Open Access. Powered by Scholars. Published by Universities.®

Digital Commons Network™

Open Access. Powered by Scholars. Published by Universities.®

Singapore Management University

Discipline
Keyword
Publication Year
Publication
Publication Type
File Type

Articles 571 - 600 of 23190

Full-Text Articles in Entire DC Network

Editorial: Special Section On Challenges And Opportunities In Retrieval-Augmented Generation For Llms: Techniques, Trends, And Applications, Philip S. Yu, Haofen Wang, Feida Zhu Jan 2026

Editorial: Special Section On Challenges And Opportunities In Retrieval-Augmented Generation For Llms: Techniques, Trends, And Applications, Philip S. Yu, Haofen Wang, Feida Zhu

Research Collection School Of Computing and Information Systems

Retrieval-Augmented Generation (RAG) represents a transformative advancement for Large Language Models (LLMs) by integrating external knowledge to substantially improve accuracy and mitigate hallucinations. As a pivotal technology in the contemporary generative Artificial Intelligence (AI) landscape, RAG addresses fundamental challenges in knowledge-intensive tasks. This special issue serves as a dedicated platform to showcase these cutting-edge advancements. It features six rigorously peer-reviewed papers that present state-of-the-art research and applications in the rapidly evolving field of RAG.

Go to article

Qualitative Study For Llm-Assisted Design Study Process: Strategies, Challenges, And Roles, Shaolun Ruan, Rui Sheng, Xiaolin Wen, Jiachen Wang, Tianyi Zhang, Yong Wang, Tim Dwyer, Jiannan Li Jan 2026

Qualitative Study For Llm-Assisted Design Study Process: Strategies, Challenges, And Roles, Shaolun Ruan, Rui Sheng, Xiaolin Wen, Jiachen Wang, Tianyi Zhang, Yong Wang, Tim Dwyer, Jiannan Li

Research Collection School Of Computing and Information Systems

Design studies aim to develop visualization solutions for real-world problems across various application domains. Recently, the emergence of large language models (LLMs) has introduced new opportunities to enhance the design study process, providing capabilities such as creative problem-solving, data handling, and insightful analysis. However, despite their growing popularity, there remains a lack of systematic understanding of how LLMs can effectively assist researchers in visualization-specific design studies. In this paper, we conducted a rnulti-stage qualitative study to fill this gap, which involved 30 design study researchers from diverse backgrounds and expertise levels. Through in-depth interviews and carefully-designed questionnaires, we investigated strategies …

Go to article

Artem: Enhancing Large Language Model Agents With Spatial-Temporal Episodic Memory, Cassandra Hui Ming Tan, Budhitama Subagdja, Ah-Hwee Tan Jan 2026

Artem: Enhancing Large Language Model Agents With Spatial-Temporal Episodic Memory, Cassandra Hui Ming Tan, Budhitama Subagdja, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

Current large language models (LLMs) exhibit significant deficiencies in episodic memory tasks including encoding, storing, and retrieving specific information from temporally dependent events over a long period of time. Recent approaches to handle memory tasks in LLMs, such as in-context learning, retrieval-augmented generation (RAG), and fine-tuning, may resolve the long-term retention issues, but are still inadequate to handle tasks requiring chronological awareness of the stored information. We introduce Agentic Retrieval with Temporal-Episodic Memory (ARTEM), a hybrid LLM-based agent architecture integrating LLMs with a self-organizing neural network named Spatial-Temporal Episodic Memory (STEM), designed to handle episodic memory tasks. Our approach employs …

Go to article

Dystop: Dynamic Staleness Control And Topology Construction For Asynchronous Decentralized Federated Learning, Yizhou Shi, Qianpiao Ma, Yan Xu, Junlong Zhou, Ming Hu, Yunming Liao Jan 2026

Dystop: Dynamic Staleness Control And Topology Construction For Asynchronous Decentralized Federated Learning, Yizhou Shi, Qianpiao Ma, Yan Xu, Junlong Zhou, Ming Hu, Yunming Liao

Research Collection School Of Computing and Information Systems

Federated Learning (FL) has emerged as a potential distributed learning paradigm that enables model training on edge devices (i.e., workers) while preserving data privacy. However, its reliance on a centralized server leads to limited scalability. Decentralized federated learning (DFL) eliminates the dependency on a centralized server by enabling peer-to-peer model exchange. Existing DFL mechanisms mainly employ synchronous communication, which may result in training inefficiencies under heterogeneous and dynamic edge environments. Although a few recent asynchronous DFL (ADFL) mechanisms have been proposed to address these issues, they typically yield stale model aggregation and frequent model transmission, leading to degraded training performance …

Go to article

Llamoco: Instruction Tuning Of Large Language Models For Optimization Code Generation, Zeyuan Ma, Yue-Jiao Gong, Hongshu Guo, Jiacheng Chen, Yining Ma, Zhiguang Cao Jan 2026

Llamoco: Instruction Tuning Of Large Language Models For Optimization Code Generation, Zeyuan Ma, Yue-Jiao Gong, Hongshu Guo, Jiacheng Chen, Yining Ma, Zhiguang Cao

Research Collection School Of Computing and Information Systems

Recently, combining the strength of large language models (LLMs) and Evolutionary Computation (EC) has shown promising results for addressing optimization problems. It typically involves either iterative next-step solution seeking or directly prompting LLMs to generate critical optimization codes. However, these methods often suffer from low computational efficiency, high sensitivity to prompt design, and a lack of domain-specific knowledge. We introduce LLaMoCo, the first instruction-tuning framework designed to adapt LLMs for solving optimization problems in a code-to-code manner. LLaMoCo features a comprehensive instruction set that includes code-style problem descriptions as input prompts and robust optimization codes from expert EC optimizers as …

Go to article

Wherefore Art Thou Competitors? How Situational Affordances Help Differentiate Among Prosocials, Individualists, And Competitors, Y. Liu, A.W. Stivers, R.O. Murphy, N.J. Van Doesum, J. Joireman, M. Gallucci, E. Aharonov-Majar, U. Athenstaedt, L. Bai, R. Böhm, N.R. Buchan, X.-P. Chen, K.B. Dumont, J.B. Engelmann, Li, Norman P., H. Euh, S. Fiedler, J. Friesen, S. Gächter, C. Garcia Jan 2026

Wherefore Art Thou Competitors? How Situational Affordances Help Differentiate Among Prosocials, Individualists, And Competitors, Y. Liu, A.W. Stivers, R.O. Murphy, N.J. Van Doesum, J. Joireman, M. Gallucci, E. Aharonov-Majar, U. Athenstaedt, L. Bai, R. Böhm, N.R. Buchan, X.-P. Chen, K.B. Dumont, J.B. Engelmann, Li, Norman P., H. Euh, S. Fiedler, J. Friesen, S. Gächter, C. Garcia

Research Collection School of Social Sciences

The Triple Dominance Measure (choosing between prosocial, individualistic, and competitive options) and the Slider Measure (“sliding” between various orientations, for example, from individualistic to prosocial) are two widely used techniques to measure social value orientation, that is, the weight individuals assign to own and others’ outcomes in interdependent situations. Surprisingly, there is only moderate correspondence between these measures, but it is unclear why and what the implications are for identifying individual differences in social value orientation. Using a dataset of 8021 participants from 31 countries and regions, this study revealed that the Slider Measure identified fewer competitors than the Triple …

Go to article

Artificial Intelligence, Fundamental Motives, And Evolutionary Mismatch, Amy J. Lim, Jose. C. Yong, Edison Sora Tan Jan 2026

Artificial Intelligence, Fundamental Motives, And Evolutionary Mismatch, Amy J. Lim, Jose. C. Yong, Edison Sora Tan

Research Collection School of Social Sciences

In recent years, the intersection of artificial intelligence (AI) and psychology has garnered unprecedented attention, particularly following the advent of generative AI tools in 2022. These tools, capable of producing human-like text, images, and even deepening our understanding of cognitive processes, have not only captured the public imagination but also sparked new concerns and debates within the psychological community. While AI has been a subject of research for decades, the emergence of its generative capabilities has truly thrust AI into the spotlight. This article explores how these advancements are reshaping our understanding of human cognition and behavior, as well as …

Go to article

Winner-Take-All In International Markets? Performance Persistence Of Social Platforms, Yang Yang, Liang Chen, Jiang Wei, Yang Liu Jan 2026

Winner-Take-All In International Markets? Performance Persistence Of Social Platforms, Yang Yang, Liang Chen, Jiang Wei, Yang Liu

Research Collection Lee Kong Chian School Of Business

Strategy research has long linked sustained competitive advantage to barriers to imitation. We highlight network effects as an alternative mechanism and adopt a geotemporal perspective to theorize how firms sustain advantage as it unfolds over time in international markets. Our study examines this question through the performance persistence of social platforms, focusing on how institutional and demand-side conditions shape the sustainability of platforms’ competitive advantages. We propose that intellectual property rights protection may restrict the degree of freedom in information dissemination, dampening the role of network effects in sustaining superior performance, whilst demand heterogeneity may enhance the value of sizable …

Go to article

Smu Launches Resilient Workforces Institute To Strengthen Singapore’S Workforce In The Age Of Ai, Singapore Management University Jan 2026

Smu Launches Resilient Workforces Institute To Strengthen Singapore’S Workforce In The Age Of Ai, Singapore Management University

SMU Press Releases and News

Singapore Management University (SMU) announced the launch of the Resilient Workforces Institute (ResWORK), a new university-level research institute advancing workforce resilience and lifelong learning amid accelerating technological change. It is among the first institutes in Singapore and the region to jointly study adult-learning and the future of work through an integrated, interdisciplinary lens spanning economics, management, behavioural science and technology.

Go to article

Algorithm Envelopment In Platform Markets, Liang Chen, Zhou Zhou, Lester Chan Jan 2026

Algorithm Envelopment In Platform Markets, Liang Chen, Zhou Zhou, Lester Chan

Research Collection Lee Kong Chian School Of Business

The theory of platform envelopment rests on network effects as the key mechanism for value creation, which nonetheless receives mixed support for its efficacy in determining competitive outcomes. We argue that the value of network effects depends on matching quality, which is a function of platform-specific algorithm technology and market-level data-driven learning. In formalizing these conceptualizations, we analyze a model that demonstrates how an entrant with a superior algorithm technology may outcompete an incumbent possessing a user base advantage, a strategy we call “algorithm envelopment.” By considering specific characteristics of data-driven learning, our analysis leads to propositions regarding the entry …

Go to article

The Effects Of Financing Green And Brown Sectors: What Do Theories And Evidence Say?, Hao Liang, Maria Teresa Punzi Jan 2026

The Effects Of Financing Green And Brown Sectors: What Do Theories And Evidence Say?, Hao Liang, Maria Teresa Punzi

Research Collection Lee Kong Chian School Of Business

This paper critically examines the economic and welfare implications of financing green and brown sectors. Drawing on a comprehensive review of recent theoretical and empirical literature, we highlight that while conventional green finance—allocating capital toward environmentally friendly (“green”) sectors and away from carbon-intensive (“brown”) sectors—can promote decarbonization, it may also produce unintended externalities. In particular, it can inadvertently incentivize higher emissions from brown firms and contribute to economic disruption. Using a dynamic stochastic general equilibrium (DSGE) model, we demonstrate that lowering the cost of capital for green sectors leads to only modest reductions in emissions, whereas raising it for brown …

Go to article

Policy Uncertainty Reduces Green Innovation, Mengyu Wang, Jeffrey Wurgler, Hong Zhang Jan 2026

Policy Uncertainty Reduces Green Innovation, Mengyu Wang, Jeffrey Wurgler, Hong Zhang

Research Collection Lee Kong Chian School Of Business

Policy uncertainty can undermine the power of government subsidies to stimulate environmentally friendly research and development. We show that Chinese firms’ green R&D falls as the uncertainty of environmental subsidies rises: Exogenous, weather-driven air pollution variability induces subsidies to fluctuate, and firms in areas with high weather-driven subsidy variability undertake less green R&D and hire fewer technical employees, controlling for the average level of subsidies. Heavy emitters and environmental technology firms are more affected. The results also illustrate how policy uncertainty can arise when policymakers are influenced by conditions that are salient but with causes that are difficult to disentangle.

Go to article

Fluid Agency In Ai Systems: A Case For Functional Equivalence In Copyright, Patent, And Tort, Anirban Mukherjee, Hannah H. Chang Jan 2026

Fluid Agency In Ai Systems: A Case For Functional Equivalence In Copyright, Patent, And Tort, Anirban Mukherjee, Hannah H. Chang

Research Collection Lee Kong Chian School Of Business

Modern Artificial Intelligence (AI) systems exhibit fluid agency in multi-step workflows: lacking human-like consciousness or culpability, yet they display behavior that is (i) stochastic (probabilistic and path‑dependent), (ii) dynamic (co‑evolving with user interaction), and (iii) adaptive (able to reorient across contexts). These properties generate valuable outputs but collapse attribution, irreducibly entangling human and machine inputs. Doctrines that assume traceable provenance—authorship, inventorship, and liability—fracture under this unmappability, yielding ownership gaps and moral “crumple zones.”This Article argues that only functional equivalence stabilizes doctrine under unmappability: Where provenance is indeterminate, legal frameworks should treat human and AI contributions as equivalent for allocating rights …

Go to article

Large Owner Expropriation Threat And Stock Option Pay's Effect On Firm Risk-­Taking Behaviors In Weak Institutions, Cuili Qian, Lipeng (Gary) Ge, Xuesong Geng, Jiatao Li, Maria Hasenhuttl Jan 2026

Large Owner Expropriation Threat And Stock Option Pay's Effect On Firm Risk-­Taking Behaviors In Weak Institutions, Cuili Qian, Lipeng (Gary) Ge, Xuesong Geng, Jiatao Li, Maria Hasenhuttl

Research Collection Lee Kong Chian School Of Business

Research Question/Issue: This study investigates the impact of managerial stock option pay on firm risk-­ taking behaviors in aweak institutional context, a critical question that has been overlooked by the literature. Specifically, we build on the comparative corporate governance perspective that emphasizes the implications of large shareholders' expropriation threat in weak institutions to develop predictions about their impact on the stock option's incentive alignment effect. We further explore the boundary conditions of such a relationship.Research Findings/Insights: Based on a sample of Chinese listed firms between 2006 and 2016, we find that a high level of large shareholders' expropriation threat weakens …

Go to article

The Charity Capacity Curse, Shirley (Lijun) Zhang, Thomas Allard, David Hardisty, Shane Wang Jan 2026

The Charity Capacity Curse, Shirley (Lijun) Zhang, Thomas Allard, David Hardisty, Shane Wang

Research Collection Lee Kong Chian School Of Business

Can displays of charity capacity cues paradoxically harm charities' fundraising efforts? Six pre-registered studies provide evidence of a charity capacity curse whereby donors penalize charities for displaying capacity cues, such as human resources capacity, organizational capacity, and financial capacity cues. This curse occurs because donors perceive higher-capacity charities as needing less help. As donors typically donate based on need, they donate less to charities with superior capacity signals. This effect is attenuated when: (1) higher-capacity charities' needs are made salient, and (2) donors reflect on how their decision impacts end-beneficiaries. This research contributes to the literature on effective altruism and …

Go to article

Multihoming And Single-Homing Complementors' Responses To Intensified Between-Platform Competition: Evidence From The Youtube–Twitch Rivalry, Niloofar Abolfathi Jan 2026

Multihoming And Single-Homing Complementors' Responses To Intensified Between-Platform Competition: Evidence From The Youtube–Twitch Rivalry, Niloofar Abolfathi

Research Collection Lee Kong Chian School Of Business

Research Summary: This paper examines how heterogeneous YouTube content creators-specifically, single-homing and multihoming channels-respond to a sudden increase in the exposure of the rival platform, Twitch. Contrary to expectations, I find that multihoming channels reduce their content creation efforts and underperform relative to single-homers. To make sense of these surprising results, I develop a formal model that characterizes the strategic behavior of different types of complementors operating across two platforms with varying levels of user exposure. The model offers a plausible explanation for the observed empirical patterns, suggesting that intensified competition between platforms, as a consequence of their changing exposure, …

Go to article

Bridging The Social Class Capital Gap: A Psychological Intervention In The Newcomer Adjustment Context, Grace J. H. Lim, Marko Pitesa Jan 2026

Bridging The Social Class Capital Gap: A Psychological Intervention In The Newcomer Adjustment Context, Grace J. H. Lim, Marko Pitesa

Research Collection Lee Kong Chian School Of Business

Motivated by observations that workers from lower social class backgrounds often experience lower career outcomes even after securing desirable jobs, we adopted an abductive approach—combining theory (the newcomer personal capital framework) with qualitative evidence from open‐ended accounts (Study 1)—to identify three challenges these workers face after joining organizations as newcomers: limited cultural capital (i.e., institutional knowledge), lower social capital (i.e., social self‐efficacy), and lower psychological capital (i.e., distress tolerance). Furthermore, in Study 1, we developed and tested a psychological intervention targeting these challenges and found that it effectively addressed them. In Studies 2 and 3, both preregistered field experiments, we …

Go to article

Tuned Out Or Dialed In: How Attributions Shape Observer Reactions To Music Listeners At Work, Oguz Gencay, Trevor Foulk, Michael Schaerer Jan 2026

Tuned Out Or Dialed In: How Attributions Shape Observer Reactions To Music Listeners At Work, Oguz Gencay, Trevor Foulk, Michael Schaerer

Research Collection Lee Kong Chian School Of Business

Music listening while working is prevalent in contemporary workplaces, with extensive research demonstrating its psychological and behavioral implications. Shifting focus from intra-individual outcomes, the present research examines the social implications of music listening at work. To do so, we adopt a novel third-party perspective to investigate how observers perceive and react to music listeners. Building on attribution theory, we argue that music listening is an ambiguous behavior that invites observers to make leisure or productivity attributions, thereby shaping how they perceive and treat music listeners. We hypothesize that observers perceive music listeners (vs. non-listeners) as less engaged if they attribute …

Go to article

Trading With The Enemy: A Coopetitive Perspective Of Resource Exchange At Arm’S Length, Zhefei Li, David Gomulya, Heli Wang Jan 2026

Trading With The Enemy: A Coopetitive Perspective Of Resource Exchange At Arm’S Length, Zhefei Li, David Gomulya, Heli Wang

Research Collection Lee Kong Chian School Of Business

Research Summary: Research on inter-firm cooperation often focuses on long-term strategic alliances while overlooking the distinct form of arm's-length resource exchanges. Our study addresses this gap using the National Basketball Association (NBA) data for precise resource exchange measurements. Contrary to strategic alliances, where competitive pressure fosters cooperation, we find that competitive pressure reduces cooperation in arm's-length exchanges. However, relational capital helps to facilitate resource exchanges. We also found that while resource bundling exacerbates the negative effect of competitive pressure, it amplifies the positive effect of relational capital. Interestingly, relational rivalry intensifies the impact of competitive pressure but does not diminish …

Go to article

Testing Mean Stability Of Heteroskedastic Time Series, Violetta Dalla, Liudas Giraitis, Peter C. B. Phillips Jan 2026

Testing Mean Stability Of Heteroskedastic Time Series, Violetta Dalla, Liudas Giraitis, Peter C. B. Phillips

Research Collection School Of Economics

Time series models are often fitted to the data without preliminary checks for stability of the mean and variance, conditions that may not hold in much economic and financial data, particularly over long periods. Ignoring such shifts may result in fitting models with spurious dynamics that lead to unsupported and controversial conclusions about time dependence, causality, and the effects of unanticipated shocks. In spite of what may seem as obvious differences between a time series of independent variates with changing variance and a stationary conditionally heteroskedastic (GARCH) process, such processes may be hard to distinguish in applied work using basic …

Go to article

Multidimensional Analysis Of Human Outdoor Comfort: Integrating Just-In-Time Adaptive Interventions (Jitais) In Urban Digital Twins, Binyu Lei, Pengyuan Liu, Kunihiko Fujiwara, Mario Frei, Clayton Miller, Xuan Yun Chua, Filip Biljecki Jan 2026

Multidimensional Analysis Of Human Outdoor Comfort: Integrating Just-In-Time Adaptive Interventions (Jitais) In Urban Digital Twins, Binyu Lei, Pengyuan Liu, Kunihiko Fujiwara, Mario Frei, Clayton Miller, Xuan Yun Chua, Filip Biljecki

Research Collection College of Integrative Studies

How can bidirectional information exchange be enhanced in urban digital twins, and support human-centric data and processes? Their key characteristic is the nearly real-time exchange of information, allowing adjustments to physical environments based on simulations and analytics within virtual models. Yet, achieving such interaction remains challenging, particularly regarding device deployment and infrastructure development. Embracing the concept of humans as sensors, this work develops a two-way framework based on the emerging concept of just-in-time adaptive interventions (JITAIs), exploring how urban digital twins can play a role in understanding and enhancing human comfort outdoors. Human comfort outdoors is inherently spatio-temporal and personalised, …

Go to article

Smart Cities Beyond Methodological Cityism: Foregrounding Developmentalism, Scalar Flexing And The Rebranding Of The 'Urban', Prerona Das, Orlando Woods, Lily Kong Jan 2026

Smart Cities Beyond Methodological Cityism: Foregrounding Developmentalism, Scalar Flexing And The Rebranding Of The 'Urban', Prerona Das, Orlando Woods, Lily Kong

Research Collection College of Integrative Studies

In this paper we show that in Southeast Asia, smart city development has become a type of digital urban transformation that is not targeted towards any particular territorial formation or settlement. Instead, development-oriented states in the region leverage the smart city rhetoric as a nationwide branding strategy to leapfrog development stages. Such smart city policies are infused with a scalar logic – which we call ‘scalar flexing’ – that captures the enmeshing of local and national scales, causing local scale smart city initiatives to be stretched across wide-ranging, national-scale urban transformation agendas. We examine the nationwide 100 smart cities policies …

Go to article

How Do You Fix Knowledge In Place? Digital Nomads And Infrastructural Stickiness, Emma Grimley, Prerona Das, Orlando Woods, Lily Kong Jan 2026

How Do You Fix Knowledge In Place? Digital Nomads And Infrastructural Stickiness, Emma Grimley, Prerona Das, Orlando Woods, Lily Kong

Research Collection College of Integrative Studies

Remote work has become increasingly accessible due to the growing acceptance of remote work arrangements following the COVID-19 pandemic. Some state actors, especially in countries which face a shortage of digital talent and have reputations as leisure-tourism destinations, have begun considering the opportunities for digital transformation or workforce upskilling that highly skilled remote workers may offer. This paper develops the case of remote workers in Thailand to argue that infrastructural stickiness is a critical determinant for effective knowledge transfer to occur. By infrastructural stickiness we refer to the regulatory and socio-material infrastructures that both attract and retain international remote workers …

Go to article

Modelling Nature? The Digital Twinning And Untwinning Of Urban Farms, Prerona Das, Orlando Woods, Lily Kong Jan 2026

Modelling Nature? The Digital Twinning And Untwinning Of Urban Farms, Prerona Das, Orlando Woods, Lily Kong

Research Collection College of Integrative Studies

Data-driven solutions and innovations have been extended to various industries, including agriculture, to increase efficiency. With the emergence of smart farming and agriculture 4.0, advanced technologies such as IoT, big data, machine learning, and cloud computing are increasingly integrated into farming. Among these, technologies such as digital models, digital twins and the metaverse are now used to control various farming activities remotely based on real-time data. However agricultural ecosystems, involving uncertain nature-based factors and complex human-nature interactions cannot always be accurately represented or predicted by digital models. Digital models aim to reduce natural environments into controllable game-like situations and undermine …

Go to article

Reinforce Trustworthiness In Multimodal Emotional Support System, Huy M. Le, Dat Tien Nguyen, Ngan T. T. Vo, Tuan D. Q. Nguyen, Nguyen Le Binh, Duy Minh Ho Nguyen, Daniel Sonntag, Lizi Liao, Binh T. Nguyen Jan 2026

Reinforce Trustworthiness In Multimodal Emotional Support System, Huy M. Le, Dat Tien Nguyen, Ngan T. T. Vo, Tuan D. Q. Nguyen, Nguyen Le Binh, Duy Minh Ho Nguyen, Daniel Sonntag, Lizi Liao, Binh T. Nguyen

Research Collection School Of Computing and Information Systems

In today's world, emotional support is increasingly essential, yet it remains challenging for both those seeking help and those offering it. Multimodal approaches to emotional support show great promise by integrating diverse data sources to provide empathetic, contextually relevant responses, fostering more effective interactions. However, current methods have notable limitations, often relying solely on text or converting other data types into text, or providing emotion recognition only, thus overlooking the full potential of multimodal inputs. Moreover, many studies prioritize response generation without accurately identifying critical emotional support elements or ensuring the reliability of outputs. To overcome these issues, we introduce …

Go to article

Cross-Modal Proxy Evolving For Ood Detection With Vision-Language Models, Hao Tang, Yu Liu, Shuanglin Yan, Fei Shen, Shengfeng He, Jing Qin Jan 2026

Cross-Modal Proxy Evolving For Ood Detection With Vision-Language Models, Hao Tang, Yu Liu, Shuanglin Yan, Fei Shen, Shengfeng He, Jing Qin

Research Collection School Of Computing and Information Systems

Reliable zero-shot detection of out-of-distribution (OOD) inputs is critical for deploying vision-language models in open-world settings. However, the lack of labeled negatives in zero-shot OOD detection necessitates proxy signals that remain effective under distribution shift. Existing negative-label methods rely on a fixed set of textual proxies, which (i) sparsely sample the semantic space beyond in-distribution (ID) classes and (ii) remain static while only visual features drift, leading to cross-modal misalignment and unstable predictions. In this paper, we propose CoEvo, a training- and annotation-free test-time framework that performs bidirectional, sample-conditioned adaptation of both textual and visual proxies. Specifically, CoEvo introduces a …

Go to article

Look, Compare And Draw: Differential Query Transformer For Automatic Oil Painting, Lingyu Liu, Yaxiong Wang, Li Zhu, Lizi Liao, Zhedong Zheng Jan 2026

Look, Compare And Draw: Differential Query Transformer For Automatic Oil Painting, Lingyu Liu, Yaxiong Wang, Li Zhu, Lizi Liao, Zhedong Zheng

Research Collection School Of Computing and Information Systems

This work introduces a new approach to automatic oil painting that emphasizes the creation of dynamic and expressive brushstrokes. A pivotal challenge lies in mitigating the duplicate and common-place strokes, which often lead to less aesthetic outcomes. Inspired by the human painting process, i.e., observing, comparing, and drawing, we incorporate differential image analysis into a neural oil painting model, allowing the model to effectively concentrate on the incremental impact of successive brushstrokes. To operationalize this concept, we propose the Differential Query Transformer (DQ-Transformer), a new architecture that leverages differentially derived image representations enriched with positional encoding to guide the stroke …

Go to article

Compendia: Automated Visual Storytelling Generation From Online Article Collection, Manusha Imesh Karunathilaka Gamage Vidana, Litian Lei, Yiming Gao, Yong Wang, Jiannan Li Jan 2026

Compendia: Automated Visual Storytelling Generation From Online Article Collection, Manusha Imesh Karunathilaka Gamage Vidana, Litian Lei, Yiming Gao, Yong Wang, Jiannan Li

Research Collection School Of Computing and Information Systems

In the digital age, readers value quantitative journalism that is clear, concise, analytical, and humancentred. To understand complex topics, they often piece together scattered facts from multiple articles. Visual storytelling can transform fragmented information into clear, engaging narratives, yet its use with unstructured online articles remains largely unexplored. To fill this gap, we present Compendia, an automated system that analyzes online articles in response to a user’s query and generates a coherent data story tailored to the user’s informational needs. Compendia addresses key challenges of storytelling from unstructured text through two modules covering: Online Article Retrieval, which gathers relevant articles; …

Go to article

Generalized Visual Relation Detection With Diffusion Models, Kaifeng Gao, Siqi Chen, Hanwang Zhang, Jun Xiao, Yueting Zhuang, Qianru Sun Jan 2026

Generalized Visual Relation Detection With Diffusion Models, Kaifeng Gao, Siqi Chen, Hanwang Zhang, Jun Xiao, Yueting Zhuang, Qianru Sun

Research Collection School Of Computing and Information Systems

Visual relation detection (VRD) aims to identify relationships (or interactions) between object pairs in an image. Although recent VRD models have achieved impressive performance, they are all restricted to pre-defined relation categories, while failing to consider the semantic ambiguity characteristic of visual relations. Unlike objects, the appearance of visual relations is always subtle and can be described by multiple predicate words from different perspectives, e.g., “ride” can be depicted as “race” and “sit on”, from the sports and spatial position views, respectively. To this end, we propose to model visual relations as continuous embeddings, and design diffusion models to achieve …

Go to article

Benchmarking Gaslighting Negation Attacks Against Reasoning Models, Bin Zhu, Hailong Yin, Jingjing Chen, Yu Gang Jiang Jan 2026

Benchmarking Gaslighting Negation Attacks Against Reasoning Models, Bin Zhu, Hailong Yin, Jingjing Chen, Yu Gang Jiang

Research Collection School Of Computing and Information Systems

Recent advances in reasoning-centric models promise improved robustness through mechanisms such as chain-of-thought prompting and test-time scaling. However, their ability to withstand gaslighting negation attacks—adversarial prompts that confidently deny correct answers—remains underexplored. In this paper, we conduct a systematic evaluation of three state-of-the-art reasoning models, i.e., OpenAI’s o4-mini, Claude-3.7-Sonnet and Gemini-2.5-Flash, across three multimodal benchmarks: MMMU, MathVista, and CharXiv. Our evaluation reveals significant accuracy drops (25–29% on average) following gaslighting negation attacks, indicating that even top-tier reasoning models struggle to preserve correct answers under manipulative user feedback. Built upon the insights of the evaluation and to further probe this vulnerability, …

Go to article

First Prev Next Last