arXiv subjects

Jia Wang

Publications and source records attributed to Jia Wang.

At least 73 records · Page 4Linked to original sources

arXiv2026

Video Compression with Hierarchical Temporal Neural Representation

Video compression has recently benefited from implicit neural representations (INRs), which model videos as continuous functions. INRs offer compact storage and flexible reconstruction, providing a promising alternative to traditional codecs. However, most existing INR-based methods treat the temporal dimension as an independent input, limiting their ability to capture complex temporal dependencies. To address this, we propose a Hierarchical Temporal Neural Representation for Videos, TeNeRV. TeNeRV integrates short- and long-term dependencies through two key components. First, an Inter-Frame Feature Fusion (IFF) module aggregates features from adjacent frames, enforcing local temporal coherence and capturing fine-grained motion. Second, a GoP-Adaptive Modulation (GAM) mechanism partitions videos into Groups-of-Pictures and learns group-specific priors. The mechanism modulates network parameters, enabling adaptive representations across different GoPs. Extensive experiments demonstrate that TeNeRV consistently outperforms existing INR-based methods in rate-distortion performance, validating the effectiveness of our proposed approach.

cs.CV↗

arXiv2026

SSPFormer: Self-Supervised Pretrained Transformer for MRI Images

The pre-trained transformer demonstrates remarkable generalization ability in natural image processing. However, directly transferring it to magnetic resonance images faces two key challenges: the inability to adapt to the specificity of medical anatomical structures and the limitations brought about by the privacy and scarcity of medical data. To address these issues, this paper proposes a Self-Supervised Pretrained Transformer (SSPFormer) for MRI images, which effectively learns domain-specific feature representations of medical images by leveraging unlabeled raw imaging data. To tackle the domain gap and data scarcity, we introduce inverse frequency projection masking, which prioritizes the reconstruction of high-frequency anatomical regions to enforce structure-aware representation learning. Simultaneously, to enhance robustness against real-world MRI artifacts, we employ frequency-weighted FFT noise enhancement that injects physiologically realistic noise into the Fourier domain. Together, these strategies enable the model to learn domain-invariant and artifact-robust features directly from raw scans. Through extensive experiments on segmentation, super-resolution, and denoising tasks, the proposed SSPFormer achieves state-of-the-art performance, fully verifying its ability to capture fine-grained MRI image fidelity and adapt to clinical application requirements.

cs.CV↗

arXiv2026

STEP3-VL-10B Technical Report

We present STEP3-VL-10B, a lightweight open-source foundation model designed to redefine the trade-off between compact efficiency and frontier-level multimodal intelligence. STEP3-VL-10B is realized through two strategic shifts: first, a unified, fully unfrozen pre-training strategy on 1.2T multimodal tokens that integrates a language-aligned Perception Encoder with a Qwen3-8B decoder to establish intrinsic vision-language synergy; and second, a scaled post-training pipeline featuring over 1k iterations of reinforcement learning. Crucially, we implement Parallel Coordinated Reasoning (PaCoRe) to scale test-time compute, allocating resources to scalable perceptual reasoning that explores and synthesizes diverse visual hypotheses. Consequently, despite its compact 10B footprint, STEP3-VL-10B rivals or surpasses models 10$\times$-20$\times$ larger (e.g., GLM-4.6V-106B, Qwen3-VL-235B) and top-tier proprietary flagships like Gemini 2.5 Pro and Seed-1.5-VL. Delivering best-in-class performance, it records 92.2% on MMBench and 80.11% on MMMU, while excelling in complex reasoning with 94.43% on AIME2025 and 75.95% on MathVision. We release the full model suite to provide the community with a powerful, efficient, and reproducible baseline.

cs.CV↗

arXiv2026

Enhancing Image Quality Assessment Ability of LMMs via Retrieval-Augmented Generation

Large Multimodal Models (LMMs) have recently shown remarkable promise in low-level visual perception tasks, particularly in Image Quality Assessment (IQA), demonstrating strong zero-shot capability. However, achieving state-of-the-art performance often requires computationally expensive fine-tuning methods, which aim to align the distribution of quality-related token in output with image quality levels. Inspired by recent training-free works for LMM, we introduce IQARAG, a novel, training-free framework that enhances LMMs' IQA ability. IQARAG leverages Retrieval-Augmented Generation (RAG) to retrieve some semantically similar but quality-variant reference images with corresponding Mean Opinion Scores (MOSs) for input image. These retrieved images and input image are integrated into a specific prompt. Retrieved images provide the LMM with a visual perception anchor for IQA task. IQARAG contains three key phases: Retrieval Feature Extraction, Image Retrieval, and Integration & Quality Score Generation. Extensive experiments across multiple diverse IQA datasets, including KADID, KonIQ, LIVE Challenge, and SPAQ, demonstrate that the proposed IQARAG effectively boosts the IQA performance of LMMs, offering a resource-efficient alternative to fine-tuning for quality assessment.

cs.CV↗

arXiv2026

Forge-and-Quench: Enhancing Image Generation for Higher Fidelity in Unified Multimodal Models

Integrating image generation and understanding into a single framework has become a pivotal goal in the multimodal domain. However, how understanding can effectively assist generation has not been fully explored. Unlike previous works that focus on leveraging reasoning abilities and world knowledge from understanding models, this paper introduces a novel perspective: leveraging understanding to enhance the fidelity and detail richness of generated images. To this end, we propose Forge-and-Quench, a new unified framework that puts this principle into practice. In the generation process of our framework, an MLLM first reasons over the entire conversational context, including text instructions, to produce an enhanced text instruction. This refined instruction is then mapped to a virtual visual representation, termed the Bridge Feature, via a novel Bridge Adapter. This feature acts as a crucial link, forging insights from the understanding model to quench and refine the generation process. It is subsequently injected into the T2I backbone as a visual guidance signal, alongside the enhanced text instruction that replaces the original input. To validate this paradigm, we conduct comprehensive studies on the design of the Bridge Feature and Bridge Adapter. Our framework demonstrates exceptional extensibility and flexibility, enabling efficient migration across different MLLM and T2I models with significant savings in training overhead, all without compromising the MLLM's inherent multimodal understanding capabilities. Experiments show that Forge-and-Quench significantly improves image fidelity and detail across multiple models, while also maintaining instruction-following accuracy and enhancing world knowledge application. Models and codes are available at https://github.com/YanbingZeng/Forge-and-Quench.

cs.CV↗

arXiv2025

Achieving High Efficiency And Enhanced Beam Quality In Laser Wakefield Acceleration

Laser wakefield acceleration, characterized by the extremely high electric field gradient exceeding 100GV/m, is regarded as a compact and cost affordable technology for the next generation of particle colliders and light sources. However, it has always been a major challenge to effectively increase the energy transfer efficiency from the laser to the accelerated beam, while ensuring the beam quality remains suitable for practical applications. This study demonstrates that the laser with shorter pulse duration allows for a two-step dechirping process of the accelerated electron beam with charge of nanocoulomb level. The electron beams with an energy spread of 1% can be generated with the energy transfer efficiency of 10% to 30% in a large parameter space. For example, one electron beam with the energy of 420MeV, the charge of 5.5nC and the RMS energy spread of 2% can be produced using an 8.3J laser pulse with 7.2fs duration.

physics.acc-ph↗

arXiv2025

Step-GUI Technical Report

Recent advances in multimodal large language models unlock unprecedented opportunities for GUI automation. However, a fundamental challenge remains: how to efficiently acquire high-quality training data while maintaining annotation reliability? We introduce a self-evolving training pipeline powered by the Calibrated Step Reward System, which converts model-generated trajectories into reliable training signals through trajectory-level calibration, achieving >90% annotation accuracy with 10-100x lower cost. Leveraging this pipeline, we introduce Step-GUI, a family of models (4B/8B) that achieves state-of-the-art GUI performance (8B: 80.2% AndroidWorld, 48.5% OSWorld, 62.6% ScreenShot-Pro) while maintaining robust general capabilities. As GUI agent capabilities improve, practical deployment demands standardized interfaces across heterogeneous devices while protecting user privacy. To this end, we propose GUI-MCP, the first Model Context Protocol for GUI automation with hierarchical architecture that combines low-level atomic operations and high-level task delegation to local specialist models, enabling high-privacy execution where sensitive data stays on-device. Finally, to assess whether agents can handle authentic everyday usage, we introduce AndroidDaily, a benchmark grounded in real-world mobile usage patterns with 3146 static actions and 235 end-to-end tasks across high-frequency daily scenarios (8B: static 89.91%, end-to-end 52.50%). Our work advances the development of practical GUI agents and demonstrates strong potential for real-world deployment in everyday digital interactions.

cs.CV↗

arXiv2025

Efficient Candidate-Free R-S Set Similarity Joins with Filter-and-Verification Trees on MapReduce

Given two different collections of sets R and S, the exact R-S set similarity join (R-S Join) finds all set pairs with similarity no less than a given threshold, which has widespread applications. Existing algorithms accelerate large-scale R-S Joins using a two-stage filter-and-verification framework along with the parallel and distributed MapReduce framework, however, they suffer from excessive candidate set pairs (candidates), leading to significant I/O and verification overhead. This paper proposes novel candidate-free R-S Join (CF-RS-Join) algorithms that integrate filtering and verification into a single stage through the filter-and-verification tree (FVT) and its linear variant (LFVT). First, CF-RS-Join with FVT (CF-RS-Join/FVT) is proposed to leverage an innovative FVT structure that compresses elements and associated sets in memory, enabling single-stage processing that eliminates candidate generation, enables fast lookups, and reduces database scans. Correctness proofs are provided. Second, CF-RS-Join with LFVT (CF-RS-Join/LFVT) is proposed to exploit a more compact Linear FVT, which compresses non-branching paths into single nodes and stores them in linear arrays for optimized traversal. Third, MR-CF-RS-Join/FVT and MR-CF-RS-Join/LFVT are proposed to extend our approaches using MapReduce for parallel processing. Extensive experiments have been conducted on the proposed algorithms against state-of-the-art (SOTA) baselines in terms of execution time, scalability, memory usage, and disk usage. The results show that MR-CF-RS-Join/LFVT outperforms the runner-up by up to 1.37x-15.78x on 7 real-world datasets.

cs.DC↗

arXiv2025

GUI Exploration Lab: Enhancing Screen Navigation in Agents via Multi-Turn Reinforcement Learning

With the rapid development of Large Vision Language Models, the focus of Graphical User Interface (GUI) agent tasks shifts from single-screen tasks to complex screen navigation challenges. However, real-world GUI environments, such as PC software and mobile Apps, are often complex and proprietary, making it difficult to obtain the comprehensive environment information needed for agent training and evaluation. This limitation hinders systematic investigation and benchmarking of agent navigation capabilities. To address this limitation, we introduce GUI Exploration Lab, a simulation environment engine for GUI agent navigation research that enables flexible definition and composition of screens, icons, and navigation graphs, while providing full access to environment information for comprehensive agent training and evaluation. Through extensive experiments, we find that supervised fine-tuning enables effective memorization of fundamental knowledge, serving as a crucial foundation for subsequent training. Building on this, single-turn reinforcement learning further enhances generalization to unseen scenarios. Finally, multi-turn reinforcement learning encourages the development of exploration strategies through interactive trial and error, leading to further improvements in screen navigation performance. We validate our methods on both static and interactive benchmarks, demonstrating that our findings generalize effectively to real-world scenarios. These findings demonstrate the advantages of reinforcement learning approaches in GUI navigation and offer practical guidance for building more capable and generalizable GUI agents.

cs.CV↗

arXiv2025

Beyond Darkness: Thermal-Supervised 3D Gaussian Splatting for Low-Light Novel View Synthesis

Under extremely low-light conditions, novel view synthesis (NVS) faces severe degradation in terms of geometry, color consistency, and radiometric stability. Standard 3D Gaussian Splatting (3DGS) pipelines fail when applied directly to underexposed inputs, as independent enhancement across views causes illumination inconsistencies and geometric distortion. To address this, we present DTGS, a unified framework that tightly couples Retinex-inspired illumination decomposition with thermal-guided 3D Gaussian Splatting for illumination-invariant reconstruction. Unlike prior approaches that treat enhancement as a pre-processing step, DTGS performs joint optimization across enhancement, geometry, and thermal supervision through a cyclic enhancement-reconstruction mechanism. A thermal supervisory branch stabilizes both color restoration and geometry learning by dynamically balancing enhancement, structural, and thermal losses. Moreover, a Retinex-based decomposition module embedded within the 3DGS loop provides physically interpretable reflectance-illumination separation, ensuring consistent color and texture across viewpoints. To evaluate our method, we construct RGBT-LOW, a new multi-view low-light thermal dataset capturing severe illumination degradation. Extensive experiments show that DTGS significantly outperforms existing low-light enhancement and 3D reconstruction baselines, achieving superior radiometric consistency, geometric fidelity, and color stability under extreme illumination.

cs.CV↗

arXiv2025

From IDs to Semantics: A Generative Framework for Cross-Domain Recommendation with Adaptive Semantic Tokenization

Cross-domain recommendation (CDR) is crucial for improving recommendation accuracy and generalization, yet traditional methods are often hindered by the reliance on shared user/item IDs, which are unavailable in most real-world scenarios. Consequently, many efforts have focused on learning disentangled representations through multi-domain joint training to bridge the domain gaps. Recent Large Language Model (LLM)-based approaches show promise, they still face critical challenges, including: (1) the \textbf{item ID tokenization dilemma}, which leads to vocabulary explosion and fails to capture high-order collaborative knowledge; and (2) \textbf{insufficient domain-specific modeling} for the complex evolution of user interests and item semantics. To address these limitations, we propose \textbf{GenCDR}, a novel \textbf{Gen}erative \textbf{C}ross-\textbf{D}omain \textbf{R}ecommendation framework. GenCDR first employs a \textbf{Domain-adaptive Tokenization} module, which generates disentangled semantic IDs for items by dynamically routing between a universal encoder and domain-specific adapters. Symmetrically, a \textbf{Cross-domain Autoregressive Recommendation} module models user preferences by fusing universal and domain-specific interests. Finally, a \textbf{Domain-aware Prefix-tree} enables efficient and accurate generation. Extensive experiments on multiple real-world datasets demonstrate that GenCDR significantly outperforms state-of-the-art baselines. Our code is available in the supplementary materials.

cs.IR↗

arXiv2025

Data Assessment for Embodied Intelligence

In embodied intelligence, datasets play a pivotal role, serving as both a knowledge repository and a conduit for information transfer. The two most critical attributes of a dataset are the amount of information it provides and how easily this information can be learned by models. However, the multimodal nature of embodied data makes evaluating these properties particularly challenging. Prior work has largely focused on diversity, typically counting tasks and scenes or evaluating isolated modalities, which fails to provide a comprehensive picture of dataset diversity. On the other hand, the learnability of datasets has received little attention and is usually assessed post-hoc through model training, an expensive, time-consuming process that also lacks interpretability, offering little guidance on how to improve a dataset. In this work, we address both challenges by introducing two principled, data-driven tools. First, we construct a unified multimodal representation for each data sample and, based on it, propose diversity entropy, a continuous measure that characterizes the amount of information contained in a dataset. Second, we introduce the first interpretable, data-driven algorithm to efficiently quantify dataset learnability without training, enabling researchers to assess a dataset's learnability immediately upon its release. We validate our algorithm on both simulated and real-world embodied datasets, demonstrating that it yields faithful, actionable insights that enable researchers to jointly improve diversity and learnability. We hope this work provides a foundation for designing higher-quality datasets that advance the development of embodied intelligence.

cs.RO↗

arXiv2025

PrediQL: Automated Testing of GraphQL APIs with LLMs

GraphQL's flexible query model and nested data dependencies expose APIs to complex, context-dependent vulnerabilities that are difficult to uncover using conventional testing tools. Existing fuzzers either rely on random payload generation or rigid mutation heuristics, failing to adapt to the dynamic structures of GraphQL schemas and responses. We present PrediQL, the first retrieval-augmented, LLM-guided fuzzer for GraphQL APIs. PrediQL combines large language model reasoning with adaptive feedback loops to generate semantically valid and diverse queries. It models the choice of fuzzing strategy as a multi-armed bandit problem, balancing exploration of new query structures with exploitation of past successes. To enhance efficiency, PrediQL retrieves and reuses execution traces, schema fragments, and prior errors, enabling self-correction and progressive learning across test iterations. Beyond input generation, PrediQL integrates a context-aware vulnerability detector that uses LLM reasoning to analyze responses, interpreting data values, error messages, and status codes to identify issues such as injection flaws, access-control bypasses, and information disclosure. Our evaluation across open-source and benchmark GraphQL APIs shows that PrediQL achieves significantly higher coverage and vulnerability discovery rates compared to state-of-the-art baselines. These results demonstrate that combining retrieval-augmented reasoning with adaptive fuzzing can transform API security testing from reactive enumeration to intelligent exploration.

cs.CR↗

arXiv2025

SocioBench: Modeling Human Behavior in Sociological Surveys with Large Language Models

Large language models (LLMs) show strong potential for simulating human social behaviors and interactions, yet lack large-scale, systematically constructed benchmarks for evaluating their alignment with real-world social attitudes. To bridge this gap, we introduce SocioBench-a comprehensive benchmark derived from the annually collected, standardized survey data of the International Social Survey Programme (ISSP). The benchmark aggregates over 480,000 real respondent records from more than 30 countries, spanning 10 sociological domains and over 40 demographic attributes. Our experiments indicate that LLMs achieve only 30-40% accuracy when simulating individuals in complex survey scenarios, with statistically significant differences across domains and demographic subgroups. These findings highlight several limitations of current LLMs in survey scenarios, including insufficient individual-level data coverage, inadequate scenario diversity, and missing group-level modeling.

cs.SI↗

arXiv2025

Hyperspherical Analysis of Dimer-Dimer Scattering in One-Dimensional Systems

We present a comprehensive analysis of four-body scattering in one-dimensional (1D) quantum systems using the adiabatic hyperspherical representation (AHR). Focusing on dimer-dimer collisions between two species of fermions interacting via the sinh-cosh potential, we implement the slow variable discretization (SVD) method to overcome numerical challenges posed by sharp avoided crossings in the potential curves. Our numerical approach is benchmarked against exact analytical results available in integrable regimes, demonstrating excellent agreement. We further explore non-integrable regimes where no analytical solutions exist, revealing novel features such as resonant enhancement of the scattering length associated with tetramer formation. These results highlight the power and flexibility of the AHR+SVD framework for accurate few-body scattering calculations in low-dimensional quantum systems, and establish a foundation for future investigations of universal few-body physics in ultracold gases.

cond-mat.quant-gas↗

arXiv2025

Formation of thiocarbonic acid (H$_2$CS$_3$) -- the sulfur counterpart of carbonic acid (H$_2$CO$_3$) - in interstellar analog ices

The first experimental formation of thiocarbonic acid (H$_2$CO$_3$) is presented in this work from low-temperature interstellar ice analogs composed of hydrogen sulfide (H$_2$S) and carbon disulfide (CS$_2$) exposed to electron irradiation simulating the impact of galactic cosmic rays (GCRs) on interstellar ices. The recent attention brought to sulfur-bearing molecules, as well as the recent detection of carbonic acid (H$_2$CO$_3$) in the interstellar medium (ISM), invites the study of the interstellar detection of the sulfur counterpart, thiocarbonic acid. However, the interstellar formation pathways of thiocarbonic acid have remained elusive. In this work, thiocarbonic acid was identified in the gas phase during the temperature programmed desorption (TPD) using isomer-selective single photoionization reflectron time-of-flight mass spectrometry (PI-ReToF-MS), suggesting that the hitherto astronomically unobserved thiocarbonic acid represents a promising candidate for future astronomical searches. The formation of H2CS3 isomers was investigated through additional isotopically labeled experiments and the formation mechanisms through quantum chemical studies. These findings unravel a key reaction pathway to thiocarbonic acid and represent a first step toward its possible formation and detection in the ISM, shedding light on the missing sulfur problem.

astro-ph.GA↗

arXiv2025

Open Vision Reasoner: Transferring Linguistic Cognitive Behavior for Visual Reasoning

The remarkable reasoning capability of large language models (LLMs) stems from cognitive behaviors that emerge through reinforcement with verifiable rewards. This work investigates how to transfer this principle to Multimodal LLMs (MLLMs) to unlock advanced visual reasoning. We introduce a two-stage paradigm built on Qwen2.5-VL-7B: a massive linguistic cold-start fine-tuning, followed by multimodal reinforcement learning (RL) spanning nearly 1,000 steps, surpassing all previous open-source efforts in scale. This pioneering work reveals three fundamental insights: 1) Behavior transfer emerges surprisingly early in cold start due to linguistic mental imagery. 2) Cold start broadly memorizes visual behaviors, while RL critically discerns and scales up effective patterns. 3) Transfer strategically favors high-utility behaviors such as visual reflection. Our resulting model, Open-Vision-Reasoner (OVR), achieves state-of-the-art performance on a suite of reasoning benchmarks, including 95.3% on MATH500, 51.8% on MathVision and 54.6% on MathVerse. We release our model, data, and training dynamics to catalyze the development of more capable, behavior-aligned multimodal reasoners.

cs.CV↗

arXiv2025

VRScout: Towards Real-Time, Autonomous Testing of Virtual Reality Games

Virtual Reality (VR) has rapidly become a mainstream platform for gaming and interactive experiences, yet ensuring the quality, safety, and appropriateness of VR content remains a pressing challenge. Traditional human-based quality assurance is labor-intensive and cannot scale with the industry's rapid growth. While automated testing has been applied to traditional 2D and 3D games, extending it to VR introduces unique difficulties due to high-dimensional sensory inputs and strict real-time performance requirements. We present VRScout, a deep learning-based agent capable of autonomously navigating VR environments and interacting with virtual objects in a human-like and real-time manner. VRScout learns from human demonstrations using an enhanced Action Chunking Transformer that predicts multi-step action sequences. This enables our agent to capture higher-level strategies and generalize across diverse environments. To balance responsiveness and precision, we introduce a dynamically adjustable sliding horizon that adapts the agent's temporal context at runtime. We evaluate VRScout on commercial VR titles and show that it achieves expert-level performance with only limited training data, while maintaining real-time inference at 60 FPS on consumer-grade hardware. These results position VRScout as a practical and scalable framework for automated VR game testing, with direct applications in both quality assurance and safety auditing.

cs.LG↗