arXiv subjects

Yiyang Wang

Publications and source records attributed to Yiyang Wang.

At least 19 recordsLinked to original sources

arXiv2026

PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives

Video world models simulate future states conditioned on current observations and user actions. Recent systems have demonstrated impressive video consistency and action controllability over long sequences. However, fairly comparing these interactive models remains challenging. In practice, a human player typically evaluates a world model by pursuing long-horizon objectives through interaction. For example, a user may turn around 360 degrees to see whether the environment remains consistent, or walk into the water and inspect whether realistic water ripples are generated. The action sequence required to achieve the same objective may vary substantially between models, making fixed action-conditioned evaluation unsuitable for cross-model comparison. To address this, we employ multi-modal Agent Players to interact with world models toward specified long-horizon objectives. Building on this paradigm, we introduce PlayWorld, a benchmark providing 171 scenarios, each with a specified objective. To evaluate performance thoroughly, we assess models along four core dimensions: geometry consistency, interaction fidelity, out-of-sight evolution, and insight evolution. In addition, we incorporate basic ability metrics for video quality and controllability. Experiments across nine state-of-the-art world models reveal that current models remain unreliable on long-horizon interactive objectives, particularly in maintaining spatial consistency and persistent state evolution. Code and data are available at https://github.com/kxding/PlayWorld.

cs.CV↗

arXiv2026

CultivAgents: Cultivating Relationship-Centered Multi-Agent Systems for Personalized Gardening

Gardening is critical to support well-being, cultural continuity, and food autonomy, yet existing digital tools often provide generic advice that overlooks gardeners' skills, local ecologies, seasons, and cultural contexts. We introduce CultivAgents, a relationship-centered multi-agent system for personalized, socio-culturally grounded gardening support. Grounded in ethics of care, CultivAgents coordinates multiple specialized agents: an Experience Agent that adapts guidance to users' skill levels, an Environmental Agent that grounds advice in local and seasonal conditions, and an Ethnobotanical Agent that connects plants to cultural knowledge and histories. We evaluated CultivAgents through a three-phase mixed-methods study with domain experts (n=3), HCI researchers (n=7), and community gardeners (n=5), analyzing expert feedback, pre/post surveys, and participatory design activities. Results suggest that CultivAgents helped gardeners translate interest into situated action: community gardeners reported increased confidence (3.00 to 3.60), motivation (4.00 to 4.40), and trust in acting on AI advice (3.20 to 4.00). Participants valued hyperlocal ecological guidance and complementary agent perspectives, while also identifying limits in cultural specificity, ecological grounding, and agent coordination. The work advances relationship-centered AI, offering design implications for multi-agent systems that support food sovereignty, community resilience, and cultural preservation.

cs.HC↗

arXiv2026

UniSD: Towards a Unified Self-Distillation Framework for Large Language Models

Self-distillation (SD) offers a promising path for adapting large language models (LLMs) without relying on stronger external teachers. However, SD in autoregressive LLMs remains challenging because self-generated trajectories are free-form, correctness is task-dependent, and plausible rationales can still provide unstable or unreliable supervision. Existing methods mainly examine isolated design choices, leaving their effectiveness, roles, and interactions unclear. In this paper, we propose UniSD, a unified framework to systematically study self-distillation. UniSD integrates complementary mechanisms that address supervision reliability, representation alignment, and training stability, including multi-teacher agreement, EMA teacher stabilization, token-level contrastive learning, feature matching, and divergence clipping. Across six benchmarks and six models from three model families, UniSD reveals when self-distillation improves over static imitation, which components drive the gains, and how these components interact across tasks. Guided by these insights, we construct UniSDfull, an integrated pipeline that combines complementary components and achieves the strongest overall performance, improving over the base model by +5.4 points and the strongest baseline by +2.8 points. Extensive evaluation highlights self-distillation as a practical and steerable approach for efficient LLM adaptation without stronger external teachers.

cs.CL↗

arXiv2026

TextReg: Mitigating Prompt Distributional Overfitting via Regularized Text-Space Optimization

Large language models (LLMs) are highly sensitive to the prompts used to specify task objectives and behavioral constraints. Many recent prompt optimization methods iteratively rewrite prompts using LLM-generated feedback, but the resulting prompts often become longer, accumulate narrow sample-specific rules, and generalize poorly beyond the training distribution. We study this failure mode as prompt distributional overfitting and argue that it reflects a lack of representation control in discrete text-space optimization. We formalize this view through representational inefficiency, a dual-factor measure that decomposes prompt inefficiency into capacity cost and scope narrowness, attributing distributional prompt overfitting to their coupled growth during optimization. We propose TextReg, a regularization framework that realizes a soft-penalty objective through regularized textual gradients, combining Dual-Evidence Gradient Purification, Semantic Edit Regularization, and Regularization-Guided Prompt Update. Across multiple reasoning benchmarks, TextReg substantially improves out-of-distribution (OOD) generalization, with accuracy gains of up to +11.8% over TextGrad and +16.5% over REVOLVE.

cs.CL↗

arXiv2026

AgentReview: Exploring Peer Review Dynamics with LLM Agents

Peer review is fundamental to the integrity and advancement of scientific publication. Traditional methods of peer review analyses often rely on exploration and statistics of existing peer review data, which do not adequately address the multivariate nature of the process, account for the latent variables, and are further constrained by privacy concerns due to the sensitive nature of the data. We introduce AgentReview, the first large language model (LLM) based peer review simulation framework, which effectively disentangles the impacts of multiple latent factors and addresses the privacy issue. Our study reveals significant insights, including a notable 37.1% variation in paper decisions due to reviewers' biases, supported by sociological theories such as the social influence theory, altruism fatigue, and authority bias. We believe that this study could offer valuable insights to improve the design of peer review mechanisms. Our code is available at https://github.com/Ahren09/AgentReview.

cs.CL↗

arXiv2026

AI Driven Soccer Analysis Using Computer Vision

Sport analysis is crucial for team performance since it provides actionable data that can inform coaching decisions, improve player performance, and enhance team strategies. To analyze more complex features from game footage, a computer vision model can be used to identify and track key entities from the field. We propose the use of an object detection and tracking system to predict player positioning throughout the game. To translate this to positioning in relation to the field dimensions, we use a point prediction model to identify key points on the field and combine these with known field dimensions to extract actual distances. For the player-identification model, object detection models like YOLO and Faster R-CNN are evaluated on the accuracy of our custom video footage using multiple different evaluation metrics. The goal is to identify the best model for object identification to obtain the most accurate results when paired with SAM2 (Segment Anything Model 2) for segmentation and tracking. For the key point detection model, we use a CNN model to find consistent locations in the soccer field. Through homography, the positions of points and objects in the camera perspective will be transformed to a real-ground perspective. The segmented player masks from SAM2 are transformed from camera perspective to real-world field coordinates through homography, regardless of camera angle or movement. The transformed real-world coordinates can be used to calculate valuable tactical insights including player speed, distance covered, positioning heatmaps, and more complex team statistics, providing coaches and players with actionable performance data previously unavailable from standard video analysis.

cs.CV↗

arXiv2026

Deep Ritz Physics-Informed Neural Network Method for Solving the Variational Inequality

Variational inequalities are widely applied in mechanical engineering, fluid penetration, transportation, and other fields. In this paper, a Deep Ritz method based on Physics-Informed Neural Networks (PINNs) is proposed to enhance the accuracy and efficiency of solving elliptic variational inequalities. The Ritz variational method is firstly utilized to transform the variational inequality problem into an optimization problem. Then Bayesian optimization is employed to tune the weights of the loss function, and a residual-based adaptive dataset update strategy is introduced to improve the convergence and accuracy of the model. Numerical experiments show that the proposed method can effectively approximate the analytical solution.

math.NA↗

arXiv2026

Deep Domain Decomposition Method for Solving the Variational Inequality Problems

By integrating physics-informed neural network (PINN) techniques with domain decomposition method, a deep domain decomposition method is presented for solving elliptic variational inequality problems. Based on the Ritz variation method, the elliptic variational inequality problem is firstly reformulated as an optimization problem, and then the subproblem in each subdomain is solved by using the Ritz-PINN method, which the parameters in the network are updated by the Adam optimizer, and the residual-adaptive training by introducing a residual-adaptive dataset update strategy to gradually guide the model to learn more complex regions. Additionally, the impact of overlapping regions on the performance of the new algorithm is explored. Numerical results demonstrate the effectiveness of the proposed algorithm, the mean square error can be reached 1.0e-07, and the number of iterations is independent of grid length h under uniform overlap conditions.

math.NA↗

arXiv2026

CompanionCast: Toward Social Collaboration with Multi-Agent Systems in Shared Experiences

Shared experiences are fundamental to social connection, yet media consumption is increasingly solitary. While AI companions offer real-time reactions and emotional regulation, existing systems either rely on single-agent designs or lack the social awareness and multi-party interaction required to replicate authentic group dynamics. We present CompanionCast, a general framework for orchestrating multiple specialized AI agents as social collaborators within a live shared context. CompanionCast integrates multimodal event detection, rolling context caching for improved grounding, and spatial audio to enhance co-presence. We validate CompanionCast through sports viewing, a domain with rich dynamics and strong social traditions. Pilot studies with soccer fans demonstrate that CompanionCast significantly improves perceived social presence and emotional sharing compared to solitary viewing. We conclude by discussing implications and open challenges for multi-agent systems as social collaborators in shared experiences.

cs.HC↗

arXiv2026

MASCOT: Towards Multi-Agent Socio-Collaborative Companion Systems

Multi-agent systems (MAS) are emerging as promising socio-collaborative companions for emotional and cognitive support. However, existing systems frequently suffer from persona collapse, where agents revert to generic, homogenized assistant behaviors, and social sycophancy, where agents produce redundant, non-constructive dialogue. We propose MASCOT, a multi-agent framework for multi-perspective socio-collaborative companions. MASCOT introduces a novel bi-level optimization strategy to harmonize individual and collective behaviors: 1) Persona-Aware Behavioral Alignment, an RLAIF-driven pipeline that fine-tunes individual agents for agent-specific identities; and 2) Collaborative Dialogue Optimization, a group-level adaptation process that promotes complementary, diverse, and productive discourse. We evaluate MASCOT using human-grounded contexts drawn across both in-domain and out-of-domain (OOD) settings against state-of-the-art baselines. MASCOT improves persona consistency by up to +14.1 and social contribution by up to +10.6. A broad evaluation suite, including human evaluation, multiple LLM judges, three-way comparisons, and automatic metrics, further shows that MASCOT produces more role-consistent and less redundant multi-agent dialogue.

cs.CL↗

arXiv2026

GDRO: Group-level Reward Post-training Suitable for Diffusion Models

Recent advancements adopt online reinforcement learning (RL) from LLMs to text-to-image rectified flow diffusion models for reward alignment. The use of group-level rewards successfully aligns the model with the targeted reward. However, it faces challenges including low efficiency, dependency on stochastic samplers, and reward hacking. The problem is that rectified flow models are fundamentally different from LLMs: 1) For efficiency, online image sampling takes much more time and dominates the time of training. 2) For stochasticity, rectified flow is deterministic once the initial noise is fixed. Aiming at these problems and inspired by the effects of group-level rewards from LLMs, we design Group-level Direct Reward Optimization (GDRO). GDRO is a new post-training paradigm for group-level reward alignment that combines the characteristics of rectified flow models. Through rigorous theoretical analysis, we point out that GDRO supports full offline training that saves the large time cost for image rollout sampling. Also, it is diffusion-sampler-independent, which eliminates the need for the ODE-to-SDE approximation to obtain stochasticity. We also empirically study the reward hacking trap that may mislead the evaluation, and involve this factor in the evaluation using a corrected score that not only considers the original evaluation reward but also the trend of reward hacking. Extensive experiments demonstrate that GDRO effectively and efficiently improves the reward score of the diffusion model through group-wise offline optimization across the OCR and GenEval tasks, while demonstrating strong stability and robustness in mitigating reward hacking.

cs.LG↗

arXiv2025

DiffCamera: Arbitrary Refocusing on Images

The depth-of-field (DoF) effect, which introduces aesthetically pleasing blur, enhances photographic quality but is fixed and difficult to modify once the image has been created. This becomes problematic when the applied blur is undesirable~(e.g., the subject is out of focus). To address this, we propose DiffCamera, a model that enables flexible refocusing of a created image conditioned on an arbitrary new focus point and a blur level. Specifically, we design a diffusion transformer framework for refocusing learning. However, the training requires pairs of data with different focus planes and bokeh levels in the same scene, which are hard to acquire. To overcome this limitation, we develop a simulation-based pipeline to generate large-scale image pairs with varying focus planes and bokeh levels. With the simulated data, we find that training with only a vanilla diffusion objective often leads to incorrect DoF behaviors due to the complexity of the task. This requires a stronger constraint during training. Inspired by the photographic principle that photos of different focus planes can be linearly blended into a multi-focus image, we propose a stacking constraint during training to enforce precise DoF manipulation. This constraint enhances model training by imposing physically grounded refocusing behavior that the focusing results should be faithfully aligned with the scene structure and the camera conditions so that they can be combined into the correct multi-focus image. We also construct a benchmark to evaluate the effectiveness of our refocusing model. Extensive experiments demonstrate that DiffCamera supports stable refocusing across a wide range of scenes, providing unprecedented control over DoF adjustments for photography and generative AI applications.

cs.CV↗

arXiv2025

Formal Degrees and Parabolic Induction: the Maximal Generic Case

We study the compatibility of the formal degree conjecture and the parabolic induction process in the simplest nontrivial case for quasi-split $p$-adic groups. For a generic discrete series $π$ induced from an irreducible supercuspidal $σ$ of a maximal Levi subgroup, we compute the quotient $d(π)/d(σ)$ of formal degrees under some assumptions. As an application, we verify the conjecture for discrete series of split $\mathrm{G}_2$ supported on maximal Levi subgroups.

math.NT↗

arXiv2025

LayerFlow: A Unified Model for Layer-aware Video Generation

We present LayerFlow, a unified solution for layer-aware video generation. Given per-layer prompts, LayerFlow generates videos for the transparent foreground, clean background, and blended scene. It also supports versatile variants like decomposing a blended video or generating the background for the given foreground and vice versa. Starting from a text-to-video diffusion transformer, we organize the videos for different layers as sub-clips, and leverage layer embeddings to distinguish each clip and the corresponding layer-wise prompts. In this way, we seamlessly support the aforementioned variants in one unified framework. For the lack of high-quality layer-wise training videos, we design a multi-stage training strategy to accommodate static images with high-quality layer annotations. Specifically, we first train the model with low-quality video data. Then, we tune a motion LoRA to make the model compatible with static frames. Afterward, we train the content LoRA on the mixture of image data with high-quality layered images along with copy-pasted video data. During inference, we remove the motion LoRA thus generating smooth videos with desired layers.

cs.CV↗

arXiv2025

SciEvo: A 2 Million, 30-Year Cross-disciplinary Dataset for Temporal Scientometric Analysis

Understanding the creation, evolution, and dissemination of scientific knowledge is crucial for bridging diverse subject areas and addressing complex global challenges such as pandemics, climate change, and ethical AI. Scientometrics, the quantitative and qualitative study of scientific literature, provides valuable insights into these processes. We introduce SciEvo, a longitudinal scientometric dataset with over two million academic publications, providing comprehensive contents information and citation graphs to support cross-disciplinary analyses. SciEvo is easy to use and available across platforms, including GitHub, Kaggle, and HuggingFace. Using SciEvo, we conduct a temporal study spanning over 30 years to explore key questions in scientometrics: the evolution of academic terminology, citation patterns, and interdisciplinary knowledge exchange. Our findings reveal critical insights, such as disparities in epistemic cultures, knowledge production modes, and citation practices. For example, rapidly developing, application-driven fields like LLMs exhibit significantly shorter citation age (2.48 years) compared to traditional theoretical disciplines like oral history (9.71 years). Our data and analytic tools can be accessed at https://github.com/Ahren09/SciEvo.

cs.DL↗

arXiv2025

DiffDoctor: Diagnosing Image Diffusion Models Before Treating

In spite of recent progress, image diffusion models still produce artifacts. A common solution is to leverage the feedback provided by quality assessment systems or human annotators to optimize the model, where images are generally rated in their entirety. In this work, we believe problem-solving starts with identification, yielding the request that the model should be aware of not just the presence of defects in an image, but their specific locations. Motivated by this, we propose DiffDoctor, a two-stage pipeline to assist image diffusion models in generating fewer artifacts. Concretely, the first stage targets developing a robust artifact detector, for which we collect a dataset of over 1M flawed synthesized images and set up an efficient human-in-the-loop annotation process, incorporating a carefully designed class-balance strategy. The learned artifact detector is then involved in the second stage to optimize the diffusion model by providing pixel-level feedback. Extensive experiments on text-to-image diffusion models demonstrate the effectiveness of our artifact detector as well as the soundness of our diagnose-then-treat design.

cs.CV↗

arXiv2024

FashionComposer: Compositional Fashion Image Generation

We present FashionComposer for compositional fashion image generation. Unlike previous methods, FashionComposer is highly flexible. It takes multi-modal input (i.e., text prompt, parametric human model, garment image, and face image) and supports personalizing the appearance, pose, and figure of the human and assigning multiple garments in one pass. To achieve this, we first develop a universal framework capable of handling diverse input modalities. We construct scaled training data to enhance the model's robust compositional capabilities. To accommodate multiple reference images (garments and faces) seamlessly, we organize these references in a single image as an "asset library" and employ a reference UNet to extract appearance features. To inject the appearance features into the correct pixels in the generated result, we propose subject-binding attention. It binds the appearance features from different "assets" with the corresponding text features. In this way, the model could understand each asset according to their semantics, supporting arbitrary numbers and types of reference images. As a comprehensive solution, FashionComposer also supports many other applications like human album generation, diverse virtual try-on tasks, etc.

cs.CV↗

arXiv2024

Strong Gravitational Lensing by Static Black Holes in Effective Quantum Gravity

We investigate strong gravitational lensing by two static black hole models (Model-1 and Model-2) within the Effective Quantum Gravity (EQG) framework, characterized by mass $M$ and parameter $ζ$. For $ζ= 0$, they reduce to the Schwarzschild solution, and depending on the parameters, they describe black holes with an event and Cauchy horizon (Model-1), a single horizon (Model-2), or no horizons. Using SMBHs Sgr A* and M87* as lenses and integrating theoretical predictions with recent EHT data, we identify significant differences in lensing signatures due to quantum corrections. For Model-1, the deviations of the lensing observables: $|δθ_{\infty}|$ of black holes in EQG from Schwarzschild black hole, for SMBHs Sgr A* and M87, can reach as much as $1.75~μ$as and $1.32~μ$as, while $|δs|$ is about $30.12$~nas for Sgr A* and $22.63$~nas for M87*. The flux ratio of the first image to all subsequent packed images indicates that EQG black hole images are brighter than their Schwarzschild counterparts, with a deviation in the brightness ratio $|δr_{mag}|$ reaching up to 2.02. The time delays between the second and first images, denoted $|δT_{2,1}|$, exhibit substantial deviations from the GR counterpart, reaching up to 1.53 min for Sgr A* and 1159.9 min for M87*. The EHT constraints on $θ_{sh}$ of Sgr A* and M87* within the $1σ$ region limit the parameters $ζ$. Our analysis concludes that EQG black holes are consistent with the EHT observations within this finite space.

astro-ph.CO↗