Search arXiv⌕ Search

arXiv subjects

Xiang Li

Publications and source records attributed to Xiang Li.

At least 37 records · Page 2Linked to original sources
arXiv2026

SciFigBench: A Benchmark for Scientific Method Figure Understanding

Scientific method figures often provide researchers with a clear visual entry point into a paper's methodology and core contributions, yet their dense layouts and complex information flows are difficult for Vision Language Models (VLMs) to interpret. Existing benchmarks predominantly focus on coarse-grained tasks, such as context-based retrieval or holistic comprehension, overlooking the fine-grained content and inter-module relations within complex methodological diagrams. To address these overlooked aspects, we introduce Scientific Figure Benchmark (SciFigBench), a benchmark for evaluating core capabilities required for scientific method figure understanding. Specifically, it targets three capabilities: dense content perception, visual relation comprehension, and information-flow tracing. For this purpose, we curate 4,352 scientific method figures from real-world research papers and construct 40,670 QA pairs to operationalize these capabilities as measurable question-answering tasks. All annotations are generated by dedicated pipelines and then manually verified. We further propose SciFigCue, a training-free visual prompting module that externalizes arrow-induced dependencies as visual relational evidence for method-figure reasoning. Experiments across open-source and proprietary VLMs show that even strong models remain substantially limited in visual relation comprehension, while SciFigCue improves performance across several tasks without parameter updates. The code and related resources are available at https://github.com/LeNcyer/SciFigBench.

cs.CV↗
arXiv2026

Shifting Adaptation from Weight Space to Memory Space: A Memory-Augmented Agent for Medical Image Segmentation

Medical image segmentation is fundamental to clinical workflows, yet models trained on a single dataset often fail to generalize across institutions, scanners, or patient populations. While vision foundation models have shown great promise in addressing this challenge, their deployment typically requires task-specific fine-tuning, which introduces substantial communication overhead in federated learning and prevents continuous knowledge evolution during deployment. In this work, we propose a memory-augmented segmentation agent (MemSeg-Agent) that shifts adaptation from weight space to memory space, enabling few-shot learning, federated supervised learning, and test-time adaptation within a unified architecture. MemSeg-Agent conditions a fixed backbone with lightweight static, few-shot, and test-time working memories, which are dynamically composed by an agentic controller. In federated settings, we update compact memory units instead of model parameters, substantially reducing communication overhead. Experiments on four public datasets demonstrate strong performance and robustness to domain shift: Static memory alone matches or surpasses strong supervised baselines with high parameter efficiency, and test-time working memory further improves in-domain and cross-domain performance without fine-tuning. Overall, MemSeg-Agent introduces a new paradigm for scalable and adaptive medical image segmentation in the era of agentic AI.

cs.CV↗
arXiv2026

Latent-WAM: Latent World Action Modeling for End-to-End Autonomous Driving

We introduce Latent-WAM, an efficient end-to-end autonomous driving framework that achieves strong trajectory planning through spatially-aware and dynamics-informed latent world representations. Existing world-model-based planners suffer from inadequately compressed representations, limited spatial understanding, and underutilized temporal dynamics, resulting in sub-optimal planning under constrained data and compute budgets. Latent-WAM addresses these limitations with two core modules: a Spatial-Aware Compressive World Encoder (SCWE) that distills geometric knowledge from a foundation model and compresses multi-view images into compact scene tokens via learnable queries, and a Dynamic Latent World Model (DLWM) that employs a causal Transformer to autoregressively predict future world status conditioned on historical visual and ego status representations. Extensive experiments on NAVSIM v2 and HUGSIM demonstrate new state-of-the-art results: 89.3 EPDMS on NAVSIM v2 and 28.9 HD-Score on HUGSIM, surpassing the best prior perception-free method by 3.2 EPDMS with significantly less training data and a compact 104M-parameter model. The code and model weights have been open-sourced at https://github.com/yukki555/Latent-WAM.

cs.CV↗
arXiv2026

Zero2Repo: Can Coding Agents Build Repositories from Scratch?

Coding agents are increasingly asked to build software rather than patch it, yet benchmarks for from-scratch repository construction are mostly limited to a single language and depend on manually curated tasks. We introduce Zero2Repo, a benchmark in which an agent receives a product requirements document, an interface contract, and an empty workspace, and must deliver a complete repository in the project's native ecosystem. Tasks are produced by a language-agnostic authoring pipeline that converts real, version-pinned open-source projects into behavioral specifications, reproducible environments, and hidden acceptance tests. Each task is validated by execution: a reference implementation derived from the upstream project must pass, and adversarial validation must show that the tests reject incorrect implementations. Evaluation runs production coding agents in isolated containers, withholds the acceptance tests until an explicit submission, and assigns a binary reward only when every test passes, with no LLM judge. The pipeline and harness make no language-specific assumptions and apply to mainstream programming ecosystems; the current release contains Python, TypeScript, Go, and C++ tasks. Even on 11 tasks drawn from repositories that frontier models have very likely seen during training, the strongest agent solves only 10, and every failing submission passes 90-99% of the hidden tests; for the two strongest agents, 67-100% of failed tests trace to a single omission or a low-frequency rule stated in the specification rather than to a missing subsystem, so each failure is a concrete target for improvement.

cs.SE↗
arXiv2026

CoRe-WM: Conditioned Response-State World Modeling for Longitudinal Treatment-Response Prediction in Breast DCE-MRI

Pathologic complete response (pCR) after neoadjuvant chemotherapy is a key prognostic indicator in aggressive breast cancer, yet existing prediction approaches either rely on pretreatment imaging alone, missing the patient's evolving response, or optimize multi-visit classifiers under direct outcome supervision, coupling representation learning with the clinical endpoint. We formulate longitudinal pCR prediction as a world-modeling problem and propose Conditioned Response-state World Modeling (CoRe-WM), which learns a latent state-transition model that predicts the next imaging state from the observed sequence. The transition function is conditioned on patient clinical profile and assigned therapy through additive shift and feature-wise modulation, enabling patient- and treatment-specific dynamics. A Factorized Response State pathway further decouples treatment-response dynamics from imaging appearance by encoding the structural prior that response manifests primarily in lesion morphology. After pretraining, all parameters are frozen and a linear readout predicts pCR at each clinical decision point. We evaluate CoRe-WM on I-SPY2 under five-fold cross-validation and assess cross-cohort transfer on I-SPY1. CoRe-WM achieves the highest discrimination among all evaluated methods at every prospective decision point, and demonstrates robust cross-cohort transfer. Ablation studies identify patient-profile conditioning and the factorized response pathway as the principal contributors. These results establish world-model-based representation learning as a promising framework for outcome prediction in oncology and beyond.

cs.CV↗
arXiv2026

AOI: Turning Failed Trajectories into Training Signals for Autonomous Cloud Diagnosis

Large language model (LLM) agents offer a promising data-driven approach to automating Site Reliability Engineering (SRE), yet their enterprise deployment is constrained by three challenges: restricted access to proprietary data, unsafe action execution under permission-governed environments, and the inability of closed systems to improve from failures. We present AOI (Autonomous Operations Intelligence), a trainable multi-agent framework formulating automated operations as a structured trajectory learning problem under security constraints. Our approach integrates three key components. First, a trainable diagnostic system applies Group Relative Policy Optimization (GRPO) to distill expert-level knowledge into locally deployed open-source models, enabling preference-based learning without exposing sensitive data. Second, a read-write separated execution architecture decomposes operational trajectories into observation, reasoning, and action phases, allowing safe learning while preventing unauthorized state mutation. Third, a Failure Trajectory Closed-Loop Evolver mines unsuccessful trajectories and converts them into corrective supervision signals, enabling continual data augmentation. Evaluated on the AIOpsLab benchmark, our contributions yield cumulative gains. (1) The AOI runtime alone achieves 66.3% best@5 success on all 86 tasks, outperforming the prior state-of-the-art (41.9%) by 24.4 points. (2) Adding Observer GRPO training, a locally deployed 14B model reaches 42.9% avg@1 on 63 held-out tasks with unseen fault types, surpassing Claude Sonnet 4.5. (3) The Evolver converts 37 failed trajectories into diagnostic guidance, improving end-to-end avg@5 by 4.8 points while reducing variance by 35%.

cs.LG↗
arXiv2026

Vector-field-guided constraint-following control for path following of uncertain mechanical systems

Achieving dynamics-level control is a critical yet challenging problem for accurate, high-speed geometric path-following of uncertain robotic systems. This paper proposes a vector-field-guided constraint-following control framework for geometric path following of uncertain mechanical systems. The key idea is to encode vector-field guidance as a velocity constraint, enabling a unified force/torque control design for a class of fully actuated and underactuated systems with desired paths that may be self-intersecting. A global, explicit error estimate connects constraint-following accuracy to path-following performance with convergence guarantees. An adaptive robust extension further addresses heterogeneous time-varying uncertainties with unknown bounds and ensures ultimately bounded path-following error. Applications to an underactuated aircraft system and a fully actuated manipulator system demonstrate the effectiveness and advantages of the framework.

eess.SY↗
arXiv2026

AlphaLedger: Portfolio-Grounded On-Policy Distillation for Regime-Aware Alpha Mining

Hosted LLMs enable automated alpha mining but incur recurring API costs and expose research context to external providers. Training a local generator requires supervision aligned with portfolio utility: a predictive factor may add little after redundancy and trading costs. This incremental value is assessed only after the factor program is complete, whereas generation proceeds sequentially. Deployment must also adapt to evolving stock-specific evidence and changes in factor effectiveness across market regimes. We propose AlphaLedger to address these challenges. Economic-context on-policy distillation (EC-OPD) conditions a frozen teacher on completed factors and their cost-adjusted marginal portfolio contributions, then distills its legal-action distributions along the student's own trajectories. The deployed generator requires neither teacher calls nor outcome inputs. EviPool constructs a dynamic stock universe from lagged multi-source evidence, while a regime-residual gate (RRG) reweights factors using matured, state-matched returns with shrinkage and bounded deviations from equal weighting. On Chinese A-shares, AlphaLedger achieves a 55.4\% annualized net return, a Sharpe ratio of 2.06, and a maximum drawdown of 10.8\%. In a common CSI500 universe, its annualized net return exceeds the strongest baseline by 43.6\% (31.6\% vs.\ 22.0\%), with a 13.5\% higher Sharpe ratio.

cs.CE↗
arXiv2026

Optimal Watermark Localization in Mixed-Source Large Language Model Texts

Watermarking provides a principled way to authenticate text generated by large language models (LLMs). In practice, however, the final text may be mixed-source, with watermark evidence surviving at only a subset of token positions after rewriting, insertion, deletion, or paraphrasing. Although prior work has studied global detection of watermark signals, when such signals can be localized remains unclear. We formulate watermark localization as a token-level multiple-testing problem based on pivotal statistics, with a latent indicator recording whether watermark dependence survives at each position. Under an asymptotic regime indexed by exponents for signal sparsity, next-token concentration, and effective-vocabulary growth, we derive a sharp boundary for global detection and phase transitions for discovery and classification within the class of coordinatewise pivot-based localization rules. We show that discovery is strictly harder than detection and that consistent classification is impossible across the parameter regime within this class. We then develop an adaptive thresholding method that does not require knowledge of the exponents or time-varying next-token distributions, but uses a data-driven estimate of the surviving watermark fraction. The method attains the optimal discovery boundary and near-optimal discovery power relative to homogeneous pivot-based rules. Simulations support the theoretical phase transitions, while experiments on model-generated texts demonstrate practical localization performance under common edit mechanisms.

stat.ME↗
arXiv2026

WinARDet: Compressed Window Auto-Regression for Long-Horizon Vision-Language Grounding

MLLM-based detectors formulate vision-language grounding as autoregressive coordinate generation, yet in dense scenes, they often collapse into a box repetition loop, repeatedly emitting identical boxes until the decoding budget is exhausted. We trace this failure to historical-answer attention takeover: Under full-history autoregression, repeated coordinate tokens in the growing answer history attract disproportionate attention, crowding out the attention sinks needed to attend to new visual evidence. % To break this loop, in this paper, we propose \textbf{Compressed Window Auto-Regression} (WinAR). This new mechanism partitions a long answer sequence into consecutive generation windows and restricts full token-level autoregression to the current window only while compressing each previous window into a fixed set of latent tokens that condition subsequent decoding. This design reduces the repeated exposure to raw historical coordinates that triggers the attention takeover, yet preserves cross-window dependencies essential for coherent multi-object generation. On RSDense, which contains 200-2,000 objects per image, our model WinARDet improves RexOmni F1-score (IoU=0.5) from 32.16 to 39.36 over the baseline, while reducing the redundancy rate from 35.48% to 13.00%. WinARDet also generalizes across decoding strategies and output formats, including parallel box decoding and polygon generation, providing a practical paradigm for long-horizon vision-language grounding across scene densities.

cs.CV↗
arXiv2026

TADDLE: A Tool-Augmented Agent for Detecting Deficient LLM-Generated Peer Reviews

LLM-generated peer reviews are increasingly common at major venues, yet their deficiencies are hard to detect because they are uniformly fluent and well-structured. Existing work either classifies authorship without judging quality, or scores quality with features designed for human-written reviews; no prior system detects deficiencies in LLM-generated reviews at the level of individual defect types. To bridge the gap, we introduce TADDLE, a Tool-Augmented Agent for Detecting Deficient LLM-Generated Peer Reviews, together with the first expert-annotated benchmark for this task. Our benchmark comprises 1,800 reviews on 50 ICLR 2025 papers, multi-label-annotated by 18 domain experts against a taxonomy of six defect categories (plus a non-deficient label). TADDLE decomposes detection into four specialized analysis tools -- Verify, Correct, Complete, and Transform -- orchestrated by an agent; an integrator synthesizes their outputs into binary and multi-label classifications via two-stage semi-supervised learning. Experiments show that TADDLE performs strongly on binary detection and the multi-label classification task. We release the benchmark and code at https://github.com/AquariusAQ/TADDLE.

cs.AI↗
arXiv2026

EmbodiSteer: Steering Embodiment-Agnostic Visuomotor Policies with Joint-Space Guidance for Zero-Shot Cross-Embodiment Deployment

Scalable robot imitation learning relies on large-scale heterogeneous data from diverse robots or body-free data, making Cartesian end-effector actions a key interface for embodiment-agnostic policy learning. However, end-effector-only abstraction leaves Cartesian policies unaware of the deployed robot body, making them brittle under robot-specific constraints such as whole-body collision avoidance. To overcome this limitation, we present EmbodiSteer, a training-free framework that steers embodiment-agnostic visuomotor policies toward zero-shot, embodiment-aware deployment. EmbodiSteer keeps policy learning in Cartesian space while efficiently lifting inference-time diffusion sampling into the target robot's joint space via forward kinematics and Jacobian-based updates. With whole-body collision-aware guidance over joint trajectories after each denoising step, the arm can be steered away from collisions while preserving learned end-effector behavior. Across 9 simulated robots, EmbodiSteer reduces collision rate by 46.1 percentage points and increases task success by 28.5 percentage points over Cartesian-only execution; on two physical robots in highly constrained scenarios, the respective improvements are 90.0 and 36.7 percentage points. Our project page is at https://frankwang67.github.io/EmbodiSteer-Page.

cs.RO↗
arXiv2026

Same Attack, Different Judge: A First Controlled Adversarial Comparison of Jev, LLM Judges, and Purpose-Built Safety Classifiers

TypeSafe AI's Jev is a new kind of safety judge. It generates no text and no label: it answers questions the developer defines in advance and returns a probability for each answer. The vendor admits that adversarial input can change these answers but gives no numbers. In this paper, we present the first controlled adversarial evaluation of Jev. We build a test harness that sends every system the same payload under the same policy, and run it on four public-data tasks, from rule-based ones (input moderation, injection detection, routing) to a judgment-based one (phishing detection), with fifty attack variants from thirteen families. The targets are Jev, five GPT and GLM judges, and two local classifiers (Llama Guard 3, Prompt Guard 2). We find that Jev is the most robust system on input moderation: eight forgeries of its structured input bypass none of its decisions, and a sentence asking only why someone might make a harmful request bypasses 9% of its blocks, against 56% for GPT-6 Astra. On phishing detection. However, the same forgeries bypass 19-29% of the emails it flags, and even without attack Jev detects only 62% of phishing emails, against 77-81% for the GLM judges. Jev is also the fastest judge, but it flags the most benign requests (27%). The local classifiers fail differently: a training-example statement makes Llama Guard 3 miss 62% of the phishing it had flagged. These results suggest that Jev resists attacks on tasks with a clear rule to check but is more exposed on tasks that call for an overall judgment.

cs.CR↗
arXiv2026

ProtoArena: Can AI Agents Discover and Exploit Real-World Cross-Component Protocol Vulnerabilities?

Large language model (LLM) agents are increasingly used for vulnerability analysis, yet existing benchmarks provide limited coverage of exploitation that depends on interactions among network protocol components. Such vulnerabilities arise from inconsistencies in message interpretation, timing, state, or trust assumptions across participants. We introduce ProtoArena, a black-box, multi-component, dynamic benchmark with 80 cross-component vulnerability-exploitation environments spanning 11 protocol and application domains and 7 attack categories. Tasks are primarily informed by published protocol-attack mechanisms. Agents interact with running systems through scoped interfaces to achieve security-effect objectives without target source code. Three information levels assess guided reproduction, exploitation with mechanism and topology context, and effect-directed black-box exploration. A hybrid verifier combines rule-based checks with LLM review of independently collected traffic and logs. We evaluate five LLMs. Claude Opus 4.8 achieves the highest observed success rates: 48.8% under L1 guidance and 12.5% under effect-directed black-box L3. In a separate DNSBomb optimization experiment, it achieves a peak downstream-response to average query-input byte-rate ratio exceeding 171,000 times. The results show that agents can achieve verifiable exploitation effects in some cross-component environments, but success with limited black-box information remains low and varies across domains and attack categories. All experiments use controlled, explicitly authorized targets; they do not test production systems or disclose previously unknown vulnerabilities.

cs.CR↗
arXiv2026

Task-evoked eye movements as candidate digital biomarkers of depression severity in adolescents and young adults

Objective digital measures of depression severity could complement conventional symptom assessments and facilitate more timely clinical evaluation. Here, we investigated whether task-evoked eye movements provide interpretable behavioral markers of depression severity in adolescents and young adults. In this single-center, cross-sectional study, 155 participants aged 14--20 years completed emotional-spatial free-viewing and memory-guided recall tasks and were stratified by 17-item Hamilton Depression Rating Scale scores. We examined 19 prespecified eye-movement features across six task-informed domains using covariate-adjusted ordinal-trend models. Greater depression severity was associated with lower gaze-allocation entropy (standardized $β=-0.599$), fewer inter-area switches ($β=-0.526$), longer first target-entry latency ($β=0.572$), and lower target dwell proportion ($β=-0.678$; all false-discovery-rate-adjusted $q<0.001$). Emotion-specific attentional-bias measures were not associated with severity, whereas associations involving fixation dispersion weakened after adjustment for anxiety symptoms. In a complementary machine-learning analysis incorporating 44 candidate eye-movement features and demographic covariates, a three-stage cascade classifier achieved the highest four-class stratification performance (accuracy, 0.665; macro-F1, 0.670; area under the receiver operating characteristic curve, 0.869), outperforming single-stage and two-stage approaches. Together, these findings identify altered visual exploration and memory-guided gaze control as candidate digital biomarkers of depression severity, rather than generalized negative attentional bias. These results support further prospective evaluation of task-evoked eye tracking as a complementary approach for timely severity assessment and clinical decision support.

eess.SP↗
arXiv2026

Contactless gait-derived features as candidate digital biomarkers of depression severity in adolescents

Timely assessment of depression severity in adolescents is essential for guiding clinical care, yet conventional evaluations rely on episodic clinical assessments and symptom rating scales. Contactless gait analysis offers a potentially scalable approach to capturing objective behavioral features associated with depression. Here, we introduce the Long-Distance 3D Adolescent Depression Gait Dataset (LD3D-ADG), comprising depth-sensor gait recordings from 119 adolescents at a single center, including 39 healthy controls, 41 with mild depression, and 39 with moderate-to-severe depression. We developed UMLPA-GCN, a graph convolutional network that models spatiotemporal gait dynamics across three anatomically defined skeletal partitions. In participant-independent five-fold cross-validation, UMLPA-GCN achieved 75.63% classification accuracy, a macro-F1 score of 75.52%, and a macro-area under the receiver operating characteristic curve of 0.839 (95% confidence interval, 0.779--0.899). Class-specific recall was 87.18%, 63.41%, and 76.92% for healthy controls, mild depression, and moderate-to-severe depression, respectively. At the participant level, reduced arm swing, slower pelvic movement, restricted joint range of motion, and greater arm-swing asymmetry were identified as leading potential digital biomarkers contributing to depression severity assessment, associated with higher 17-item Hamilton Depression Rating Scale scores. These findings provide proof-of-concept evidence for contactless digital depression severity stratification in adolescents. Further external and prospective validation is needed to establish their utility for longitudinal monitoring and timely clinical assessment.

eess.SP↗
arXiv2026

Negative Advantages Is a Double-Edged Sword: Calibrating advantages in GRPO for Search Agents

Search agents achieve strong question-answering performance through multi-turn interactions with search engines, with Group Relative Policy Optimization (GRPO) being a widely used training algorithm. However, GRPO-style algorithms still face several challenges in multi-hop search settings. First, correct intermediate steps are often penalized when the final answer is wrong. Second, training is highly unstable, often causing degradation of natural language ability or even catastrophic training collapse. Our analysis attributes these issues to coarse-grained advantage assignment and an imbalance between positive and negative advantages. To address these problems, we propose CalibAdv, an advantage calibration method specifically designed for search agents that enables more accurate and more stable modeling of penalties and rewards. Specifically, CalibAdv leverages the correctness of intermediate steps to downscale excessive negative advantages at a fine-grained level. It then further rebalances positive and negative advantages to improve training stability. Importantly, CalibAdv adopts a lightweight design that calibrates advantages from standard rollout signals, making it simple and easy to deploy. Extensive experiments across three models and seven benchmarks demonstrate that CalibAdv improves both model performance and training stability. Our code is available at https://github.com/wujwyi/CalibAdv.

cs.CL↗
arXiv2026

Breaking the Cloak! Unveiling Chinese Cloaked Toxicity with Homophone Graph and Toxic Lexicon

Social media platforms have experienced a significant rise in toxic content, including abusive language and discriminatory remarks, presenting growing challenges for content moderation. Some users evade censorship by deliberately disguising toxic words through homophonic cloak, which necessitates the task of unveiling cloaked toxicity. Existing methods are mostly designed for English texts, while Chinese cloaked toxicity unveiling has not been solved yet. To tackle the issue, we propose C$^2$TU, a novel training-free and prompt-free method for Chinese cloaked toxic content unveiling. It first employs substring matching to identify candidate toxic words based on Chinese homophone graph and toxic lexicon. Then it filters those candidates that are non-toxic and corrects cloaks to be their corresponding toxicities. Specifically, we develop two model variants for filtering, which are based on BERT and LLMs, respectively. For LLMs, we address the auto-regressive limitation in computing word occurrence probability and utilize the full semantic contexts of a text sequence to reveal cloaked toxic words. Extensive experiments demonstrate that C$^2$TU can achieve superior performance on two Chinese toxic datasets. In particular, our method outperforms the best competitor by up to 71% on the F1 score and 35% on accuracy, respectively. Our code and data are available at https://github.com/XDxc-cuber/C2TU-Chinese-cloaked-toxicity-unveiling.

cs.CL↗