Digital Commons Network^™

Mease: Multi-Agent Episodic Action Sequence Explanation, Phyo Wai Khaing, Minghong Geng, Shubham Pateria, Budhitama Subagdja, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Multi-agent reinforcement learning (MARL) achieves remarkable performance in complex coordination tasks, yet interpreting the emergent behaviors of trained agents remains a fundamental challenge. Most current explainability methods focus on individual agent decisions, overlooking the critical interplay of joint strategiesand temporal coordination patterns that define successful multi-agent policies. We present MEASE (Multi-agent Episodic Action Sequence Explanation), a novel explainable MARL (XMARL) framework that explains trained MARL policies as human-interpretable emergent cooperative joint behaviors. MEASE employs a cognition-inspired episodic memory model to learn spatio-temporal multi-agent interaction patterns, coupled with abstraction algorithms that identify significant cooperative agent behaviors. We evaluate MEASE on diverse …

Collaborative Practices And Tool Utilization In Software Development Projects: A Student Perspective, Yi Meng Lau, Muhammad Syahmi Bin Abbas, Lingxiao Jiang

Research Collection School Of Computing and Information Systems

Software development is a collaborative activity that depends on effective teamwork, shared understanding, and coordinated use of development practices and tools. While these aspects are well studied in professional environments, they are less frequently examined within software engineering education. This study investigates how students collaborate in group projects, focusing on collaborative practices, tool usage, and their perceptions of software quality. We conducted a quantitative post-project survey with 143 second-year undergraduate students enrolled in a software development course. The results show that students actively share information and often establish team norms to support coordination and collaboration. However, students face challenges in …

Enhancing Action And Ingredient Modeling For Semantically Grounded Recipe Generation, Guoshan Liu, Bin Zhu, Yian Li, Jingjing Chen, Chong-Wah Ngo, Yu-Gang Jiang

Research Collection School Of Computing and Information Systems

Recent advances in Multimodal Large Language Models (MLMMs) have enabled recipe generation from food images, yet outputs often contain semantically incorrect actions or ingredients despite high lexical scores (e.g., BLEU, ROUGE). To address this gap, we propose a semantically grounded framework that predicts and validates actions and ingredients as internal context for instruction generation. Our two-stage pipeline combines supervised fine-tuning (SFT) with reinforcement fine-tuning (RFT): SFT builds foundational accuracy using an Action-Reasoning dataset and ingredient corpus, while RFT employs frequency-aware rewards to improve long-tail action prediction and ingredient generalization. A Semantic Confidence Scoring and Rectification (SCSR) module further filters and …

Relationship Between Change In Unrealized Gains Or Losses Of Afs Securities And Credit Spreads Of U.S. Life Insurance Companies, Kwang Liang Yeo

Dissertations and Theses Collection (Open Access)

This study evaluates the transmission of unrealized fair value accounting losses to credit default risk within the United States life insurance sector during a period of unprecedented monetary tightening. Integrating a pre-specified confirmatory econometric framework with a text-based exploratory analysis of regulatory filings, we examine how changes in Other Comprehensive Income (OCI) and benchmark interest rate dynamics shape market perceptions of insurer default probability. Our findings reveal that credit default swap markets do not price OCI variations contemporaneously; instead, they reflect a structural one-quarter pricing lag that aligns with the regulatory timeline of financial disclosures. Furthermore, the relationship between interest …

A Study On The Capital Market Valuation Effects Of Corporate Ai Adoption: An Empirical Analysis Of Manufacturing Firms In China's A-Share Market, Libin Ma

Dissertations and Theses Collection (Open Access)

This dissertation examines whether and how corporate adoption of artificial intelligence (AI) affects capital market valuation in manufacturing firms. As AI is increasingly embedded in R&D design, production organization, supply chain coordination, and operational management, its value implications depend not only on whether firms adopt AI, but also on how AI is deployed within business processes. Existing research has mainly focused on productivity, innovation, labor substitution, and digital transformation, while the enterprise value consequences of differentiated AI application strategies remain insufficiently examined.

Using Chinese A-share listed manufacturing firms from 2015 to 2024 as the sample, this study identifies firm-level AI …

From Suppliers’ Esg Performance To Corporate Bargaining Power With Customers: A Supply Chain Advantage Transmission Perspective, Minghao Du

Dissertations and Theses Collection (Open Access)

With the growing awareness of sustainable development, Environmental, Social, and Governance (ESG) practices has increasingly become a critical source for firms to meet stakeholder demands and gain new competitive advantages. In particular, under the backdrop of rising supply chain risks, stakeholder concerns are no longer confined to the ESG performance of individual firms; rather, they have expanded to encompass the entire supply chain, attaching equal importance to the ESG performance of both focal firms and their supply chain partners. With the continuous tightening of ESG regulatory requirements, customers may move beyond managing the ESG practices of first-tier suppliers to paying …

Research On The Impact Of Industrial-Financial Cooperation On Corporate R&D Investment, Jiayin Yin

Dissertations and Theses Collection (Open Access)

The national industrial-financial cooperation pilot policy represents a key strategic initiative to implement China's strategy of building a "financial powerhouse" and guide financial resources toward empowering high-quality development in the real economy. In December 2016, four central government departments in China jointly announced the first batch of national industrial-financial cooperation pilot cities. Leveraging this quasi-natural experiment, this study comprehensively evaluates the impact of the pilot policy on corporate R&D investment activities using a sample of Chinese A-share listed companies from 2014 to 2019.

Research findings indicate: (1) The industrial-financial cooperation pilot policy significantly increases enterprises' R&D investment levels. Specifically, compared …

Be Responsible In Your Answers! Monitoring Out-Of-Domain Behaviors In Domain-Specific Llms, Boquan Li, Chenzhe Lou, Zhe Ren, Peixin Zhang, Zirui Fu, Jun Sun, Yaowen Zheng

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) have accelerated the rapid development of chatbot web applications in various domains, such as coding, biomedicine and psychology. Compared to general LLMs like ChatGPT, domain-specific LLMs require a greater sense of responsibility. For instance, if a programming LLM casually answers medical or psychological questions, it not only misleads the public but also poses legal risks. This highlights new demands for monitoring and preventing such irresponsible behaviors. Existing efforts attempt to monitor LLMs from multiple aspects, such as lying, jailbreaks, and toxic content, while overlooking out-of-domain behaviors. In this work, we propose an innovative LLM domain monitoring …

Architecture-Agnostic Test-Time Adaptation Via Backprop-Free Embedding Alignment, Xiao Ma, Young D. Kwon, Pan Zhou, Dong Ma

PhD Student’s Publications Collection

Test-Time Adaptation (TTA) adapts a deployed model during online inference to mitigate the impact of domain shift. While achieving strong accuracy, most existing methods rely on backpropagation, which is memory and computation intensive, making them unsuitable for resource-constrained devices. Recent attempts to reduce this overhead often suffer from high latency or are tied to specific architectures such as ViT-only or CNN-only. In this work, we revisit domain shift from an embedding perspective. Our analysis reveals that domain shift induces three distinct structural changes in the embedding space: translation (mean shift), scaling (variance shift), and rotation (covariance shift). Based on this …

Measuring Research Impact At Singapore Management University, Pin Pin Yeo

Research Collection Library

Describes the support and services provided by SMU Libraries to the institution for research assessment.

Penforge: On-The-Fly Expert Agent Construction For Automated Penetration Testing, Huihui Huang, Jieke Shi, Junkai Chen, Ting Zhang, Yikun Li, Chengran Yang, Eng Lieh Ouh, Lwin Khin Shar, David Lo

Research Collection School Of Computing and Information Systems

Penetration testing is essential for identifying vulnerabilities in web applications before real adversaries can exploit them. Recent work has explored automating this process with Large Language Model (LLM)-powered agents, but existing approaches either rely on a single generic agent that struggles in complex scenarios or narrowly specialized agents that cannot adapt to diverse vulnerability types. We therefore introduce PenForge, a framework that dynamically constructs expert agents during testing rather than relying on those prepared beforehand. By integrating automated reconnaissance of potential attack surfaces with agents instantiated on the fly for context-aware exploitation, PenForge achieves a 30.0% exploit success rate (12/40) …

Finding Missing Input Validation In Tees Via Llm-Assisted Symbolic Execution, Chengyan Ma, Jieke Shi, Ruidong Han, Ye Liu, Yuqing Niu, David Lo

Research Collection School Of Computing and Information Systems

Trusted Execution Environments (TEEs) provide hardware-enforced isolation that protects sensitive code and data from untrusted software. Despite their strong security guarantees, analyzing TEE applications remains challenging due to the high cost and complexity of configuring complete TEE build and runtime environments, as well as the limited observability imposed by hardware isolation. This paper presents SymTEE, a novel large language model (LLM)-assisted symbolic execution framework for detecting missing input validation issues in TEE applications without requiring real TEE setups. SymTEE begins by leveraging Abstract Syntax Tree (AST) analysis to extract TEE code slices that may lack sufficient input validation, and then …

The Impact Reporting Playbook, Julian Tang

Asian Management Insights

Ditch the proverbial laundry list and build impactful reports with evidence of change that leaders, partners and the public can trust and act on.

When Performance Is Not Enough: Managing Ngos In An Age Of Digital Mistrust, Christianna Sirindah Parr

Asian Management Insights

As NGOs lack formal authority, their influence depends on legitimacy, which can be contested at scale online. Leaders must treat social media as a governance decision, with clear scope, speaking authority and response protocols.

The Determinants And Consequences Of Green Credit In Commercial Banks: A Dual-Pillar Perspective Of Asset Quality And Esg, Jianmei Xia

Dissertations and Theses Collection (Open Access)

As China’s “dual carbon” strategy advances, green credit has increasinglybecome a crucial financial instrument driving the green transition of China’s industrial structure. From a dual perspective of asset quality and ESGperformance, this paper systematically examines the determinants andconsequences of green credit practices among commercial banks. Employing a multiple-method approach that combines comparative case study, in-depthinterview, and large-sample empirical analysis, this study aims to address twocore questions: First, what are the key factors driving commercial banks toprovide green credit, and what is the logic behind their differentiated decision- making? Second, how does green credit affect banks’ financial and non-financial performance? This …

Generative Ai In Enterprises: Optimizing Applications With Large Language Models, Donghao Huang

Dissertations and Theses Collection (Open Access)

This dissertation investigates how to deploy Large Language Models (LLMs) effectively in enterprise settings, where accuracy, reliability, cost, privacy, and operational constraints often matter more than benchmark performance alone. Drawing on seventeen peer-reviewed publications (eleven published and six accepted for publication), the work develops and validates optimization strategies across three connected themes: retrieval-augmented generation (RAG), agentic AI for workflow automation, and deployment guidelines for real-world enterprise environments.

First, we study RAG optimization through systematic evaluation of open and proprietary models, highlighting conditions under which efficient open-weight models can match or exceed proprietary alternatives. To address a pervasive failure mode in …

Enhancing Market Acceptance Of Sports Rehabilitation Services: A Perspective Of Public Health Literacy And Information Acquisition, Na Li

Dissertations and Theses Collection (Open Access)

This article takes public health literacy as its point of departure, constructs a theoretical model of “Health Literacy–Health Information Acquisition–Market Acceptance”, and introduces rehabilitation experience and service accessibility as moderating variables, in order to systematically examine the formation pathways and underlying mechanisms of market acceptance in the sports rehabilitation service sector.

This study adopts a combination of questionnaire survey and empirical analysis, collecting data from public samples across multiple regions and conducting statistical tests. The findings indicate that, Firstly, public health literacy has a statistically significant positive influence on the market acceptance of sports rehabilitation services. As the level of …

The Impact Of Mortality Awareness On Task And Relationship Leadership Styles, Sang Yeol Lee

Dissertations and Theses Collection (Open Access)

In current organizational management practice, the question of how leaders can transform the universal awareness of human finitude into constructive leadership is an issue that embodies both profound humanistic concern and significant managerial implications. Based on transcendence management theory and leadership style theory, this study explores how mortality awareness affects leaders’ task orientation and relationship orientation styles, revealing its underlying mechanisms and boundary conditions. By conducting a three-wave, matched-pair questionnaire survey involving 203 enterprise managers and their subordinates, the theoretical hypotheses were tested. The findings are as follows: First, mortality awareness has a significant direct positive effect on both task-oriented …

Compliance Supervision And Firm Performance: An Empirical Study On Digital Transformation Of The Pharmaceutical Manufacturing Industry, Jun Ma

Dissertations and Theses Collection (Open Access)

The biopharmaceutical industry, characterized by stringent global regulatory oversight, faces a pivotal theoretical tension regarding the nexus between regulatory compliance and corporate performance. The traditional perspectives often frame compliance as an “operational burden”, while the emerging viewpoints posit it as a “strategic asset”. Grounded in an integrated framework of the Resource-Based View (RBV) and Dynamic Capability (DC), this study investigates the moderating role of digital transformation in the relationship between Good Manufacturing Practice (GMP) compliance and corporate performance.

Employing a mixed-method design, this paper conducted two complementary studies. Study 1 adopted a grounded theory based on interviews with nine industry …

Agentspec: Customizable Runtime Enforcement For Safe And Reliable Llm Agents, Haoyu Wang, Christopher M. Poskitt, Jun Sun

Research Collection School Of Computing and Information Systems

Agents built on LLMs are increasingly deployed across diverse domains, automating complex decision-making and task execution. However, their autonomy introduces safety risks, including security vulnerabilities, legal violations, and unintended harmful actions. Existing mitigation methods, such as model-based safeguards and early enforcement strategies, fall short in robustness, interpretability, and adaptability. To address these challenges, we propose AgentSpec, a lightweight domain-specific language for specifying and enforcing runtime constraints on LLM agents. With AgentSpec, users define structured rules that incorporate triggers, predicates, and enforcement mechanisms, ensuring agents operate within predefined safety boundaries. We implement AgentSpec across multiple domains, including code execution, embodied agents, …

Hypersiniel: Guaranteed Output Delivery Comes (Almost) Free In Private Delegation Of Zksnarks, Yunbo Yang, Yuejia Cheng, Junkai Liang, Kailun Wang, Xuanming Liu, Xiaoguo Li, Jianfei Sun, Jiachen Shen, Xiaolei Dong, Zhenfu Cao, Meng Hao, Guomin Yang, Deng, Robert H., Kui Ren

Research Collection School Of Computing and Information Systems

Zero-knowledge Succinct Non-interactive Argument of Knowledge (zkSNARK) is a powerful cryptographic primitive that enables a prover to convince a verifier that something is true without leaking the private witness.Current zkSNARKs face significant computational costs in generating proofs, which restricts their use in areas like private payments, confidential smart contracts, and anonymous credentials. Private delegation offers a practical solution by outsourcing the heavy computation to powerful external workers without leaking any private information. In this work, we propose HyperSiniel, an efficient private delegation framework for general zkSNARKs that achieves a new feature called guaranteed output delivery (GOD). HyperSiniel is designed to …

Bridging Draft Policy Misalignment: Group Tree Optimization For Speculative Decoding, Shijing Hu, Jingyang Li, Zhihui Lu, Pan Zhou

Research Collection School Of Computing and Information Systems

Speculative decoding accelerates large language model (LLM) inference by letting a lightweight draft model propose multiple tokens that the target model verifies in parallel. Yet existing training objectives optimize only a single greedy draft path, while decoding follows a tree policy that re-ranks and verifies multiple branches. This draft policy misalignment limits achievable speedups. We introduce Group Tree Optimization (GTO), which aligns training with the decoding-time tree policy through two components: (i) Draft Tree Reward, a sampling-free objective equal to the expected acceptance length of the draft tree under the target model, directly measuring decoding performance; (ii) Group-based Draft Policy …

Dragging With Geometry: From Pixels To Geometry-Guided Image Editing, Xinyu Pu, Hongsong Wang, Jie Gui, Pan Zhou

Research Collection School Of Computing and Information Systems

Interactive point-based image editing serves as a controllable editor, enabling precise and flexible manipulation of image content. However, most drag-based methods operate primarily on the 2D pixel plane with limited use of 3D cues. As a result, they often produce imprecise and inconsistent edits, particularly in geometry-intensive scenarios such as rotations and perspective transformations. To address these limitations, we propose a novel geometry-guided drag-based image editing method—GeoDrag, which addresses three key challenges: 1) incorporating 3D geometric cues into pixel-level editing, 2) mitigating discontinuities caused by geometry-only guidance, and 3) resolving conflicts arising from multi-point dragging. Built upon a unified displacement …

From Spatial To Actions: Grounding Vision-Language-Action Model In Spatial Foundation Priors, Zhengshen Zhang, Hao Li, Yalun Dai, Zhengbang Zhu, Lei Zhou, Chenchen Liu, Dong Wang, Francis E. H. Tay, Sijin Chen, Ziwei Liu, Yuxiao Liu, Xinghang Li, Pan Zhou

Research Collection School Of Computing and Information Systems

Existing vision-language-action (VLA) models act in 3D real-world but are typically built on 2D encoders, leaving a spatial reasoning gap that limits generalization and adaptability. Recent 3D integration techniques for VLAs either require specialized sensors and transfer poorly across modalities, or inject weak cues that lack geometry and degrade vision-language alignment. In this work, we introduce FALCON (From Spatial to Action), a novel paradigm that injects rich 3D spatial tokens into the action head. FALCON leverages spatial foundation models to deliver strong geometric priors from RGB alone, and includes an Embodied Spatial Model that can optionally fuse depth, or pose …

Advancing The Understanding Of Palliative Care In Singapore: Knowledge, Attitudes, Receptiveness, And The Moderating Role Of Media Information-Seeking Preferences, Su Lin Yeo, Angel Lee, Raymond Han Lip Ng, May O. Lwin, Yumin Lin

Research Collection Lee Kong Chian School Of Business

Introduction: Despite the growing demand for palliative care, this specialized medical care remains severely underutilized, with almost 50% of countries worldwide lacking access. Given that misperceptions and stigma continue to hinder progress in healthcare policy, this study aims to examine the relationships between palliative care knowledge, attitudes, and receptiveness to better understand how public health communication can enhance awareness and understanding of the topic in Singapore. Applying the adapted knowledge-attitude-practice (KAP) model, it further extends theory by examining the moderating role of media information-seeking between knowledge and receptiveness.Methods: Mixed-mode surveys involving 1,226 participants (926 online and 300 in-person respondents), representing …

Impending-Exit Period And Employee Performance: Rethinking Human Capital Disruption, Yea Hee Ko, Charlie O. Trevor

Research Collection Lee Kong Chian School Of Business

The well-established disruptive effects of employee turnover on firms have typically been attributed to post-exit dynamics, such as losses of human and social capital. Little is known, however, about leavers’ pre-exit job performance, which, if declining in sufficient magnitude as separation nears, may drive some of this disruption. Drawing on career concerns research, we argue that impending exit weakens incentives to improve future career prospects at the firm, thereby resulting in reduced performance. Our analysis reveals strikingly large negative relationships, as job performance during the impending-exit period declines by 53.9% and 79.8% across two performance measures. Additionally, we predict and …

The Robin Hood Effect In Transgressions Against Firms: How Political Ideology Shapes Consumer Justifications, Jason D. Lin, Anat Keinan, Hannah H. Chang, Donald R. Lehmann

Research Collection Lee Kong Chian School Of Business

Consumer transgressions against firms involve moral violations in which consumers take advantage of loopholes in companies’ return and satisfaction-guaranteed policies or the lack of strict enforcement against password sharing, piracy, shoplifting, wardrobing, coupon stacking, promotion abuse, and fraud. Drawing on the timeless tale of Robin Hood, the legendary outlaw who stands up against injustice, the authors demonstrate that consumers use a Robin Hood justification to legitimize their transgressions against firms. Building on moral foundations theory, this research demonstrates that political ideology impacts consumers’ transgression behavior and the perceived morality of these practices. Six studies show that under certain conditions in …

Fixed Effects Estimation Of Spatial Panel Model With Missing Responses: An Application To Us State Tax Competition, Xiaoyu Meng, Zhenlin Yang

Research Collection School Of Economics

We consider estimation and inferences for general spatial panel data models with randomly missing observations on responses. It allows for unobserved spatiotemporal heterogeneity, time-varying endogenous and contextual spatial interactions, time-varying cross-sectional error dependence, and serial correlation. A general M-estimation method is proposed for model estimation and a novel corrected plug-in method is proposed for model inference. Both take into account the estimation of fixed effects. Asymptotic properties of the proposed methods are studied, and finite sample properties are investigated. An empirical application is given using U.S. state tax competition data. The proposed methods apply to matrix exponential spatial specification and …

Genuinely Unbalanced Spatial Panel Data Models: Fixed Effects M-Estimation And Inference, Xiaoyu Meng, Zhenlin Yang

Research Collection School Of Economics

We consider spatial panel data models with genuine unbalancedness arising from the non-presence of some spatial units in certain time periods. General M-estimation methods are proposed for model estimation, which take into account the estimation of the incidental fixed effects parameters and allow for spatiotemporal heteroskedasticity and high-order time-varying spatial effects. Corrected plug-in methods are proposed for standard error estimation. The proposed estimation and inference methods are rigorously studied for their asymptotic properties and finite sample performance. An application to China’s provincial FDI inflows shows that properly accounting for genuine unbalancedness uncovers significant positive spatial spillovers that are masked when …

Conflogger: Enhance Systems’ Configuration Diagnosability Through Configuration Logging, Shiwen Shan, Yintong Huo, Yuxin Su, Zhining Wang, Dan Li, Zibin Zheng

Research Collection School Of Computing and Information Systems

Modern configurable systems offer customization via intricate configuration spaces, yet such flexibility introduces pervasive configuration-related issues such as misconfigurations and latent softwarebugs. Existing diagnosability supports focus on post-failure analysis of software behavior to identify configuration issues, but none of these approaches look into whether the software clue sufficient failure information for diagnosis. To fill in the blank, we propose the idea of configuration logging to enhance existing logging practices at the source code level. We develop ConfLogger, the first tool that unifies configuration-aware static taint analysis with LLM-based log generation to enhance software configuration diagnosability. Specifically, our method 1) identifies …