasks such as mental health counseling. In these tasks, the challenge is to ensure that the client’s actions (i.e., interactions with the counselor) are consistent with with its stipulated profiles and negative behavior settings. In this paper, we propose a novel framework that supports consistent client simulation for mental health counseling. Our framework tracks the mental state of a simulated client, controls its state transitions, and …

Assessing The Robustness Of Test Selection Methods For Deep Neural Networks, Qiang Hu, Yuejun Guo, Xiaofei Xie, Maxime Cordy, Wei Ma, Mike Papadakis, Lei Ma, Yves Le Traon

Research Collection School Of Computing and Information Systems

Regularly testing deep learning-powered systems on newly collected data is critical to ensure their reliability, robustness, and efficacy in real-world applications. This process is demanding due to the significant time and human effort required for labeling new data. While test selection methods alleviate manual labor by labeling and evaluating only a subset of data while meeting testing criteria, we observe that such methods with reported promising results are simply evaluated, e.g., testing on original test data. The question arises: are they always reliable? In this article, we explore when and to what extent test selection methods fail. First, we identify …

Think Both Ways: Teacher-Student Bidirectional Reasoning Enhances Mcq Generation And Distractor Quality, Yimiao Qiu, Yang Deng, Quanming Yao, Zhimeng Zhang, Zhiang Dong, Chang Yao, Jingyuan Chen

Research Collection School Of Computing and Information Systems

Generating high-quality Multiple Choice Questions (MCQs) remains challenging for educational tools due to the need for contextual relevance and plausible distractors. Existing methods still struggle with these dual requirements, leading to questions that lack depth and distractors that are either too obvious or irrelevant. In this paper, we propose BiFlow, a novel framework that integrates bidirectional reasoning perspectives: teacher reasoning generates contextually relevant questions and plausible distractors, while student reasoning evaluates question clarity and the misleading nature of the distractors. To further enhance reasoning, we introduce PathFinder, a mechanism that employs breadth-first search and Chainof-Thought (CoT) strategies to explore diverse …

Non-Homophilic Graph Pre-Training And Prompt Learning, Xingtong Yu, Jie Zhang, Yuan Fang, Renhe Jiang

Research Collection School Of Computing and Information Systems

Graphs are ubiquitous for modeling complex relationships between objects across various fields. Graph neural networks (GNNs) have become a mainstream technique for graph-based applications, but their performance heavily relies on abundant labeled data. To reduce labeling requirement, pre-training and prompt learning has become a popular alternative. However, most existing prompt methods do not distinguish between homophilic and heterophilic characteristics in graphs. In particular, many real-world graphs are non-homophilic-neither strictly nor uniformly homophilic-as they exhibit varying homophilic and heterophilic patterns across graphs and nodes. In this paper, we propose ProNoG, a novel pre-training and prompt learning framework for such non-homophilic graphs. …

Optimizing Group Utility In Itinerary Planning: A Strategic And Crowd-Aware Approach, Junhua Liu, Aldy Gunawan, Kristin L. Wood, Kwan Hui Lim

Research Collection School Of Computing and Information Systems

Itinerary recommendation is a complex sequence prediction problem with numerous practical applications. The task becomes significantly more challenging when optimizing multiple factors simultaneously, such as user queuing times, crowd levels, attraction popularity, walking durations, and operating hours. These factors, combined with the dynamic and unpredictable nature of visitor flow, introduce substantial complexities, particularly when accounting for collective user behavior. Existing solutions often adopt a single-user perspective, overlooking critical challenges arising from natural crowd dynamics. For example, the Selfish Routing problem illustrates how individual decision-making can lead to suboptimal outcomes for the group as a whole. To address these challenges, we …

Use Of A Preliminary Artificial Intelligence-Based Laryngeal Cancer Screening Framework For Low-Resource Settings: Development And Validation Study, Shao Wei Sean Lam, Min Hun Lee, Michael Dorosan, Samuel Altonji, Hiang Khoon Tan, Walter T. Lee

Research Collection School Of Computing and Information Systems

Background: Early-stage diagnosis of laryngeal cancer significantly improves patient survival and quality of life. However, the scarcity of specialists in low-resource settings hinders the timely review of flexible nasopharyngoscopy (FNS) videos, which are essential for accurate triage of at-risk patients.Objective: We introduce a preliminary AI-based screening framework to address this challenge for the triaging of at-risk patients in low-resource settings. This formative research addresses multiple challenges common in high-dimensional FNS videos: (1) selecting clear, informative images; (2) deriving regions within frames that show an anatomical landmark of interest; and (3) classifying patients into referral grades based on the FNS video …

The 6th International Workshop On Talent And Management Computing (Tmc 2025), Hengshu Zhu, Yong Ge, Hui Xiong, Ee-Peng Lim

Research Collection School Of Computing and Information Systems

In today's competitive and fast-evolving business environment, it is a critical time for organizations to rethink how to deal with talent and management-related tasks in a quantitative manner. Indeed, thanks to the era of big data, the availability of large-scale talent data provides unparalleled opportunities for leaders to deliver intelligence for effective management for organizations. In the past few years, talent and management computing have increasingly attracted attention from KDD communities, and a number of research/applied data science efforts have been devoted. To this end, the purpose of this workshop, i.e., the 6th International Workshop on Talent and Management Computing …

Inference-Time Gaze Refinement For Micro-Expression Recognition: Enhancing Event-Based Eye Tracking With Motion-Aware Post-Processing, Panahetipola Mudiyanselage Nuwan Bandara, Thivya Kandappu, Archan Misra

Research Collection School Of Computing and Information Systems

Event-based eye tracking holds significant promise for fine-grained cognitive state inference, offering high temporal resolution and robustness to motion artifacts, critical features for decoding subtle mental states such as attention, confusion, or fatigue. In this work, we introduce a model-agnostic, inference-time refinement framework designed to enhance the output of existing event-based gaze estimation models without modifying their architecture or requiring retraining. Our method comprises two key post-processing modules: (i) Motion-Aware Median Filtering, which suppresses blink-induced spikes while preserving natural gaze dynamics, and (ii) Optical Flow-Based Local Refinement, which aligns gaze predictions with cumulative event motion to reduce spatial jitter and …

Prism: To Fortify Widget Based User‑App Data Exchanges Using Android Virtualization Framework, Yingtat Ng, Zhe Chen, Haiqing Qiu, Xuhua Ding

Research Collection School Of Computing and Information Systems

We present Prism, an UI hardening technique for an Android app to safeguard its widgets against a corrupted kernel. Prism ensures secure interface rendering and allows for visual authentication, which developers could use to enable user intent confidentiality protection. Our design leverages the recent Android Virtualization Framework with minimal changes to the existing UI framework and graphics subsystem. It is much easier to deploy and use Prism on Android phones than TrustZone-based secure UI schemes, because the apps are not admitted to the Secure World and retain their full rights to manage and control their own interfaces. We have implemented …

Oblivious Digital Tokens, Mihael Liskij, Xuhua Ding, Gene Tsudik, David A. Basin

Research Collection School Of Computing and Information Systems

A computing device typically identifies itself by exhibiting unique measurable behavior or by proving its knowledge of a secret. In both cases, the identifying device must reveal information to a verifier. Considerable research has focused on protecting identifying entities (provers) and reducing the amount of leaked data. However, little has been done to conceal the fact that the verification occurred.We show how this problem naturally arises in the context of digital emblems, which were recently proposed by the International Committee of the Red Cross to protect digital resources during cyber-conflicts. To address this new and important open problem, we define …

Trust In Healthcare Ai Can’T Just Be Designed – It Must Be Felt By Clinicians And Patients, Adriana Banozic-Tang, Heng Wang

Research Collection Yong Pung How School Of Law

Trust in healthcare AI currently over-relies on system design, not lived medical realities.Continuous feedback loops are necessary to embed trust in healthcare AI that is responsive to clinician and patient needs.Initiatives in South-East Asia show how trust in technology can be extended from policy to practice.

Ai In The Judiciary: The Singapore Case, Nydia Remolina Leon

Research Collection Yong Pung How School Of Law

This paper examines the integration of Artificial Intelligence (AI) within the judicial system of Singapore. Singapore's judiciary has embraced AI not as a tool for adjudication, but as an augmentative instrument for legal research, procedural efficiency, and access to justice. It provides a detailed account of AI use cases in the courts, including case summarization, evidence review, assistance for selfrepresented litigants, and tools like the Divorce Assets Informative Division Estimator. The discussion then turns to the legal profession, exploring how law firms in Singapore are adopting AI technologies. The paper also addresses how AI implementation in the judicial system is …

Fact-Audit: An Adaptive Multi-Agent Framework For Dynamic Fact-Checking Evaluation Of Large Language Models, Hongzhan Lin, Yang Deng, Yuxuan Gu, Wenxuan Zhang, Jing Ma, See-Kiong Ng, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) have significantly advanced the fact-checking studies. However, existing automated fact-checking evaluation methods rely on static datasets and classification metrics, which fail to automatically evaluate the justification production and uncover the nuanced limitations of LLMs in fact-checking. In this work, we introduce FACT-AUDIT, an agent-driven framework that adaptively and dynamically assesses LLMs’ fact-checking capabilities. Leveraging importance sampling principles and multi-agent collaboration, FACT-AUDIT generates adaptive and scalable datasets, performs iterative model-centric evaluations, and updates assessments based on model-specific responses. By incorporating justification production alongside verdict prediction, this framework provides a comprehensive and evolving audit of LLMs’ factual reasoning …

Analysis Of Extended Producer Responsibility In E-Waste Management: Policy Drivers And Challenges In Singapore, Aldy Gunawan, Aidan Marc Wong, Tasaporn Visawameteekul, Minh Phuong Huynh, Linh Chi Tran

Research Collection School Of Computing and Information Systems

This paper examines the role of the Extended Producer Responsibility (EPR) scheme in electronic waste (e-waste) management in Singapore. It investigates the policy drivers and challenges of e-waste management, using data from an online survey to explore the attitudes and behaviors of young consumers, with a particular focus on youth. We employ the Theory of Reasoned Action (TRA) and the Theory of Planned Behavior (TPB) frameworks to develop a model that examines the relationships among attitudes, perceived norms, awareness, and perceived convenience in relation to EPR awareness and perception. The findings highlight the need for customized policies tailored to different …