arXiv subjects

Zijian Zhang

Publications and source records attributed to Zijian Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2025

Solution for Meta KDD Cup'25: A Comprehensive Three-Step Framework for Vision Question Answering

Vision Large Language Models (VLLMs) have improved multi-modal understanding and visual question answering (VQA), but still suffer from hallucinated answers. Multi-modal Retrieval-Augmented Generation (RAG) helps address these issues by incorporating external information, yet challenges remain in visual context comprehension, multi-source retrieval, and multi-turn interactions. To address these challenges, Meta constructed the CRAG-MM benchmark and launched the CRAG-MM Challenge at KDD Cup 2025, which consists of three tasks. This paper describes the solutions of all tasks in Meta KDD Cup'25 from BlackPearl team. We use a single model for each task, with key methods including data augmentation, RAG, reranking, and multi-task fine-tuning. Our solution achieve automatic evaluation rankings of 3rd, 3rd, and 1st on the three tasks, and win second place in Task3 after human evaluation.

cs.IR↗

arXiv2025

TreeReader: A Hierarchical Academic Paper Reader Powered by Language Models

Efficiently navigating and understanding academic papers is crucial for scientific progress. Traditional linear formats like PDF and HTML can cause cognitive overload and obscure a paper's hierarchical structure, making it difficult to locate key information. While LLM-based chatbots offer summarization, they often lack nuanced understanding of specific sections, may produce unreliable information, and typically discard the document's navigational structure. Drawing insights from a formative study on academic reading practices, we introduce TreeReader, a novel language model-augmented paper reader. TreeReader decomposes papers into an interactive tree structure where each section is initially represented by an LLM-generated concise summary, with underlying details accessible on demand. This design allows users to quickly grasp core ideas, selectively explore sections of interest, and verify summaries against the source text. A user study was conducted to evaluate TreeReader's impact on reading efficiency and comprehension. TreeReader provides a more focused and efficient way to navigate and understand complex academic literature by bridging hierarchical summarization with interactive exploration.

cs.HC↗

arXiv2025

CWNet: Causal Wavelet Network for Low-Light Image Enhancement

Traditional Low-Light Image Enhancement (LLIE) methods primarily focus on uniform brightness adjustment, often neglecting instance-level semantic information and the inherent characteristics of different features. To address these limitations, we propose CWNet (Causal Wavelet Network), a novel architecture that leverages wavelet transforms for causal reasoning. Specifically, our approach comprises two key components: 1) Inspired by the concept of intervention in causality, we adopt a causal reasoning perspective to reveal the underlying causal relationships in low-light enhancement. From a global perspective, we employ a metric learning strategy to ensure causal embeddings adhere to causal principles, separating them from non-causal confounding factors while focusing on the invariance of causal factors. At the local level, we introduce an instance-level CLIP semantic loss to precisely maintain causal factor consistency. 2) Based on our causal analysis, we present a wavelet transform-based backbone network that effectively optimizes the recovery of frequency information, ensuring precise enhancement tailored to the specific attributes of wavelet transforms. Extensive experiments demonstrate that CWNet significantly outperforms current state-of-the-art methods across multiple datasets, showcasing its robust performance across diverse scenes. Code is available at https://github.com/bywlzts/CWNet-Causal-Wavelet-Network.

cs.CV↗

arXiv2025

Nonlinear dynamics of soliton molecules in a Kerr micro-ring

The optical Kerr micro-ring provides an ideal platform for the study of dissipative optical solitons. Dissipative solitons are localized waves produced by a precise equilibrium between dispersion and nonlinearity, as well as gain and loss. Dissipative brilliant solitons are vulnerable to external noise,but dissipative dark solitons exhibit greater robustness against noise and losses. This study discusses the division of the input pump light field into transverse electric (TE) and transverse magnetic(TM) modes. TE mode generates bright solitons. TM mode forms dark solitons via cross-phase modulation (XPM), which induces a self-focusing effect. The bright and dark solitons bind into a soliton molecule pair through mutual interactions. We scan the entire detuning interval; within the positive small detuning interval, two modes simultaneously generate bright-dark soliton pair Turing roll states. Within the larger detuning interval, the excited Brillouin scattering noise induces a high suppression ratio (SR) between the dark and bright soliton pumps, which is unfavorable for the formation of the bright-dark soliton pair. The system provides a new multi-soliton manipulation scheme for optical communications. The soliton molecule pairs hold major implications for high precision measurement and all-optical controlling fields.

physics.optics↗

arXiv2025

HKD4VLM: A Progressive Hybrid Knowledge Distillation Framework for Robust Multimodal Hallucination and Factuality Detection in VLMs

Driven by the rapid progress in vision-language models (VLMs), the responsible behavior of large-scale multimodal models has become a prominent research area, particularly focusing on hallucination detection and factuality checking. In this paper, we present the solution for the two tracks of Responsible AI challenge. Inspirations from the general domain demonstrate that a smaller distilled VLM can often outperform a larger VLM that is directly tuned on downstream tasks, while achieving higher efficiency. We thus jointly tackle two tasks from the perspective of knowledge distillation and propose a progressive hybrid knowledge distillation framework termed HKD4VLM. Specifically, the overall framework can be decomposed into Pyramid-like Progressive Online Distillation and Ternary-Coupled Refinement Distillation, hierarchically moving from coarse-grained knowledge alignment to fine-grained refinement. Besides, we further introduce the mapping shift-enhanced inference and diverse augmentation strategies to enhance model performance and robustness. Extensive experimental results demonstrate the effectiveness of our HKD4VLM. Ablation studies provide insights into the critical design choices driving performance gains.

cs.CV↗

arXiv2025

Agents for self-driving laboratories applied to quantum computing

Fully automated self-driving laboratories are promising to enable high-throughput and large-scale scientific discovery by reducing repetitive labour. However, effective automation requires deep integration of laboratory knowledge, which is often unstructured, multimodal, and difficult to incorporate into current AI systems. This paper introduces the k-agents framework, designed to support experimentalists in organizing laboratory knowledge and automating experiments with agents. Our framework employs large language model-based agents to encapsulate laboratory knowledge including available laboratory operations and methods for analyzing experiment results. To automate experiments, we introduce execution agents that break multi-step experimental procedures into agent-based state machines, interact with other agents to execute each step and analyze the experiment results. The analyzed results are then utilized to drive state transitions, enabling closed-loop feedback control. To demonstrate its capabilities, we applied the agents to calibrate and operate a superconducting quantum processor, where they autonomously planned and executed experiments for hours, successfully producing and characterizing entangled quantum states at the level achieved by human scientists. Our knowledge-based agent system opens up new possibilities for managing laboratory knowledge and accelerating scientific discovery.

cs.AI↗

arXiv2025

Adaptive Plan-Execute Framework for Smart Contract Security Auditing

Large Language Models (LLMs) have shown great promise in code analysis and auditing; however, they still struggle with hallucinations and limited context-aware reasoning. We introduce SmartAuditFlow, a novel Plan-Execute framework that enhances smart contract security analysis through dynamic audit planning and structured execution. Unlike conventional LLM-based auditing approaches that follow fixed workflows and predefined steps, SmartAuditFlow dynamically generates and refines audit plans based on the unique characteristics of each smart contract. It continuously adjusts its auditing strategy in response to intermediate LLM outputs and newly detected vulnerabilities, ensuring a more adaptive and precise security assessment. The framework then executes these plans step by step, applying a structured reasoning process to enhance vulnerability detection accuracy while minimizing hallucinations and false positives. To further improve audit precision, SmartAuditFlow integrates iterative prompt optimization and external knowledge sources, such as static analysis tools and Retrieval-Augmented Generation (RAG). This ensures audit decisions are contextually informed and backed by real-world security knowledge, producing comprehensive security reports. Extensive evaluations across multiple benchmarks demonstrate that SmartAuditFlow outperforms existing methods, achieving 100 percent accuracy on common and critical vulnerabilities, 41.2 percent accuracy for comprehensive coverage of known smart contract weaknesses in real-world projects, and successfully identifying all 13 tested CVEs. These results highlight SmartAuditFlow's scalability, cost-effectiveness, and superior adaptability over traditional static analysis tools and contemporary LLM-based approaches, establishing it as a robust solution for automated smart contract auditing.

cs.CR↗

arXiv2025

NTIRE 2025 challenge on Text to Image Generation Model Quality Assessment

This paper reports on the NTIRE 2025 challenge on Text to Image (T2I) generation model quality assessment, which will be held in conjunction with the New Trends in Image Restoration and Enhancement Workshop (NTIRE) at CVPR 2025. The aim of this challenge is to address the fine-grained quality assessment of text-to-image generation models. This challenge evaluates text-to-image models from two aspects: image-text alignment and image structural distortion detection, and is divided into the alignment track and the structural track. The alignment track uses the EvalMuse-40K, which contains around 40K AI-Generated Images (AIGIs) generated by 20 popular generative models. The alignment track has a total of 371 registered participants. A total of 1,883 submissions are received in the development phase, and 507 submissions are received in the test phase. Finally, 12 participating teams submitted their models and fact sheets. The structure track uses the EvalMuse-Structure, which contains 10,000 AI-Generated Images (AIGIs) with corresponding structural distortion mask. A total of 211 participants have registered in the structure track. A total of 1155 submissions are received in the development phase, and 487 submissions are received in the test phase. Finally, 8 participating teams submitted their models and fact sheets. Almost all methods have achieved better results than baseline methods, and the winning methods in both tracks have demonstrated superior prediction performance on T2I model quality assessment.

cs.CV↗

arXiv2025

Lyman-Break Galaxies in the Mpc-Scale Environments Around Three $z\sim7.5$ Quasars With JWST Imaging

We study the Mpc-scale environments of the three highest redshift luminous quasars at $z\geq 7.5$ (J031343.84-180636.40, J134208.11+092838.61, and J100758.27+211529.21) to understand their connection to large-scale structure. Cosmological simulations show that these early supermassive black holes (SMBHs) are expected to form in the most massive dark matter halos. Therefore, it is expected that they are anchors of galaxy overdensities if luminous matter traces the underlying dark matter structure of the Universe. Using JWST NIRCam (F090W/F115W/F250M/F360M/F430M) imaging, we observe the large-scale structure out to $\sim13$ comoving Mpc around these quasars. We select F090W-dropout Lyman Break galaxies (LBGs) and F430M-excess [OIII] emitters in the three fields. We find 18, 21, and 6 LBG candidates in the fields of J0313, J1342, and J1007, respectively, resulting in a wide range of overdensities ($1+δ\sim 19,\,24,$ and $7$). The photometric redshifts indicate serendipitous foreground and background overdensities in the J0313 field. The joint angular autocorrelation of the combined LBG sample shows significant clustering on $<1.8$ comoving Mpc scales, demonstrating that the selected galaxies are likely associated with the large-scale structure surrounding the quasars. This first systematic study of $z\sim 7.5$ quasars shows a diverse set of quasar environments at the onset of their formation, providing empirical data to help constrain theoretical predictions of early structure formation.

astro-ph.GA↗

arXiv2025

Anyprefer: An Agentic Framework for Preference Data Synthesis

High-quality preference data is essential for aligning foundation models with human values through preference learning. However, manual annotation of such data is often time-consuming and costly. Recent methods often adopt a self-rewarding approach, where the target model generates and annotates its own preference data, but this can lead to inaccuracies since the reward model shares weights with the target model, thereby amplifying inherent biases. To address these issues, we propose Anyprefer, a framework designed to synthesize high-quality preference data for aligning the target model. Anyprefer frames the data synthesis process as a cooperative two-player Markov Game, where the target model and the judge model collaborate together. Here, a series of external tools are introduced to assist the judge model in accurately rewarding the target model's responses, mitigating biases in the rewarding process. In addition, a feedback mechanism is introduced to optimize prompts for both models, enhancing collaboration and improving data quality. The synthesized data is compiled into a new preference dataset, Anyprefer-V1, consisting of 58K high-quality preference pairs. Extensive experiments show that Anyprefer significantly improves model alignment performance across four main applications, covering 21 datasets, achieving average improvements of 18.55% in five natural language generation datasets, 3.66% in nine vision-language understanding datasets, 30.05% in three medical image analysis datasets, and 16.00% in four visuo-motor control tasks.

cs.LG↗

arXiv2025

Bridge the Domains: Large Language Models Enhanced Cross-domain Sequential Recommendation

Cross-domain Sequential Recommendation (CDSR) aims to extract the preference from the user's historical interactions across various domains. Despite some progress in CDSR, two problems set the barrier for further advancements, i.e., overlap dilemma and transition complexity. The former means existing CDSR methods severely rely on users who own interactions on all domains to learn cross-domain item relationships, compromising the practicability. The latter refers to the difficulties in learning the complex transition patterns from the mixed behavior sequences. With powerful representation and reasoning abilities, Large Language Models (LLMs) are promising to address these two problems by bridging the items and capturing the user's preferences from a semantic view. Therefore, we propose an LLMs Enhanced Cross-domain Sequential Recommendation model (LLM4CDSR). To obtain the semantic item relationships, we first propose an LLM-based unified representation module to represent items. Then, a trainable adapter with contrastive regularization is designed to adapt the CDSR task. Besides, a hierarchical LLMs profiling module is designed to summarize user cross-domain preferences. Finally, these two modules are integrated into the proposed tri-thread framework to derive recommendations. We have conducted extensive experiments on three public cross-domain datasets, validating the effectiveness of LLM4CDSR. We have released the code online.

cs.IR↗

arXiv2025

Analysis of Multi-epoch JWST Images of $\sim 300$ Little Red Dots: Tentative Detection of Variability in a Minority of Sources

James Webb Space Telescope (JWST) has revealed a population of red and compact sources at $z \gtrsim 5$ known as ``Little Red Dots'' (LRDs) that are likely active galactic nuclei (AGNs). Here we present a comprehensive study of the variability of 314 LRDs with multi-epoch JWST observations in five deep fields: UDS, GOODS-S, GOODS-N, Abell 2744, and COSMOS. Our analyses use all publicly available JWST NIRCam imaging data in these fields, together with multi-epoch JWST MIRI images available. We measure the significance (signal-to-noise ratio or ${\rm SNR}_{\rm var}$) of the variabilities for all LRDs and statistically evaluate their variabilities using the ${\rm SNR}_{\rm var}$ distributions. We pay particular attention to the systematic offsets of photometric zero points among different epochs that seem to commonly exist. The derived ${\rm SNR}_{\rm var}$ distributions of the LRDs, including those with broad H$α$/H$β$ emission lines, follow the standard Gaussian distribution, and are generally consistent with those of the comparison samples of objects detected in the same images. This finding suggests that the LRD population on average does not show strong variability, which can be explained by super-Eddington accretion of the black holes in AGNs. Alternatively, many of them may be dominated by galaxies. We also find eight strongly variable LRD candidates with variability amplitudes of 0.24 -- 0.82 mag. The rest-frame optical SEDs of these variable LRDs should have significant AGN contribution. Future JWST observations will provide more variability information of LRDs.

astro-ph.GA↗

arXiv2025

TokenFocus-VQA: Enhancing Text-to-Image Alignment with Position-Aware Focus and Multi-Perspective Aggregations on LVLMs

While text-to-image (T2I) generation models have achieved remarkable progress in recent years, existing evaluation methodologies for vision-language alignment still struggle with the fine-grained semantic matching. Current approaches based on global similarity metrics often overlook critical token-level correspondences between textual descriptions and visual content. To this end, we present TokenFocus-VQA, a novel evaluation framework that leverages Large Vision-Language Models (LVLMs) through visual question answering (VQA) paradigm with position-specific probability optimization. Our key innovation lies in designing a token-aware loss function that selectively focuses on probability distributions at pre-defined vocabulary positions corresponding to crucial semantic elements, enabling precise measurement of fine-grained semantical alignment. The proposed framework further integrates ensemble learning techniques to aggregate multi-perspective assessments from diverse LVLMs architectures, thereby achieving further performance enhancement. Evaluated on the NTIRE 2025 T2I Quality Assessment Challenge Track 1, our TokenFocus-VQA ranks 2nd place (0.8445, only 0.0001 lower than the 1st method) on public evaluation and 2nd place (0.8426) on the official private test set, demonstrating superiority in capturing nuanced text-image correspondences compared to conventional evaluation methods.

cs.CV↗

arXiv2025

Fast Prediction of Phase shifters Phase Error Distributions Using Gaussian Processes

We present an efficient method for evaluating random phase errors in phase shifters within photonic integrated circuits, avoiding the computational cost of traditional Monte Carlo simulations. By modeling spatially correlated manufacturing variations as a stationary Gaussian process and applying linear functional theory, our approach enables rapid prediction of phase difference distributions through numerical integration. We validate the method using tolerance optimization and virtual manufacturing experiments, demonstrating its accuracy and significant reduction in computational burden. This framework offers a practical and scalable alternative for error analysis and robust design in photonic systems.

physics.optics↗

arXiv2025

Ice-Filling: Near-Optimal Channel Estimation for Dense Array Systems

By deploying a large number of antennas with sub-half-wavelength spacing in a compact space, dense array systems (DASs) can fully unleash the multiplexing and diversity gains of limited apertures. To acquire these gains, accurate channel state information acquisition is necessary but challenging due to the large antenna numbers. To overcome this obstacle, this paper reveals that designing the observation matrix to exploit the high spatial correlation of DAS channels is crucial for realizing near-optimal Bayesian channel estimation. Specifically, we prove that the observation matrix design for channel estimation is equivalent to a time-domain duality of point-to-point multiple-input multiple-output precoding, except for the change in the total power constraint on the precoding matrix to the pilot-wise discrete power constraint on the observation matrix. Inspired by Bayesian regression, a novel ice-filling algorithm is proposed to design amplitude-and-phase controllable observation matrices, and a majorization-minimization algorithm is proposed to address the phase-only controllable case. Particularly, we prove that the ice-filling algorithm can be interpreted as a ``quantized" water-filling algorithm, wherein the latter's continuous power-allocation process is converted into the former's discrete pilot-assignment process. To support the near-optimality of the proposed designs, we provide comprehensive analyses on the achievable mean square errors and their asymptotic expressions. Finally, numerical results confirm that our proposed designs achieve the near-optimal channel estimation performance and outperform existing approaches significantly.

cs.IT↗

arXiv2025

HunyuanVideo: A Systematic Framework For Large Video Generative Models

Recent advancements in video generation have significantly impacted daily life for both individuals and industries. However, the leading video generation models remain closed-source, resulting in a notable performance gap between industry capabilities and those available to the public. In this report, we introduce HunyuanVideo, an innovative open-source video foundation model that demonstrates performance in video generation comparable to, or even surpassing, that of leading closed-source models. HunyuanVideo encompasses a comprehensive framework that integrates several key elements, including data curation, advanced architectural design, progressive model scaling and training, and an efficient infrastructure tailored for large-scale model training and inference. As a result, we successfully trained a video generative model with over 13 billion parameters, making it the largest among all open-source models. We conducted extensive experiments and implemented a series of targeted designs to ensure high visual quality, motion dynamics, text-video alignment, and advanced filming techniques. According to evaluations by professionals, HunyuanVideo outperforms previous state-of-the-art models, including Runway Gen-3, Luma 1.6, and three top-performing Chinese video generative models. By releasing the code for the foundation model and its applications, we aim to bridge the gap between closed-source and open-source communities. This initiative will empower individuals within the community to experiment with their ideas, fostering a more dynamic and vibrant video generation ecosystem. The code is publicly available at https://github.com/Tencent/HunyuanVideo.

cs.CV↗

arXiv2025

Large Language Model Enhanced Recommender Systems: A Survey

Large Language Model (LLM) has transformative potential in various domains, including recommender systems (RS). There have been a handful of research that focuses on empowering the RS by LLM. However, previous efforts mainly focus on LLM as RS, which may face the challenge of intolerant inference costs by LLM. Recently, the integration of LLM into RS, known as LLM-Enhanced Recommender Systems (LLMERS), has garnered significant interest due to its potential to address latency and memory constraints in real-world applications. This paper presents a comprehensive survey of the latest research efforts aimed at leveraging LLM to enhance RS capabilities. We identify a critical shift in the field with the move towards incorporating LLM into the online system, notably by avoiding their use during inference. Our survey categorizes the existing LLMERS approaches into three primary types based on the component of the RS model being augmented: Knowledge Enhancement, Interaction Enhancement, and Model Enhancement. We provide an in-depth analysis of each category, discussing the methodologies, challenges, and contributions of recent studies. Furthermore, we highlight several promising research directions that could further advance the field of LLMERS.

cs.IR↗

arXiv2025

Personalized Recommendation Models in Federated Settings: A Survey

Federated recommender systems (FedRecSys) have emerged as a pivotal solution for privacy-aware recommendations, balancing growing demands for data security and personalized experiences. Current research efforts predominantly concentrate on adapting traditional recommendation architectures to federated environments, optimizing communication efficiency, and mitigating security vulnerabilities. However, user personalization modeling, which is essential for capturing heterogeneous preferences in this decentralized and non-IID data setting, remains underexplored. This survey addresses this gap by systematically exploring personalization in FedRecSys, charting its evolution from centralized paradigms to federated-specific innovations. We establish a foundational definition of personalization in a federated setting, emphasizing personalized models as a critical solution for capturing fine-grained user preferences. The work critically examines the technical hurdles of building personalized FedRecSys and synthesizes promising methodologies to meet these challenges. As the first consolidated study in this domain, this survey serves as both a technical reference and a catalyst for advancing personalized FedRecSys research.

cs.IR↗