arXiv subjects

Chuang Liu

Publications and source records attributed to Chuang Liu.

At least 19 recordsLinked to original sources

arXiv2026

MemGuard: Persisting Verifier Signals for LLM-Agent Memory Governance

LLM agents are moving from single-prompt use to long task streams in which reusable memory becomes a core capability for terminal, software-engineering, and web tasks. Such memory is useful only when stored experience remains reliable across hundreds of interactions, but two failure modes break that assumption in practice. The first is unreliable admission: failed trajectories,accidental successes, and misleading observations enter memory because they appear relevant, then mislead later decisions. The second is memory drift: long-running banks accumulate duplicate, stale, and conflicting records that retrieval alone cannot repair. MemGuard's key distinction is to treat verifier output not as a one-shot filter, but as persistent lifecycle metadata. It converts multi-criteria score-token verification into reward, confidence, label, and uncertainty descriptors that are attached to every candidate before activation and reused during retrieval, conflict resolution, summarization, and archival. We evaluate MemGuard on Terminal-Bench 2.0, SWE-Bench Verified, WebArena, and Mind2Web across four backbones, comparing against four memory baselines plus a verifier-only control under matched runtime budgets. Averaged over five seeds, MemGuard achieves the best success metric and lowest average steps in all 16 backbone-benchmark settings, improving over ReasoningBank, the strongest prior baseline among the memory methods we evaluate, with a largest gain of 7.9 success-rate points on WebArena, 5.6 step-success-rate points on Mind2Web, and 2.4-3.5 points on terminal and software-engineering benchmarks. Code is available at https://github.com/whyyyyy123/MemGuard.

cs.AI↗

arXiv2026

When Images Look Right and Retrieve Wrong: Coverage-Guided Cross-Scale Re-Indexing for Knowledge-Faithful Generative Perception

Multimodal information systems increasingly route generated visual content back through the same vision-language index that informed its production, so the output must remain retrievable by the queries it was meant to serve. When the scene contains entities at vastly different scales, existing language-guided generators condition on a single, globally pooled text embedding and quietly drop scale-specific concepts, breaking concept-query retrieval even when pixel fidelity is high. We formalise this failure as semantic collapse and propose CERES, a closed-loop multimodal indexing framework that builds a three-level semantic pyramid, mines implicit concepts via a co-occurrence-aware router, performs scale-routed cross-attention into a lightweight U-Net generator, and verifies coverage by re-indexing the generated image with the same frozen VLM. A continuously differentiable soft-Jaccard coverage objective returns dense gradients to the 0.39 M-parameter generator under explicit non-degeneracy conditions, and coverage is verified by an independent DINOv2 linear probe trained only on external scene and object labels. On four pansharpening benchmarks across seven settings, CERES delivers the new state of the art with the largest gains where scale variation is most extreme (+4.64% relative Q2n and +9.7 mAP for DOTA detection). It also improves concept-query retrieval Recall@5 by +14.0 points and image-text mean reciprocal rank by 0.19 over the strongest baseline, showing that the closed loop preserves queryable content rather than self-referential feature consistency.

cs.MM↗

arXiv2026

GenMatch: An End-to-End Generative Matching Framework for Micro-View Order-Dispatching in Ride-Hailing

Micro-View Order-Dispatching assigns available drivers to passenger orders within each dispatch batch and is critical to the service quality and operational efficiency of ride-hailing platforms. Mainstream industrial solutions follow a multi-stage paradigm of model prediction, value calculation, and dispatch matching. Although dispatch quality is determined by the final batch-level assignment, these stages optimize different intermediate objectives. This cross-stage objective inconsistency means that improving a single stage does not necessarily improve the overall dispatch result. We therefore formulate Micro-View Order-Dispatching as a generative matching problem and propose GenMatch, an end-to-end Generative Matching framework and the first such framework deployed in a real-world production environment. Applying generative modeling to this problem introduces three challenges. First, each dispatch batch forms a dynamic sparse bipartite graph, requiring efficient structured batch-level encoding. Second, replacing the hand-crafted value function requires learning unified business utility from heterogeneous feedback. Third, directly generating an assignment requires tracking the evolving matching state because each selected order-driver pair changes the remaining feasible candidates. GenMatch addresses these challenges with a Context-Aware Bipartite Encoder, a Business-Aware Utility Learner, and a State-Aware Pointer Decoder. Extensive offline evaluations and online A/B tests in five cities across DiDi's international ride-hailing markets show consistent improvements over competitive baselines, confirming the effectiveness and practicality of GenMatch for industrial order-dispatching.

cs.AI↗

arXiv2026

EMMR: Emotion-Mediated Multimodal Reasoning for Personality Assessment in Asynchronous Video Interviews

Asynchronous Video Interviews (AVIs) have become increasingly popular for personality assessment. Recent large language models (LLMs) have shown potential for personality assessment from transcribed interview responses. However, text-centered methods may overlook non-verbal behavioral cues conveyed through visual and audio modalities, even though such cues are highly relevant to personality assessment. In particular, emotion-related cues provide important social and affective evidence for understanding candidates' behavior related to personality traits. Thus, we propose EMMR (Emotion-Mediated Multimodal Reasoning), a two-stage framework for MLLMs-based personality assessment for AVIs. EMMR extracts emotion-related cues from multimodal interview data and incorporates them into personality assessment through structured reasoning as auxiliary social and behavioral evidence. Experiments on two AVIs datasets, OPVA and AVI-6, show that EMMR improves MAE, MSE, and PCC compared with baselines. Further analysis indicates that semantic descriptions of emotion cues enhance personality assessment, while their quality affects personality assessment reliability. These results suggest that integrating emotion-related cues into multimodal reasoning is a promising direction for more interpretable MLLMs-based personality assessment in AVIs.

cs.HC↗

arXiv2026

GraphSculptor: Sculpting Pre-training Coreset for Graph Self-supervised Learning

Graph self-supervised learning typically relies on large-scale unlabeled datasets, heavily inflating computational costs. However, empirical evidence suggests that these datasets contain substantial redundancy-our analysis reveals that uniformly subsampling 50% of graphs retains over 96% of downstream performance. To exploit this redundancy, we introduce GraphSculptor for pre-training coreset construction. Unlike methods dependent on additional training-time signals or limited solely to topological statistics, GraphSculptor provides a label-free solution that constructs coresets via two complementary perspectives: intrinsic structure and contextual semantics. Concretely, structural diversity is quantified using intrinsic graph statistics, yielding a structural feature vector for each graph, while semantic diversity is captured by utilizing a pre-trained language model to encode descriptions generated via graph-to-text. GraphSculptor integrates these signals into a unified metric space and performs cluster-aware selection to preserve joint structural-semantic diversity. We further derive a theoretical bound on the loss gap between coreset and full-data pre-training, offering theoretical motivation for our selection formulation. Extensive experiments demonstrate that GraphSculptor effectively sculpts the dataset: a 10% coreset achieves 99.6% of full-data performance while reducing pre-training time by nearly 90%, offering a scalable solution for data-efficient graph pre-training.

cs.LG↗

arXiv2026

SwiftGS: Episodic Priors for Immediate Satellite Surface Recovery

Rapid, large-scale 3D reconstruction from multi-date satellite imagery is vital for environmental monitoring, urban planning, and disaster response, yet remains difficult due to illumination changes, sensor heterogeneity, and the cost of per-scene optimization. We introduce SwiftGS, a meta-learned system that reconstructs 3D surfaces in a single forward pass by predicting geometry-radiation-decoupled Gaussian primitives together with a lightweight SDF, replacing expensive per-scene fitting with episodic training that captures transferable priors. The model couples a differentiable physics graph for projection, illumination, and sensor response with spatial gating that blends sparse Gaussian detail and global SDF structure, and incorporates semantic-geometric fusion, conditional lightweight task heads, and multi-view supervision from a frozen geometric teacher under an uncertainty-aware multi-task loss. At inference, SwiftGS operates zero-shot with optional compact calibration and achieves accurate DSM reconstruction and view-consistent rendering at significantly reduced computational cost, with ablations highlighting the benefits of the hybrid representation, physics-aware rendering, and episodic meta-training.

cs.CV↗

arXiv2026

Directing the Narrative: A Finetuning Method for Controlling Coherence and Style in Story Generation

Story visualization requires generating sequential imagery that aligns semantically with evolving narratives while maintaining rigorous consistency in character identity and visual style. However, existing methodologies often struggle with subject inconsistency and identity drift, particularly when depicting complex interactions or extended narrative arcs. To address these challenges, we propose a cohesive two-stage framework designed for robust and consistent story generation. First, we introduce Group-Shared Attention (GSA), a mechanism that fosters intrinsic consistency by enabling lossless cross-sample information flow within attention layers. This allows the model to structurally encode identity correspondence across frames without relying on external encoders. Second, we leverage Direct Preference Optimization (DPO) to align generated outputs with human aesthetic and narrative standards. Unlike conventional methods that rely on conflicting auxiliary losses, our approach simultaneously enhances visual fidelity and identity preservation by learning from holistic preference data. Extensive evaluations on the ViStoryBench benchmark demonstrate that our method establishes a new state-of-the-art, significantly outperforming strong baselines with gains of +10.0 in Character Identity (CIDS) and +18.7 in Style Consistency (CSD), all while preserving high-fidelity generation.

cs.CV↗

arXiv2026

DEP: A Decentralized Large Language Model Evaluation Protocol

With the rapid development of Large Language Models (LLMs), a large number of benchmarks have been proposed. However, most benchmarks lack unified evaluation standard and require the manual implementation of custom scripts, making results hard to ensure consistency and reproducibility. Furthermore, mainstream evaluation frameworks are centralized, with datasets and answers, which increases the risk of benchmark leakage. To address these issues, we propose a Decentralized Evaluation Protocol (DEP), a decentralized yet unified and standardized evaluation framework through a matching server without constraining benchmarks. The server can be mounted locally or deployed remotely, and once adapted, it can be reused over the long term. By decoupling users, LLMs, and benchmarks, DEP enables modular, plug-and-play evaluation: benchmark files and evaluation logic stay exclusively on the server side. In remote setting, users cannot access the ground truth, thereby achieving data isolation and leak-proof evaluation. To facilitate practical adoption, we develop DEP Toolkit, a protocol-compatible toolkit that supports features such as breakpoint resume, concurrent requests, and congestion control. We also provide detailed documentation for adapting new benchmarks to DEP. Using DEP toolkit, we evaluate multiple LLMs across benchmarks. Experimental results verify the effectiveness of DEP and show that it reduces the cost of deploying benchmark evaluations. As of February 2026, we have adapted over 60 benchmarks and continue to promote community co-construction to support unified evaluation across various tasks and domains.

cs.CL↗

arXiv2026

From Subtle to Significant: Prompt-Driven Self-Improving Optimization in Test-Time Graph OOD Detection

Graph Out-of-Distribution (OOD) detection aims to identify whether a test graph deviates from the distribution of graphs observed during training, which is critical for ensuring the reliability of Graph Neural Networks (GNNs) when deployed in open-world scenarios. Recent advances in graph OOD detection have focused on test-time training techniques that facilitate OOD detection without accessing potential supervisory information (e.g., training data). However, most of these methods employ a one-pass inference paradigm, which prevents them from progressively correcting erroneous predictions to amplify OOD signals. To this end, we propose a \textbf{S}elf-\textbf{I}mproving \textbf{G}raph \textbf{O}ut-\textbf{o}f-\textbf{D}istribution detector (SIGOOD), which is an unsupervised framework that integrates continuous self-learning with test-time training for effective graph OOD detection. Specifically, SIGOOD generates a prompt to construct a prompt-enhanced graph that amplifies potential OOD signals. To optimize prompts, SIGOOD introduces an Energy Preference Optimization (EPO) loss, which leverages energy variations between the original test graph and the prompt-enhanced graph. By iteratively optimizing the prompt by involving it into the detection model in a self-improving loop, the resulting optimal prompt-enhanced graph is ultimately used for OOD detection. Comprehensive evaluations on 21 real-world datasets confirm the effectiveness and outperformance of our SIGOOD method. The code is at https://github.com/Ee1s/SIGOOD.

cs.LG↗

arXiv2025

Entropy-Driven Sensor Deployment and Source Detection in Hypergraphs

Identifying the diffusion source in complex networks is critical for understanding and controlling epidemic spread. In realistic settings, full observation of node states is rarely available, making sensor-based source detection a practical alternative. However, existing sensor-based methods are often confined to simple networks, failing to capture the higher-order group dynamics of real-world spreading process. By deploying a limited number of sensors to monitor the diffusion process, one can infer the origin from partial observations. Yet, determining optimal sensor placement is challenging, i.e., poor deployment leads to redundant or noisy data, while optimal placement must balance coverage diversity and information value under limited resources. To address these challenges, we propose a dedicated framework termed Sensor-based Source Detection in Hypergraphs (SSDH). Specifically, we introduce a novel entropy-driven sensor deployment strategy that effectively captures critical early-stage diffusion signals by maximizing information gain under limited resources. Furthermore, we develop a source localization algorithm that quantifies propagation uncertainty through a newly defined path uncertainty-based score. By integrating this score with topological distance, SSDH enables accurate and robust source identification. Extensive experiments on both synthetic and empirical hypergraphs demonstrate that SSDH consistently outperforms competing algorithms by 5%--30% across different sensor ratios, final spreading ratios, and infection probabilities. These results validate the effectiveness of SSDH and highlight its superior capability to tackle source localization in complex systems characterized by higher-order interactions.

physics.soc-ph↗

arXiv2025

Synergistic Enhancement of Requirement-to-Code Traceability: A Framework Combining Large Language Model based Data Augmentation and an Advanced Encoder

Automated requirement-to-code traceability link recovery, essential for industrial system quality and safety, is critically hindered by the scarcity of labeled data. To address this bottleneck, this paper proposes and validates a synergistic framework that integrates large language model (LLM)-driven data augmentation with an advanced encoder. We first demonstrate that data augmentation, optimized through a systematic evaluation of bi-directional and zero/few-shot prompting strategies, is highly effective, while the choice among leading LLMs is not a significant performance factor. Building on the augmented data, we further enhance an established, state-of-the-art pre-trained language model based method by incorporating an encoder distinguished by a broader pre-training corpus and an extended context window. Our experiments on four public datasets quantify the distinct contributions of our framework's components: on its own, data augmentation consistently improves the baseline method, providing substantial performance gains of up to 26.66%; incorporating the advanced encoder provides an additional lift of 2.21% to 11.25%. This synergy culminates in a fully optimized framework with maximum gains of up to 28.59% on $F_1$ score and 28.9% on $F_2$ score over the established baseline, decisively outperforming ten established baselines from three dominant paradigms. This work contributes a pragmatic and scalable methodology to overcome the data scarcity bottleneck, paving the way for broader industrial adoption of data-driven requirement-to-code traceability.

cs.SE↗

arXiv2025

OR-Toolformer: Modeling and Solving Operations Research Problems with Tool Augmented Large Language Models

Large language models (LLMs) demonstrate strong mathematical reasoning, but reliance on closed-source APIs for OR tasks raises privacy concerns, and training open-source models from scratch incurs high compute costs. We introduce OR-Toolformer, which fine-tunes Llama-3.1-8B-Instruct with a semi-automatic data synthesis pipeline that generates diverse OR problem-answer pairs and augments the model with external solvers to produce API calls. On three of four standard benchmarks, OR-Toolformer achieves up to 80.1% execution accuracy, exceeding size-matched baselines by over 4.3%. In zero-shot evaluation on two unseen OR problem types, it attains 54% average accuracy, a 21 percentage-point improvement over the strongest baseline. These findings validate the efficacy of tool-augmented fine-tuning LLMs for accurate and generalizable OR problem modeling and solving.

cs.AI↗

arXiv2025

Joint-octamamba:an octa joint segmentation network based on feature enhanced mamba

OCTA is a crucial non-invasive imaging technique for diagnosing and monitoring retinal diseases like diabetic retinopathy, age-related macular degeneration, and glaucoma. Current 2D-based methods for retinal vessel (RV) segmentation offer insufficient accuracy. To address this, we propose RVMamba, a novel architecture integrating multiple feature extraction modules with the Mamba state-space model. Moreover, existing joint segmentation models for OCTA data exhibit performance imbalance between different tasks. To simultaneously improve the segmentation of the foveal avascular zone (FAZ) and mitigate this imbalance, we introduce FAZMamba and a unified Joint-OCTAMamba framework. Experimental results on the OCTA-500 dataset demonstrate that Joint-OCTAMamba outperforms existing models across evaluation metrics.The code is available at https://github.com/lc-sfis/Joint-OCTAMamba.

cs.CV↗

arXiv2025

HIP: Model-Agnostic Hypergraph Influence Prediction via Distance-Centrality Fusion and Neural ODEs

Predicting user influence in social networks is a critical problem, and hypergraphs, as a prevalent higher-order modeling approach, provide new perspectives for this task. However, the absence of explicit cascade or infection probability data makes it particularly challenging to infer influence in hypergraphs. To address this, we introduce HIP, a unified and model-independent framework for influence prediction without knowing the underlying spreading model. HIP fuses multi-dimensional centrality indicators with a temporally reinterpreted distance matrix to effectively represent node-level diffusion capacity in the absence of observable spreading. These representations are further processed through a multi-hop Hypergraph Neural Network (HNN) to capture complex higher-order structural dependencies, while temporal correlations are modeled using a hybrid module that combines Long Short-Term Memory (LSTM) networks and Neural Ordinary Differential Equations (Neural ODEs). Notably, HIP is inherently modular: substituting the standard HGNN with the advanced DPHGNN, and the LSTM with xLSTM, yields similarly strong performance, showcasing its architectural generality and robustness. Empirical evaluations across 14 real-world hypergraph datasets demonstrate that HIP consistently surpasses existing baselines in prediction accuracy, resilience, and identification of top influencers, all without relying on any diffusion trajectories or prior knowledge of the spreading model. These findings underline HIP's effectiveness and adaptability as a general-purpose solution for influence prediction in complex hypergraph environments.

cs.SI↗

arXiv2025

Structural-Aware Key Node Identification in Hypergraphs via Representation Learning and Fine-Tuning

Evaluating node importance is a critical aspect of analyzing complex systems, with broad applications in digital marketing, rumor suppression, and disease control. However, existing methods typically rely on conventional network structures and fail to capture the polyadic interactions intrinsic to many real-world systems. To address this limitation, we study key node identification in hypergraphs, where higher-order interactions are naturally modeled as hyperedges. We propose a novel framework, AHGA, which integrates an Autoencoder for extracting higher-order structural features, a HyperGraph neural network-based pre-training module (HGNN), and an Active learning-based fine-tuning process. This fine-tuning step plays a vital role in mitigating the gap between synthetic and real-world data, thereby enhancing the model's robustness and generalization across diverse hypergraph topologies. Extensive experiments on eight empirical hypergraphs show that AHGA outperforms classical centrality-based baselines by approximately 37.4%. Furthermore, the nodes identified by AHGA exhibit both high influence and strong structural disruption capability, demonstrating their superiority in detecting multifunctional nodes.

cs.SI↗

arXiv2025

Source Detection in Hypergraph Epidemic Dynamics using a Higher-Order Dynamic Message Passing Algorithm

Source detection is crucial for capturing the dynamics of real-world infectious diseases and informing effective containment strategies. Most existing approaches to source detection focus on conventional pairwise networks, whereas recent efforts on both mathematical modeling and analysis of contact data suggest that higher-order (e.g., group) interactions among individuals may both account for a large fraction of infection events and change our understanding of how epidemic spreading proceeds in empirical populations. In the present study, we propose a message-passing algorithm, called the HDMPN, for source detection for a stochastic susceptible-infectious dynamics on hypergraphs. By modulating the likelihood maximization method by the fraction of infectious neighbors, HDMPN aims to capture the influence of higher-order structures and do better than the conventional likelihood maximization. We numerically show that, in most cases, HDMPN outperforms benchmarks including the likelihood maximization method without modification.

physics.soc-ph↗

arXiv2025

Transforming Acidic Corrosion and Embrittlement into a Hydrogen-Trapping Cage

The vision of a hydrogen economy demands efficient platforms to close the gap between sustainable proton sources and solid-state hydrogen carriers. Metal hydrides serve as key carriers, yet their synthesis remains constrained by the energy-intensive use of high-pressure H2, which fragments the hydrogen chain. Here, we overturn this paradigm by transforming two classic degradation mechanisms, acidic corrosion and hydrogen embrittlement, into a constructive materials-design strategy. We demonstrate that synergistic control of these processes in acid enables the in-situ engineering of a "hydrogen-trapping cage" (HTC) microstructure within metals. Composed of a dense defect network, this cage directly captures and stabilizes protons as hydrides under mild conditions, guided by the universal criterion |DeltaPeq| > DeltaPph. Using this platform, we synthesize over 20 hydrides, including challenging targets such as LiH and NaH, and showcase its functional power with a cage-rich titanium hydride electrocatalyst. This catalyst achieves an exceptional current density of 1.07 A cm-2 for nitrate-to-ammonia conversion, attributed to rapid H- transport within the engineered cage. This work establishes a transformative "failure-to-function" paradigm, delivering an integrated platform that unifies hydrogen capture, stabilization, and conversion.

cond-mat.mtrl-sci↗

arXiv2025

BPO: Revisiting Preference Modeling in Direct Preference Optimization

Direct Preference Optimization (DPO) have emerged as a popular method for aligning Large Language Models (LLMs) with human preferences. While DPO effectively preserves the relative ordering between chosen and rejected responses through pairwise ranking losses, it often neglects absolute reward magnitudes. This oversight can decrease the likelihood of chosen responses and increase the risk of generating out-of-distribution responses, leading to poor performance. We term this issue Degraded Chosen Responses (DCR).To address this issue, we propose Balanced Preference Optimization (BPO), a novel framework that dynamically balances the optimization of chosen and rejected responses through two key components: balanced reward margin and gap adaptor. Unlike previous methods, BPO can fundamentally resolve DPO's DCR issue, without introducing additional constraints to the loss function. Experimental results on multiple mathematical reasoning tasks show that BPO significantly outperforms DPO, improving accuracy by +10.1% with Llama-3.1-8B-Instruct (18.8% to 28.9%) and +11.7% with Qwen2.5-Math-7B (35.0% to 46.7%). It also surpasses DPO variants by +3.6% over IPO (43.1%), +5.0% over SLiC (41.7%), and +3.1% over Cal-DPO (43.6%) on the same model. Remarkably, our algorithm requires only a single line of code modification, making it simple to implement and fully compatible with existing DPO-based frameworks.

cs.CL↗