Searcharxiv⌕ Search

arXiv subjects

Liang Peng

Publications and source records attributed to Liang Peng.

At least 19 recordsLinked to original sources
arXiv2026

RoboPhys-3D: A Comprehensive Embodied World Model Evaluation via 3D Reconstruction

Video world models increasingly serve as data engines, action planners, and simulators for embodied AI, but conventional embodied world model (EWM) benchmarks lack a unified 3D-grounded protocol for establishing whether generated rollouts preserve the underlying 3D scene state or translate into executable actions. We introduce RoboPhys-3D, a 3D-grounded EWM benchmark built on RoboTwin 2.0, covering 50 manipulation tasks across four regimes, with 5,000 episodes and 25,000 multi-view ground-truth videos. A defining feature of RoboPhys-3D is that generated and ground-truth videos are processed through the same 3D reconstruction pipeline, enabling reconstruction-induced error to be distinguished from generation-induced error. The RoboPhys-3D benchmark organizes 50 complementary metrics into 18 sub-dimensions across four levels: pixel-level fidelity, 3D geometry consistency, state-level understanding, and task-level completeness. We further introduce Average Full Score, a hierarchical score averaging all 50 metrics for comprehensive evaluation, and RoboPhyscore, a compact task-aligned score averaging the metrics most strongly correlated with task success. Among the four representative video world models, Cosmos 3 achieves the highest RoboPhyscore (0.6330, 92.7% of ground truth), while state- and execution-grounded metrics reveal substantial failures that perceptual and vision-language model-based judgments fail to capture. RoboPhyscore further exhibits strong agreement with human evaluation (Pearson r = 0.9761 and Spearman \r{ho} = 0.8962), demonstrating the importance of grounded, execution-aware evaluation for EWM capability.

cs.RO↗
arXiv2026

A General-Purpose VLM Can Teach an Astronomy Foundation Model to Better Recognize Galaxy Morphology

Existing astronomy foundation models provide strong galaxy representations, but adapting them to new survey conditions and survey-specific morphology recognition tasks still requires substantial human supervision. We show that VLM-based VQA systems contain meaningful visual-semantic priors that can serve as weak supervision for downstream morphology classifiers and improve morphology classification under limited human-label budgets. We first introduce a survey-oriented VQA benchmark spanning two representative imaging regimes and evaluate state-of-the-art VLMs on galaxy morphology questions. The results show that these models capture useful morphology signals and informative uncertainty, but are not sufficiently reliable to replace human annotators. Motivated by this finding, we use a general-purpose VLM as a morphology teacher for Zoobot, an astronomy foundation model pretrained on large-scale Galaxy Zoo annotations. Across two survey domains and multiple annotation budgets, the VLM teacher consistently improves Zoobot's downstream morphology classification. These results demonstrate that a general-purpose VLM provides knowledge complementary to an astronomy foundation model and can teach it to better recognize galaxy morphology under limited human supervision. The resulting pipeline is designed for label-efficient adaptation to forthcoming large-scale surveys, including the Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) and the Nancy Grace Roman Space Telescope. The benchmark and code are publicly available at https://github.com/fw-ic/VLM-morphology-teacher.

cs.CV↗
arXiv2026

PVCap: Towards Accurate 3D Dense Captioning via PseudoCap and VoxelCapNet

3D dense captioning, an emerging vision-language task, aims to generate descriptive sentences for each object in the 3D scene. Despite the impressive results achieved by previous methods, they suffer from two limitations. First, current research often employs global rigid transformations, such as rotation, to augment scenes without changing their spatial layouts. However, diverse spatial layouts are crucial for training a 3D dense captioning model to describe spatial relations between objects. Second, previous works mainly focus on the design of the caption generation pipeline while utilizing a simple network architecture for other components, i.e., backbone and detection head, which is crucial for extracting rich semantic information for captioning. In this paper, we propose PVCap to alleviate the aforementioned problems. Our PVCap consists of PseudoCap and VoxelCapNet. Specifically, PseudoCap employs a random mixing technique on instances within the dataset, generating numerous pseudo frames with diverse spatial layouts at the instance level. By utilizing a teacher-student framework, PseudoCap obtains pseudo caption labels for these pseudo frames. This data augmentation approach significantly increases the number of training samples and enhances the model's ability to describe the environment effectively. Regarding VoxelCapNet, we introduce a robust caption network that utilizes voxel features and adapts the caption head to the voxel-based network architecture. Our VoxelCapNet can serve as a competitive baseline for future research on 3D dense captioning. Extensive experiments are conducted on two prevalent benchmarks, i.e., ScanRefer and Nr3D. Notably, our method surpasses current state-of-the-art by 11.41% and 13.99% in CIDEr@0.5IoU, respectively. Codes will be made publicly available.

cs.CV↗
arXiv2026

GPU-Accelerated Matrix-Based Hough Transform for Online Track Reconstruction in the STCF MDC

The Super Tau-Charm Facility (STCF) is a proposed next-generation high-luminosity electron-positron collider operating at center-of-mass energies of 2-7 GeV for precision studies of tau-charm physics. Its high event rate, detector occupancy, and background level impose stringent requirements on real-time track reconstruction in MDC, particularly for low-transverse-momentum particles with strongly curved or multi-turn trajectories. To address this challenge, we develop a GPU-accelerated matrix-based Hough transform method for online track reconstruction in the STCF MDC. Following an algorithm-architecture co-design paradigm, the data representation and computational workflow of the conformal Hough transform are reformulated for GPU execution. The original irregular parameter-space computations are organized into regular matrix-based operations, and the core computations are adapted to CUDA thread organization and the GPU memory hierarchy to exploit the inherent parallelism of the Hough transform and reduce computational and data-transfer overhead. Tests on five representative simulated physics channels with nominal background overlay show an average signal retention ratio of 93.04%, while reducing the retained hit volume to 34.92% of the original level. The GPU implementation processes 1,000 events in approximately 0.14s, achieving a speedup of 151.57 x compared with the CPU baseline. These results demonstrate that the proposed method substantially improves track reconstruction throughput while preserving track-associated hits, providing a new methodological perspective for real-time track reconstruction in future high-luminosity particle-collider experiments.

hep-ex↗
arXiv2026

Neural-Network-Assisted Binary Template Construction for Matrix-Based Pattern Matching in the STCF MDC

The Super Tau-Charm Facility, operating at high luminosity, will produce high event rates and high data throughput, imposing stringent requirements on fast track finding and data reduction and compression algorithms in the High-Level Trigger. Local track segment finding in the Main Drift Chamber underpins subsequent segment combination and full track reconstruction, yet high background rates and limited detection efficiency can significantly increase the risk of false triggers and signal loss in pattern matching algorithms. This paper presents a neural-network-assisted framework for constructing binary template libraries used in matrix-based pattern matching for MDC local track segment finding. The framework formulates template construction as a differentiable multi-objective optimization problem, employing a neural network to jointly learn template parameters under multiple constraints. After training, only binary template pairs are exported and deployed into the existing bitwise pattern matching routine, requiring no neural network inference at runtime and thus preserving the deterministic, fast, and parallelizable nature of the online algorithm. Experimental results based on simulation samples demonstrate that, under limited detection efficiency, the resulting template library maintains relatively high signal retention across different transverse momentum ranges and background levels, and can be flexibly tailored to adjust the coverage range according to practical requirements. The proposed approach decouples the physics performance from the computational speed by combining the improved physics performance brought by offline neural-network-based optimization with the determinism and high speed of a conventional online algorithm, suggesting a new research direction for artificial-intelligence-enhanced online data processing in high-luminosity particle collider experiments.

hep-ex↗
arXiv2026

Qwen-Image-2.0-RL Technical Report

We present Qwen-Image-2.0-RL, a post-training pipeline that applies reinforcement learning from human feedback (RLHF) and on-policy distillation (OPD) to improve both the visual quality and instruction-following capability of the Qwen-Image-2.0 diffusion model. To provide reliable reward signals, we construct task-specific composite reward models by fine-tuning vision-language models with a pointwise scoring paradigm and chain-of-thought reasoning. For text-to-image generation, the reward models cover alignment, aesthetics, and portrait fidelity dimensions. For image editing tasks, the reward system addresses instruction-following accuracy and face identity preservation. Building on this reward system, we develop a scalable GRPO-based RL training framework, incorporating a hybrid classifier-free guidance (CFG) strategy to preserve pre-trained knowledge, prompt curation via intra-group reward range filtering, and per-category reward weight calibration. To merge the task-specialized RL policies for T2I and editing, we propose on-policy distillation as the final training stage, which consolidates multiple teachers into a single student model through trajectory-level velocity matching. Extensive evaluation shows that Qwen-Image-2.0-RL achieves 57.84 overall score on Qwen-Image-Bench (+2.61 over the base model), Elo ratings of 1193 in text-to-image arena (+78) and 1349 in image edit arena (+93), demonstrating consistent gains in aesthetic quality, prompt adherence, and editing accuracy.

cs.CV↗
arXiv2026

Qwen-Image-Flash: Beyond Objective Design

Few-step distillation has become an effective strategy for accelerating advanced visual generative models, yet prior work has largely focused on distillation objectives. In this work, we revisit few-step distillation from a complementary perspective, focusing on the training recipe that critically shapes student performance. Using Qwen-Image-2.0 as a representative case, we systematically investigate three factors in unified text-to-image generation and instruction-guided image editing distillation: data composition, teacher guidance, and task mixture. Our empirical analysis reveals several non-obvious behaviors, which motivate the development of Qwen-Image-Flash. Overall, our results suggest that effective few-step distillation requires not only carefully designed objectives, but also principled organization of the broader training pipeline.

cs.CV↗
arXiv2026

Roughness-controlled Tribocharging Governs Friction in Dry Glass Contacts

Friction is commonly reduced by polishing surfaces, based on the idea that roughness enhances mechanical interlocking and thus friction. Here we show that, for dry glass-glass contacts, increasing nanoscale roughness can instead reduce friction because it suppresses triboelectric adhesion. Using rheometer-based friction measurements in dry nitrogen, super-resolution imaging of the real contact area, soft x-ray discharge, and Faraday-cup electrometry, we demonstrate that sliding generates substantial tribocharges whose electrostatic attraction contributes significantly to friction. As the root-mean-square surface slope h'_rms of the glass ball is increased from 0.01 to 0.09, the real contact area and retained tribocharge both decrease strongly, while the average contact pressure increases by a factor of three; nevertheless, the friction coefficient drops by about 30%. Discharging the interface with soft x-rays largely removes the roughness dependence of friction. Our results show that nanoscale roughness controls tribocharging and electroadhesion in dielectric contacts, inverting the classical relation between roughness and friction and identifying triboelectric effects as a key design parameter for friction control.

cond-mat.soft↗
arXiv2026

Quantifying Social Inflation in Liability Insurance with Advanced Statistical Methods

Social inflation, which is the rise in liability claim costs beyond general economic inflation, has become a major concern for insurers and reinsurers, yet it is difficult to quantify because litigation outcomes are heavy-tailed and the mix of cases reaching verdict versus settlement changes over time. Using a large database of US jury verdicts and settlements, we develop case-mix-adjusted social inflation measures through multiple channels that matter to reinsurers: plaintiff win rates (a frequency-type channel), settlement propensity (a frequency-type channel), and verdict/settlement severity. The approach combines rolling-window logistic regression for probabilities and quantile (value-at-risk) regression for severities, with uncertainty quantified via a random-weighted bootstrap. We find statistically significant relative increases in plaintiff win probability of approximately 20%-30% from 2009 to 2024, alongside a statistically significant relative decline in settlement probability of more than 10% over the same period. The dominant channel is verdict severity: Even after controlling for explanatory variables, verdict awards show a sharp rise after 2020, increasing by more than 100% from 2020 to 2024, whereas settlement amounts show limited and often statistically insignificant inflation. Therefore, inflation in total amounts payable to plaintiffs closely tracks verdict severity. Social inflation is more pronounced in corporate-defendant and uninsured-defendant cases and in states without tort caps or third-party litigation funding regulation. In addition, we find that social inflation has impacts not only on "nuclear verdicts" but also, in a similar manner, on moderate losses.

econ.GN↗
arXiv2026

Qwen-Image-VAE-2.0 Technical Report

We present Qwen-Image-VAE-2.0, a suite of high-compression Variational Autoencoders (VAEs) that achieve significant advances in both reconstruction fidelity and diffusability. To address the reconstruction bottlenecks of high compression, we adopt an improved architecture featuring Global Skip Connections (GSC) and expanded latent channels. Moreover, we scale training to billions of images and incorporate a synthetic rendering engine to improve performance in text-rich scenarios. To tackle the convergence challenges of high-dimensional latent space, we implement an enhanced semantic alignment strategy to make the latent space highly amenable to diffusion modeling. To optimize computational efficiency, we leverage an asymmetric and attention-free encoder-decoder backbone to minimize encoding overhead. We present a comprehensive evaluation of Qwen-Image-VAE-2.0 on public reconstruction benchmarks. To evaluate performance in text-rich scenarios, we propose OmniDoc-TokenBench, a new benchmark comprising a diverse collection of real-world documents coupled with specialized OCR-based evaluation metrics. Qwen-Image-VAE-2.0 achieves state-of-the-art reconstruction performance, demonstrating exceptional capabilities in both general domains and text-rich scenarios at high compression ratio. Furthermore, downstream DiT experiments reveal our models possess superior diffusability, significantly accelerating convergence compared to existing high-compression baselines. These establish Qwen-Image-VAE-2.0 as a leading model with high compression, superior reconstruction, and exceptional diffusability.

cs.CV↗
arXiv2026

Qwen-Image-2.0 Technical Report

We present Qwen-Image-2.0, an omni-capable image generation foundation model that unifies high-fidelity generation and precise image editing within a single framework. Despite recent progress, existing models still struggle with ultra-long text rendering, multilingual typography, high-resolution photorealism, robust instruction following, and efficient deployment, especially in text-rich and compositionally complex scenarios. Qwen-Image-2.0 addresses these challenges by coupling Qwen3-VL as the condition encoder with a Multimodal Diffusion Transformer for joint condition-target modeling, supported by large-scale data curation and a customized multi-stage training pipeline. This enables strong multimodal understanding while preserving flexible generation and editing capabilities. The model supports instructions of up to 1K tokens for generating text-rich content such as slides, posters, infographics, and comics, while significantly improving multilingual text fidelity and typography. It also enhances photorealistic generation with richer details, more realistic textures, and coherent lighting, and follows complex prompts more reliably across diverse styles. Extensive human evaluations show that Qwen-Image-2.0 substantially outperforms previous Qwen-Image models in both generation and editing, marking a step toward more general, reliable, and practical image generation foundation models.

cs.CV↗
arXiv2026

Physics-Grounded Understanding of Thermal Boundary Conductance between Ga$_2$O$_3$ and SiC from a Feedforward Neural Network Potential

Ga$_2$O$_3$/SiC heterointegration is attractive for ultra-wide-bandgap power electronics, but interfacial thermal boundary conductance (TBC) remains a major heat-removal bottleneck. Direct experimental access to intrinsic atomistic interfacial transport remains limited, particularly for ideally synthesized materials with defect-free interfacial contact. First-principles simulations are too expensive at relevant length and time scales, while empirical Molecular Dynamics (MD) potentials often lack transferability across oxide and carbide bonding environments. We develop a unified feedforward neural network potential and validate it against density-functional data, bulk phonon dispersions, and anisotropic thermal-conductivity trends in both $\beta$-Ga$_2$O$_3$ and SiC. Nonequilibrium simulations show that TBC decreases with transport length, increases with temperature, and is consistently higher for Ga$_2$O$_3$$(\bar{2}01)$/SiC(0001) than for Ga$_2$O$_3$(100)/SiC(0001). These trends are explained by attenuation of long-mean-free-path carriers, enhanced incoherent and anharmonic interfacial exchange within broadly unchanged spectral channels, and stronger bonding and vibrational coupling at the $(\bar{2}01)$ interface. The results show how a single transferable feedforward neural network potential can enable large-scale transport prediction and physics-grounded mechanistic understanding of thermal boundary conductance. Code for NEP training and simulation workflows is available at the project repository https://github.com/knowhow07/TBC_Ga2O3_SiC.git

cond-mat.mtrl-sci↗
arXiv2026

Towards Visual Query Localization in the 3D World

Visual query localization (VQL) aims to predict the spatio-temporal response of the most recent occurrence in a sequence given a query. Currently, most research focuses on visual query localization in 2D videos, while its counterpart in 3D space has received little attention. In this paper, we make the first attempt to address visual query localization in the 3D world by introducing a novel benchmark, dubbed 3DVQL. Specifically, 3DVQL contains 2,002 sequences with around 170,000 frames and 6.4K response track segments from 38 object categories. Each sequence in 3DVQL is provided with multiple modalities, including point clouds, RGB images, and depth images, to support flexible research. To ensure high-quality annotations, each sequence is manually annotated with multiple rounds of verification and refinement. To the best of our knowledge, 3DVQL is the first benchmark for 3D multimodal visual query localization. To facilitate comparison in subsequent research, we implement a series of representative 3D multimodal VQL baselines using point clouds and RGB images. The experimental results show that existing methods exhibit significant performance variations across different fusion modules. To encourage future research, we propose a lift-and-attention fusion algorithm named LaF, which significantly outperforms existing baseline models. Our benchmark and model will be publicly released at https://github.com/wuhengliangliang/3DVQL.

cs.CV↗
arXiv2026

Project Imaging-X: A Survey of 1000+ Open-Access Medical Imaging Datasets for Foundation Model Development

Foundation models have demonstrated remarkable success across diverse domains and tasks, primarily due to the thrive of large-scale, diverse, and high-quality datasets. However, in the field of medical imaging, the curation and assembling of such medical datasets are highly challenging due to the reliance on clinical expertise and strict ethical and privacy constraints, resulting in a scarcity of large-scale unified medical datasets and hindering the development of powerful medical foundation models. In this work, we present the largest survey to date of medical image datasets, covering over 1,000 open-access datasets with a systematic catalog of their modalities, tasks, anatomies, annotations, limitations, and potential for integration. Our analysis exposes a landscape that is modest in scale, fragmented across narrowly scoped tasks, and unevenly distributed across organs and modalities, which in turn limits the utility of existing medical image datasets for developing versatile and robust medical foundation models. To turn fragmentation into scale, we propose a metadata-driven fusion paradigm (MDFP) that integrates public datasets with shared modalities or tasks, thereby transforming multiple small data silos into larger, more coherent resources. Building on MDFP, we release an interactive discovery portal that enables end-to-end, automated medical image dataset integration, and compile all surveyed datasets into a unified, structured table that clearly summarizes their key characteristics and provides reference links, offering the community an accessible and comprehensive repository. By charting the current terrain and offering a principled path to dataset consolidation, our survey provides a practical roadmap for scaling medical imaging corpora, supporting faster data discovery, more principled dataset creation, and more capable medical foundation models.

cs.CV↗
arXiv2026

MedTextWeaver: Procedural Knowledge Evolution in Agentic Medical Text Editing

Medical text editing is essential for improving communication among diverse stakeholders in clinical settings. However, adapting LLM agents to this task remains challenging because expert supervision is often sparse, fragmented, and distributed across interacting quality dimensions. We identify that direct accumulation or retrieval of individual feedback is insufficient for effective adaptation, as fragmented evaluations do not directly translate into a coherent understanding of medical text quality. Based on this observation, we propose MedTextWeaver, a training-free framework that transforms fragmented evaluative evidence into global quality principles and actionable procedural knowledge for medical text editing. Across three clinical text datasets and a real-world validation experiment, MedTextWeaver consistently improves performance over strong LLM baselines and existing memory-based adaptation approaches. Further analysis demonstrates that the learned knowledge enables more effective adaptation under limited supervision while providing an explicit and interpretable interface between expert evaluations and LLM editing behavior.

cs.CL↗
arXiv2026

RISER: Orchestrating Latent Reasoning Skills for Adaptive Activation Steering

Recent work on domain-specific reasoning with large language models (LLMs) often relies on training-intensive approaches that require parameter updates. While activation steering has emerged as a parameter efficient alternative, existing methods apply static, manual interventions that fail to adapt to the dynamic nature of complex reasoning. To address this limitation, we propose RISER (Router-based Intervention for Steerable Enhancement of Reasoning), a plug-and-play intervention framework that adaptively steers LLM reasoning in activation space. RISER constructs a library of reusable reasoning vectors and employs a lightweight Router to dynamically compose them for each input. The Router is optimized via reinforcement learning under task-level rewards, activating latent cognitive primitives in an emergent and compositional manner. Across seven diverse benchmarks, RISER yields 3.4-6.5% average zero-shot accuracy improvements over the base model while surpassing CoT-style reasoning with 2-3x higher token efficiency and robust accuracy gains. Further analysis shows that RISER autonomously combines multiple vectors into interpretable, precise control strategies, pointing toward more controllable and efficient LLM reasoning.

cs.AI↗
arXiv2025

SMART: Semantic Matching Contrastive Learning for Partially View-Aligned Clustering

Multi-view clustering has been empirically shown to improve learning performance by leveraging the inherent complementary information across multiple views of data. However, in real-world scenarios, collecting strictly aligned views is challenging, and learning from both aligned and unaligned data becomes a more practical solution. Partially View-aligned Clustering aims to learn correspondences between misaligned view samples to better exploit the potential consistency and complementarity across views, including both aligned and unaligned data. However, most existing PVC methods fail to leverage unaligned data to capture the shared semantics among samples from the same cluster. Moreover, the inherent heterogeneity of multi-view data induces distributional shifts in representations, leading to inaccuracies in establishing meaningful correspondences between cross-view latent features and, consequently, impairing learning effectiveness. To address these challenges, we propose a Semantic MAtching contRasTive learning model (SMART) for PVC. The main idea of our approach is to alleviate the influence of cross-view distributional shifts, thereby facilitating semantic matching contrastive learning to fully exploit semantic relationships in both aligned and unaligned data. Extensive experiments on eight benchmark datasets demonstrate that our method consistently outperforms existing approaches on the PVC problem.

cs.CV↗
arXiv2025

Trustworthy Neighborhoods Mining: Homophily-Aware Neutral Contrastive Learning for Graph Clustering

Recently, neighbor-based contrastive learning has been introduced to effectively exploit neighborhood information for clustering. However, these methods rely on the homophily assumption-that connected nodes share similar class labels and should therefore be close in feature space-which fails to account for the varying homophily levels in real-world graphs. As a result, applying contrastive learning to low-homophily graphs may lead to indistinguishable node representations due to unreliable neighborhood information, making it challenging to identify trustworthy neighborhoods with varying homophily levels in graph clustering. To tackle this, we introduce a novel neighborhood Neutral Contrastive Graph Clustering method, NeuCGC, that extends traditional contrastive learning by incorporating neutral pairs-node pairs treated as weighted positive pairs, rather than strictly positive or negative. These neutral pairs are dynamically adjusted based on the graph's homophily level, enabling a more flexible and robust learning process. Leveraging neutral pairs in contrastive learning, our method incorporates two key components: (1) an adaptive contrastive neighborhood distribution alignment that adjusts based on the homophily level of the given attribute graph, ensuring effective alignment of neighborhood distributions, and (2) a contrastive neighborhood node feature consistency learning mechanism that leverages reliable neighborhood information from high-confidence graphs to learn robust node representations, mitigating the adverse effects of varying homophily levels and effectively exploiting highly trustworthy neighborhood information. Experimental results demonstrate the effectiveness and robustness of our approach, outperforming other state-of-the-art graph clustering methods. Our code is available at https://github.com/THPengL/NeuCGC.

cs.SI↗