arXiv subjects

Weiyu Zhang

Publications and source records attributed to Weiyu Zhang.

At least 19 recordsLinked to original sources

arXiv2026

"It's Like Drinking from a Fire Hose": Understanding and Characterizing Video Learning Experiences for Individuals with ADHD

Video lectures have become increasingly prevalent for education and professional development, yet their static visuals, dense information, and long duration pose attentional challenges for individuals with ADHD. While adaptive learning offers opportunities towards ADHD-accessible video learning, little is known about how to suitably adapt such videos: What components in multimodal video lectures are challenging for ADHD viewers? How do these experiences surface in behavioral signals to trigger an adaptation? What presentations do they prefer? To answer these questions, we conducted an eye-tracking-based retrospective think-aloud study with 16 participants with ADHD, who watched and reflected on a curated set of video lecture segments. Our study uncovered video design elements that hindered learning and revealed participants' coping strategies along with their limitations. By jointly analyzing behavioral signals and retrospective reflections, we characterized how these experiences manifested in behavioral patterns. We further surfaced participants' practices for addressing learning needs beyond the video watching process, and derived design implications for future ADHD-friendly adaptive video learning systems.

cs.HC↗

arXiv2026

PIConGPU modeling of nanoplasma formation in helium nanodroplets irradiated by intense femtosecond laser pulses

Helium nanodroplets provide a unique and versatile platform for investigating strong-field-driven nanoplasma dynamics. In this work, we present large-scale, GPU-accelerated particle-in-cell simulations using \textsc{PIConGPU} to study the interaction of pure helium nanodroplets containing up to $10^{6}$ atoms with intense near-infrared femtosecond laser pulses, and compare the results with single-shot velocity-map electron imaging and ion measurements. The simulations describe the plasma evolution from the first ionization events to collective electron motion, nanoplasma formation, and early expansion. We show that the calculated electron and ion observables reproduce the main features of the measured spectra in systems with similar cluster sizes and laser intensities. Our results demonstrate that \textsc{PIConGPU} captures the essential physics of nanoplasma formation previously addressed mainly with molecular-dynamics or TDDFT approaches, while remaining computationally efficient and applicable to much larger systems. This establishes \textsc{PIConGPU} as a powerful and scalable tool for connecting nanoplasma theory with experimentally accessible observables.

physics.plasm-ph↗

arXiv2026

ContiGuard: A Framework for Continual Toxicity Detection Against Evolving Evasive Perturbations

Toxicity detection mitigates the dissemination of toxic content (e.g., hateful comments, posts, and messages within online social actions) to safeguard a healthy online social environment. However, malicious users persistently develop evasive perturbations to disguise toxic content and evade detectors. Traditional detectors or methods are static over time and are inadequate in addressing these evolving evasion tactics. Thus, continual learning emerges as a logical approach to dynamically update detection ability against evolving perturbations. Nevertheless, disparities across perturbations hinder the detector's continual learning on perturbed text. More importantly, perturbation-induced noises distort semantics to degrade comprehension and also impair critical feature learning to render detection sensitive to perturbations. These amplify the challenge of continual learning against evolving perturbations. In this work, we present ContiGuard, the first framework tailored for continual learning of the detector on time-evolving perturbed text (termed continual toxicity detection) to enable the detector to continually update capability and maintain sustained resilience against evolving perturbations. Specifically, to boost the comprehension, we present an LLM-powered semantic enriching strategy, where we dynamically incorporate possible meaning and toxicity-related clues excavated by LLM into the perturbed text to improve the comprehension. To mitigate non-critical features and amplify critical ones, we propose a discriminability-driven feature learning strategy, where we strengthen discriminative features while suppressing the less-discriminative ones to shape a robust classification boundary for detection...

cs.CL↗

arXiv2026

What Does Vision Tool-Use Reinforcement Learning Really Learn? Disentangling Tool-Induced and Intrinsic Effects for Crop-and-Zoom

Vision tool-use reinforcement learning (RL) can equip vision language models with visual operators such as crop-and-zoom and achieves strong performance gains, yet it remains unclear whether these gains are driven by improvements in tool use or evolving intrinsic capabilities. We introduce MED (Measure--Explain--Diagnose), a coarse-to-fine framework that disentangles intrinsic capability changes from tool-induced effects, decomposes the tool-induced performance difference into gain and harm terms, and probes the mechanisms driving their evolution. Across checkpoint-level analyses in the crop-and-zoom setting on two VLMs with different tool priors and six benchmarks, we find that improvements are dominated by intrinsic learning, while tool-use RL mainly reduces tool-induced harm (e.g., fewer call-induced errors and weaker tool schema interference) and yields limited progress in tool-based correction of intrinsic failures. Overall, in the crop-and-zoom setting studied here, current vision tool-use RL learns to coexist safely with tools rather than master them.

cs.CV↗

arXiv2026

Uni-RS: A Spatially Faithful Unified Understanding and Generation Model for Remote Sensing

Unified remote sensing multimodal models exhibit a pronounced spatial reversal curse: Although they can accurately recognize and describe object locations in images, they often fail to faithfully execute the same spatial relations during text-to-image generation, where such relations constitute core semantic information in remote sensing. Motivated by this observation, we propose Uni-RS, the first unified multimodal model tailored for remote sensing, to explicitly address the spatial asymmetry between understanding and generation. Specifically, we first introduce explicit Spatial-Layout Planning to transform textual instructions into spatial layout plans, decoupling geometric planning from visual synthesis. We then impose Spatial-Aware Query Supervision to bias learnable queries toward spatial relations explicitly specified in the instruction. Finally, we develop Image-Caption Spatial Layout Variation to expose the model to systematic geometry-consistent spatial transformations. Extensive experiments across multiple benchmarks show that our approach substantially improves spatial faithfulness in text-to-image generation, while maintaining strong performance on multimodal understanding tasks like image captioning, visual grounding, and VQA tasks.

cs.CV↗

arXiv2025

A Survey on Training-free Alignment of Large Language Models

The alignment of large language models (LLMs) aims to ensure their outputs adhere to human values, ethical standards, and legal norms. Traditional alignment methods often rely on resource-intensive fine-tuning (FT), which may suffer from knowledge degradation and face challenges in scenarios where the model accessibility or computational resources are constrained. In contrast, training-free (TF) alignment techniques--leveraging in-context learning, decoding-time adjustments, and post-generation corrections--offer a promising alternative by enabling alignment without heavily retraining LLMs, making them adaptable to both open-source and closed-source environments. This paper presents the first systematic review of TF alignment methods, categorizing them by stages of pre-decoding, in-decoding, and post-decoding. For each stage, we provide a detailed examination from the viewpoint of LLMs and multimodal LLMs (MLLMs), highlighting their mechanisms and limitations. Furthermore, we identify key challenges and future directions, paving the way for more inclusive and effective TF alignment techniques. By synthesizing and organizing the rapidly growing body of research, this survey offers a guidance for practitioners and advances the development of safer and more reliable LLMs.

cs.CL↗

arXiv2025

A Gravity-informed Spatiotemporal Transformer for Human Activity Intensity Prediction

Human activity intensity prediction is crucial to many location-based services. Despite tremendous progress in modeling dynamics of human activity, most existing methods overlook physical constraints of spatial interaction, leading to uninterpretable spatial correlations and over-smoothing phenomenon. To address these limitations, this work proposes a physics-informed deep learning framework, namely Gravity-informed Spatiotemporal Transformer (Gravityformer) by integrating the universal law of gravitation to refine transformer attention. Specifically, it (1) estimates two spatially explicit mass parameters based on spatiotemporal embedding feature, (2) models the spatial interaction in end-to-end neural network using proposed adaptive gravity model to learn the physical constraint, and (3) utilizes the learned spatial interaction to guide and mitigate the over-smoothing phenomenon in transformer attention. Moreover, a parallel spatiotemporal graph convolution transformer is proposed for achieving a balance between coupled spatial and temporal learning. Systematic experiments on six real-world large-scale activity datasets demonstrate the quantitative and qualitative superiority of our model over state-of-the-art benchmarks. Additionally, the learned gravity attention matrix can be not only disentangled and interpreted based on geographical laws, but also improved the generalization in zero-shot cross-region inference. This work provides a novel insight into integrating physical laws with deep learning for spatiotemporal prediction.

cs.LG↗

arXiv2025

SRLCG: Self-Rectified Large-Scale Code Generation with Multidimensional Chain-of-Thought and Dynamic Backtracking

Large language models (LLMs) have revolutionized code generation, significantly enhancing developer productivity. However, for a vast number of users with minimal coding knowledge, LLMs provide little support, as they primarily generate isolated code snippets rather than complete, large-scale project code. Without coding expertise, these users struggle to interpret, modify, and iteratively refine the outputs of LLMs, making it impossible to assemble a complete project. To address this issue, we propose Self-Rectified Large-Scale Code Generator (SRLCG), a framework that generates complete multi-file project code from a single prompt. SRLCG employs a novel multidimensional chain-of-thought (CoT) and self-rectification to guide LLMs in generating correct and robust code files, then integrates them into a complete and coherent project using our proposed dynamic backtracking algorithm. Experimental results show that SRLCG generates code 15x longer than DeepSeek-V3, 16x longer than GPT-4, and at least 10x longer than other leading CoT-based baselines. Furthermore, they confirm its improved correctness, robustness, and performance compared to baselines in large-scale code generation.

cs.SE↗

arXiv2025

Distilling human mobility models with symbolic regression

Human mobility is a fundamental aspect of social behavior, with broad applications in transportation, urban planning, and epidemic modeling. Represented by the gravity model and the radiation model, established analytical models for mobility phenomena are often discovered by analogy to physical processes. Such discoveries can be challenging and rely on intuition, while the potential of emerging social observation data in model discovery is largely unexploited. Here, we propose a systematic approach that leverages symbolic regression to automatically discover interpretable models from human mobility data. Our approach finds several well-known formulas, such as the distance decay effect and classical gravity models, as well as previously unknown ones, such as an exponential-power-law decay that can be explained by the maximum entropy principle. By relaxing the constraints on the complexity of model expressions, we further show how key variables of human mobility are progressively incorporated into the model, making this framework a powerful tool for revealing the underlying mathematical structures of complex social phenomena directly from observational data.

physics.soc-ph↗

arXiv2024

BianCang: A Traditional Chinese Medicine Large Language Model

The surge of large language models (LLMs) has driven significant progress in medical applications, including traditional Chinese medicine (TCM). However, current medical LLMs struggle with TCM diagnosis and syndrome differentiation due to substantial differences between TCM and modern medical theory, and the scarcity of specialized, high-quality corpora. To this end, in this paper we propose BianCang, a TCM-specific LLM, using a two-stage training process that first injects domain-specific knowledge and then aligns it through targeted stimulation to enhance diagnostic and differentiation capabilities. Specifically, we constructed pre-training corpora, instruction-aligned datasets based on real hospital records, and the ChP-TCM dataset derived from the Pharmacopoeia of the People's Republic of China. We compiled extensive TCM and medical corpora for continual pre-training and supervised fine-tuning, building a comprehensive dataset to refine the model's understanding of TCM. Evaluations across 11 test sets involving 31 models and 4 tasks demonstrate the effectiveness of BianCang, offering valuable insights for future research. Code, datasets, and models are available on https://github.com/QLU-NLP/BianCang.

cs.CL↗

arXiv2024

Percolation transition of k-frequent destinations network for urban mobility

Urban spatial interactions are a complex aggregation of routine visits and random explorations by individuals. The inherent uncertainty of these random visitations poses significant challenges to understanding urban structures and socioeconomic developments. To capture the core dynamics of urban interaction networks, we analyze the percolation structure of the $k$-most frequented destinations of intracity place-to-place flows from mobile phone data of eight major U.S. cities at a Census Block Group (CBG) level. Our study reveals a consistent percolation transition at $k^* = 130$, a critical threshold for the number of frequently visited destinations necessary to maintain a cohesive urban network. This percolation threshold proves remarkably consistent across diverse urban configurations, sizes, and geographical settings over a 48-month study period, and can largely be interpreted as the joint effect of the emergence of hubness and the level of mixing of residents. Furthermore, we examine the socioeconomic profiles of residents from different origin areas categorized by the fulfillment level of $k^*=130$ principal destinations, revealing a pronounced distinction in the origins' socioeconomic advantages. These insights offer a nuanced understanding of how urban spaces are interconnected and the determinants of travel behavior. Our findings contribute to a deeper comprehension of the structural dynamics that govern urban spatial interactions.

physics.soc-ph↗

arXiv2024

Help Me Reflect: Leveraging Self-Reflection Interface Nudges to Enhance Deliberativeness on Online Deliberation Platforms

The deliberative potential of online platforms has been widely examined. However, little is known about how various interface-based reflection nudges impact the quality of deliberation. This paper presents two user studies with 12 and 120 participants, respectively, to investigate the impacts of different reflective nudges on the quality of deliberation. In the first study, we examined five distinct reflective nudges: persona, temporal prompts, analogies and metaphors, cultural prompts and storytelling. Persona, temporal prompts, and storytelling emerged as the preferred nudges for implementation on online deliberation platforms. In the second study, we assess the impacts of these preferred reflectors more thoroughly. Results revealed a significant positive impact of these reflectors on deliberative quality. Specifically, persona promotes a deliberative environment for balanced and opinionated viewpoints while temporal prompts promote more individualised viewpoints. Our findings suggest that the choice of reflectors can significantly influence the dynamics and shape the nature of online discussions.

cs.HC↗

arXiv2023

Hierarchical accompanying and inhibiting patterns on the spatial arrangement of taxis' local hotspots

The spatial arrangement of taxi hotspots indicates their inherent distribution relationships, reflecting their spatial organization structure, and has received attention in urban studies. Previous studies have primarily explored large-scale hotspots through visual analysis or simple indices, which typically spans hundreds or even thousands of meters. However, the spatial arrangement patterns of small-scale hotspots representing specific popular pick-up and drop-off locations have been largely overlooked. In this study, we quantitatively examine the spatial arrangement of local hotspots in Wuhan and Beijing, China, using taxi trajectory data. Local hotspots are small-scale hotspots with the highest density near the center. Their optimal radius is adaptively calculated based on the data, which is 90 m * 90 m and 110 m * 110 m in Wuhan and Beijing, respectively. Popular hotspots are typically surrounded by less popular ones, although regions with many popular hotspots inhibit the presence of less popular ones. These configurations are termed as hierarchical accompanying and inhibiting patterns. Finally, inspired by both patterns, a KNN-based model is developed to describe these relationships and successfully reproduce the spatial distribution of less popular hotspots based on the most popular ones. These insights enhance our understanding of local urban structures and support urban planning.

cs.CY↗

arXiv2023

Hybrid deliberation: Citizen dialogues in a post-pandemic era

This report first provides a brief review of various forms of dialogue-based participation, e.g., Citizen Assembly, Citizen Lottery, Citizen Jury, Deliberative Polling, and Participatory Budgeting. Challenges associated with these long-lasting practices are identified and hybrid deliberation is proposed as a concept to address the challenges. The report then analyzes six leading examples of digital or hybrid formats of citizen dialogues. Through the comparison of the cases, the report concludes about the hurdles/risks, success factors/opportunities, and best practices for a complementary use of digital and analogue participation formats. Hybrid deliberation is proposed to be the future direction for dialogue-based participation that involves masses and generates high-quality outcomes.

cs.CY↗

arXiv2022

Examining the Impact of Source-product Congruence and Sponsorship Disclosure on the Communicative Effectiveness of Instagram Influencers

Guided by the Persuasion Knowledge Model and the Attribution Theory, this study investigates the perceived source expertise-product attribute congruence and sponsorship disclosure as pertinent factors affecting the communicative effectiveness of influencers. Instagram, with an immense influencer market value projected at USD2.3 billion in 2020, was chosen as the platform context. The study utilised a 2 (source expertise) x2 (product category) x2 (sponsorship disclosure) experiment to examine the roles of source-product congruence and sponsorship disclosure in affecting consumers' perception of extrinsic and intrinsic source motives, consumer resistance and ultimately, advertising effectiveness. Results revealed that the presence of a sponsorship disclosure generated stronger perceptions of extrinsic source motives but did not impact consumer resistance and advertising effectiveness, indicating that the activation of consumers' conceptual persuasion knowledge may not necessarily affect attitudinal persuasion knowledge. Source-product congruence, on the other hand, had main impacts on intrinsic motives, consumer resistance and ad effectiveness. In addition, hierarchical multiple regressions found that source-product congruence triggers a multi-stage process where consumers' perception of intrinsic source motives mediates consumer resistance which subsequently, mediates the relationship between source-product congruence and ad effectiveness.

cs.SI↗

arXiv2022

Young women's cognition of commercial digital signage in shopping malls: A situated action approach

Existing literature on digital signage is growing but has not always emphasized the cognitive processes of the audience. This research aims to address this gap by studying how young women in Singapore cognize commercial digital signage in shopping malls and what cause them to do so. Using cognitive ethnography and taking the situated action approach, our findings suggest a comprehensive list of factors, both external and internal, that influence young women's cognition of commercial digital signage in both positive and negative ways. The research's practical implications are discussed.

cs.HC↗

arXiv2022

MetroGAN: Simulating Urban Morphology with Generative Adversarial Network

Simulating urban morphology with location attributes is a challenging task in urban science. Recent studies have shown that Generative Adversarial Networks (GANs) have the potential to shed light on this task. However, existing GAN-based models are limited by the sparsity of urban data and instability in model training, hampering their applications. Here, we propose a GAN framework with geographical knowledge, namely Metropolitan GAN (MetroGAN), for urban morphology simulation. We incorporate a progressive growing structure to learn hierarchical features and design a geographical loss to impose the constraints of water areas. Besides, we propose a comprehensive evaluation framework for the complex structure of urban systems. Results show that MetroGAN outperforms the state-of-the-art urban simulation methods by over 20% in all metrics. Inspiringly, using physical geography features singly, MetroGAN can still generate shapes of the cities. These results demonstrate that MetroGAN solves the instability problem of previous urban simulation GANs and is generalizable to deal with various urban attributes.

cs.CY↗

arXiv2022

A Review of Research on Civic Technology: Definitions, Theories, History and Insights

There have been initiatives that take advantage of information and communication technologies to serve civic purposes, referred to as civic technologies (Civic Tech). In this paper, we present a review of 224 papers from the ACM Digital Library focusing on Computer Supported Cooperative Work and Human-Computer Interaction, the key fields supporting the building of Civic Tech. Through this review, we discuss the concepts, theories and history of civic tech research and provide insights on the technological tools, social processes and participation mechanisms involved. Our work seeks to direct future civic tech efforts to the phase of by the citizens.

cs.CY↗