Searcharxiv⌕ Search

arXiv subjects

Shuang Zhou

Publications and source records attributed to Shuang Zhou.

At least 19 recordsLinked to original sources
arXiv2026

UniMod: Enhancing Multi-Modal Medical Diagnosis through Cross-Modality and Within-Modality Alignment

Multi-modal learning combining medical images and clinical text is promising for disease diagnosis. However, standard multi-modal training leads to shortcut learning: models exploit the easier modality (e.g., diagnostic cues in text) while neglecting harder-to-learn features (e.g., subtle visual patterns). We propose UniMod, a framework that mitigates shortcut learning by requiring each modality to predict the diagnosis on its own. It supervises image-only, text-only, and multi-modal classification simultaneously, so each modality must extract diagnostic features. We add cross-modality alignment for knowledge transfer and within-modality supervised contrastive alignment over same-diagnosis patients. On Harvard-Glaucoma, UniMod reaches 0.850 AUC, outperforming OGM-GE and Gradient Blending by 1.6-1.8%; on CheXpert Plus, it reaches 0.966 AUC, surpassing them by over 5%. UniMod also extends to 5-class multi-label diagnosis without architectural change, improving mean AUC by 0.097 over CGGM.

cs.CV↗
arXiv2026

Mapping Dust Attenuation at Kiloparsec Scales. IV. A Dust-model Interpretation of Attenuation Curves in Nearby Galaxies

In this fourth paper on kiloparsec-scale dust attenuation, we ask whether the empirical trends found in Papers I--III can be translated into effective dust properties. Using attenuation curves for 2487 high-continuum-S/N spaxels in 91 SwiM v4.2 galaxies, we construct a grid of uniform-screen dust models composed of astronomical silicate and graphite grains with MRN-like size distributions. We fit the normalized attenuation-curve shape to constrain model parameters and then use the attenuation amplitude to estimate the model-dependent dust mass surface density. The inferred dust masses, compositions, and small-grain fractions are therefore effective quantities defined within the adopted attenuation model. The fitted models reproduce the main attenuation-curve variations and provide a direct bridge to Papers I--III: within this model, the relative 2175\AA\ bump sequence maps mainly onto the effective fraction of small graphitic/carbonaceous grains, while the NUV-slope sequence maps onto the effective small-silicate grain fraction and total silicate mass fraction. Non-SF regions have higher dust mass surface densities but lower dust-to-stellar mass ratios than SF regions, separating absolute dust content from dust content per unit stellar mass. Regions with larger specific H$\alpha$ surface brightness have larger dust-to-stellar mass ratios but lower inferred small-grain fractions, especially lower small-silicate fractions. In non-SF regions this quantity is interpreted as ionized-gas emission per unit stellar mass rather than as a direct sSFR. These model-dependent trends support a picture in which local dust processing changes the relative abundance of small grains and thereby shapes the attenuation-curve variations found across the series.

astro-ph.GA↗
arXiv2026

Mapping Dust Attenuation at Kiloparsec Scales. III. The 2175\AA\ Bump

We combine the SwiM_v4.2 Swift/UVOT+MaNGA catalog with 2MASS $K_s$ imaging to map the 2175{\AA} attenuation bump at kiloparsec scales in nearby galaxies. We use two complementary estimators: an ultraviolet-to-near-infrared attenuation-curve method, yielding $A_{bump}^{UOIR}$ and $B$ for 2487 high-continuum-S/N spaxels, and the NUV-only method of Battisti et al. (2025), yielding $A_{bump}^{NUV}$ and $k_{bump}$ for 7934 spaxels. The two absolute bump estimates agree well where they overlap. We compare bump strength with local stellar-population, emission-line, attenuation-curve, and geometric diagnostics after separating star-forming (SF) and non-SF regions. The strongest bumps occur at low specific H$\alpha$ surface brightness, $\Sigma_{\text{H}\alpha}/\Sigma_\ast$, especially in non-SF regions, where this ratio traces ionized-gas emission per unit stellar mass rather than sSFR. The bump also weakens with EW(H$\alpha$) and strengthens with $D_n4000$ and stellar age. In contrast, metallicity, inclination, galactocentric radius, $A_V$, and optical attenuation-curve slope are secondary predictors. The absolute strength $A_{bump}^{NUV}$ increases with $\Sigma_{\text{H}\alpha}$ and $\Sigma_\ast$, while the relative strengths $k_{bump}$ and $B$ do not, indicating that absolute bump amplitude partly follows dust column whereas normalized strengths better trace effective bump prominence. These results support local radiation-field processing of the 2175{\AA} carriers.

astro-ph.GA↗
arXiv2026

IntentTune: Using user demand and personalization to resolve "unknown" query intents for e-commerce search

Understanding user intent is fundamental to delivering relevant search results in e-commerce. However, substantial fraction of real-world queries are under-specified (e.g., "watch" or "shirt"), lacking explicit attributes such as gender or age group. This ambiguity poses a significant challenge for query intent detection models in e-commerce search systems, which must accurately infer latent user intent (e.g., age, gender) to support effective downstream retrieval. We introduce IntentTune, a framework for resolving ambiguous or under-specified query intents by leveraging either (1) user-specific behavioral signals including search history, browsing activity, and profile attributes or (2) population-level demand patterns aggregated across all users. Through experiments on real-world e-commerce data, we first demonstrate that population-level demand patterns alone are insufficient to reliably infer intent in under-specified queries. We then demonstrate that user-specific behavioral signals -- particularly prior search queries -- outperform both population-level statistics and static profile information for inferring gender, age group, product category, and size intent from underspecified queries.

cs.IR↗
arXiv2026

DynamicMem: A Long-Horizon Memory Benchmark in Real-World Settings

LLM agents increasingly act as personal assistants that must remember a user's profile over months: who they are (attributes), what they routinely do (habits), and what they prefer (preferences), and keep it updated as jobs, routines, and tastes drift. Existing benchmarks evaluate this "memory" ability through short, simplified interactions, missing three core properties of real behavior: the profile is heterogeneous, with attributes, habits, and preferences evolving on different timelines; changes are driven by external context such as seasons and life events; and evidence is rarely stated explicitly, instead scattered across many small actions in different apps that a memory system must infer from. We introduce DynamicMem, a synthetic benchmark that constructs 15 months of activity per user, providing long-term multi-app data that real users' privacy keeps out of reach. It provides user-consistent trajectories averaging 2.2M tokens and 1,772 grounded events per user across 16 applications such as e-commerce, fitness, and social platforms. The profile evolves over this period and is never given explicitly: each attribute, habit, or preference must be inferred from small signals scattered across apps. We evaluate at five quarterly checkpoints to track how systems scale as history grows. Benchmarking five representative systems exposes problems a single accuracy score hides: (i) profile reconstruction degrades with history length while service-task accuracy stays flat, despite both drawing on the same memory; (ii) no system both keeps facts that stay true and replaces facts that change, with errors clustering on preferences and on naming the exact referent; and (iii) over 93% of failures trace to what the memory retrieves, not to the model writing the answer, so the largest room for improvement lies in memory itself. Code: https://wenyaxie023.github.io/DynamicMem/

cs.CL↗
arXiv2026

Let LLMs Judge Each Other: Multi-Agent Peer-Reviewed Reasoning for Medical Question Answering

Objective: To enhance the accuracy, interpretability, and robustness of large language models (LLMs) in medical question answering (MedQA). Method: We designed a multi-agent peer-reviewed reasoning method in which multiple LLM agents independently generate chain-of-thought reasoning with candidate answers, then act as peer reviewers to evaluate each other's reasoning for factual correctness and logical soundness. The highest-rated reasoning chain is selected to produce the final answer. Experiments were conducted with five state-of-the-art LLMs (Llama-3.1-8B, Qwen2.5-7B, Phi-4, DeepSeek-LLM-7B, GPT-oss-20B) on three benchmark datasets: HeadQA, MedQA-USMLE, and PubMedQA. Performance was compared against single-model chain-of-thought reasoning and chain-of-thought-based majority voting. Results: Peer-reviewed reasoning consistently outperformed both baselines. The best model combination achieved an average accuracy of 0.820 across datasets, exceeding the strongest single model (0.777) and majority voting ensembles (up to 0.789). The method also scaled effectively with more participating models, while peer assessments reliably distinguished high- from low-quality reasoning chains. Conclusion: The proposed multi-agent peer-reviewed reasoning method enables LLMs to act as both solvers and evaluators, yielding superior performance in MedQA. By emphasizing reasoning quality rather than answer agreement alone, this approach improves accuracy, interpretability, and robustness, offering a promising direction for trustworthy biomedical AI systems.

cs.CL↗
arXiv2026

Iterating Toward Better Search: A Two-Agent Simulation Framework for Evaluating Agentic Search Architectures in E-Commerce

We present a modular two-agent simulation framework for evaluating conversational shopping assistant architectures. An independent buyer agent, configured with personas, missions, and patience levels, is paired with an interchangeable responder that integrates with a real e-commerce search API. Holding the buyer constant across experiments enables controlled comparison of responder designs on identical scenarios. Using 2011 conversations across 14 persona buckets, we establish four empirical findings. First, rolling-window memory outperforms intent-extraction memory on all quality metrics while being 35% faster per query. Second, illustrating rapid evidence-driven iteration, a systematic failure analysis of a responder version enables targeted fixes that reduce failure and near-failure rates by 62% across the full dataset. Third, swapping the responder LLM backbone from Gemini~2.5 to Llama~3.3~70B costs 0.16--0.45 points despite identical architecture. Finally, we document systematic philosophical disagreement between frontier LLM judges: Gemini rewards process correctness while Claude demands concrete outcomes, despite using the same evaluation prompt.

cs.AI↗
arXiv2026

Asuka-Bench: Benchmarking Code Agents on Underspecified User Intent and Multi-Round Refinement

Existing code-generation benchmarks score a single mapping from a complete prompt to a one-shot output. However, real web development is different. Users seldom write a full spec at the start; many requirements only become clear once they look at an intermediate result and react to it. We present Asuka-Bench, a benchmark that pairs underspecified user intent with multi-round refinement, grounded in browser-rendered behavior. Each task is resolved through a closed loop: a Code Agent generates a web project, a UI Agent executes test cases on the deployed site, and a User LLM turns evaluation outcomes into natural-language feedback for the next round. The benchmark comprises 50 web tasks with 784 evaluation criteria and 2402 expected outcomes. We benchmark 8 LLMs across 2 agent frameworks. The results separate models clearly: weighted Task Pass Rate varies by 38 percentage points and models also differ substantially in their ability to repair from feedback. Asuka-Bench is also far from saturated: even the strongest model completes only 52% of projects after three rounds.

cs.SE↗
arXiv2026

Bayesian Region Selection and Prediction in Poisson Regression with Spatially Dependent Global-Local Shrinkage Prior

High-dimensional spatially correlated covariates are common in regression models encountered in environmental sciences and other fields. In such models, the regression coefficients often exhibit a sparse structure with spatial dependence. Although standard variable selection approaches can help detect the sparse structure, incorporating the dependence into variable selection helps recover spatially contiguous signals and improves prediction accuracy. Motivated by a real-world challenge in hurricane count prediction, we propose a novel neighborhood-structured global-local shrinkage prior for prediction and region selection in Poisson regression with spatial covariates. The proposed prior combines the Conditional Auto-Regressive (CAR) prior with a Super Heavy-tailed prior to introduce spatial dependence among the coefficients while ensuring appropriate shrinkage effects for covariate selection. We develop an efficient Metropolis-within-Gibbs sampler for computation that accommodates the count data. Extensive simulation studies demonstrate that the proposed model excels when signals are weak and adjacent and the spatial dependence in covariates is strong. In the application of hurricane prediction from the north Atlantic, our method outperforms traditional regression-based approaches and rivals the benchmark oracle model.

stat.ME↗
arXiv2026

OralMLLM-Bench: Evaluating Cognitive Capabilities of Multimodal Large Language Models in Dental Practice

Multimodal large language models (MLLMs) have emerged as a promising paradigm for dental image analysis. However, their ability to capture the multi-level cognitive processes required for radiographic analysis remains unclear. Here, we present a comprehensive benchmark to evaluate the cognitive capabilities of MLLMs in dental radiographic analysis. It spans three critical imaging modalities, i.e., periapical, panoramic, and lateral cephalometric radiographs, and defines four cognitive categories: perception, comprehension, prediction, and decision-making. The benchmark comprises 27 clinically grounded tasks derived from public datasets, with manually curated annotations and 3,820 clinician assessments for evaluation. Six frontier MLLMs, including GPT-5.2 and GLM-4.6, are evaluated. We demonstrate the performance gap between MLLMs and clinicians in dental practice, delineate model strengths and limitations, characterize failure patterns, and provide recommendations for improvement. This data resource will facilitate the development of next-generation artificial intelligence systems aligned with clinical cognition, safety requirements, and workflow complexity in dental practice.

cs.CL↗
arXiv2026

General365: Benchmarking General Reasoning in Large Language Models Across Diverse and Challenging Tasks

Contemporary large language models (LLMs) have demonstrated remarkable reasoning capabilities, particularly in specialized domains like mathematics and physics. However, their ability to generalize these reasoning skills to more general and broader contexts--often termed general reasoning--remains under-explored. Unlike domain-specific reasoning, general reasoning relies less on expert knowledge but still presents formidable reasoning challenges, such as complex constraints, nested logical branches, and semantic interference. To address this gap, we introduce General365, a benchmark specifically designed to assess general reasoning in LLMs. By restricting background knowledge to a K-12 level, General365 explicitly decouples reasoning from specialized expertise. The benchmark comprises 365 seed problems and 1,095 variant problems across eight categories, ensuring both high difficulty and diversity. Evaluations across 26 leading LLMs reveal that even the top-performing model achieves only 62.8% accuracy, in stark contrast to the near-perfect performances of LLMs in math and physics benchmarks. These results suggest that the reasoning abilities of current LLMs are heavily domain-dependent, leaving significant room for improvement in broader applications. We envision General365 as a catalyst for advancing LLM reasoning beyond domain-specific tasks toward robust, general-purpose real-world scenarios. Code, Dataset, and Leaderboard: https://general365.github.io

cs.CL↗
arXiv2026

PRIME: Prototype-Driven Multimodal Pretraining for Cancer Prognosis with Missing Modalities

Multimodal self-supervised pretraining offers a promising route to cancer prognosis by integrating histopathology whole-slide images, gene expression, and pathology reports, yet most existing approaches require fully paired and complete inputs. In practice, clinical cohorts are fragmented and often miss one or more modalities, limiting both supervised fusion and scalable multimodal pretraining. We propose PRIME, a missing-aware multimodal self-supervised pretraining framework that learns robust and transferable representations from partially observed cohorts. PRIME maps heterogeneous modality embeddings into a unified token space and introduces a shared prototype memory bank for latent-space semantic imputation via patient-level consensus retrieval, producing structurally aligned tokens without reconstructing raw signals. Two complementary pretraining objectives: inter-modality alignment and post-fusion consistency under structured missingness augmentation, jointly learn representations that remain predictive under arbitrary modality subsets. We evaluate PRIME on The Cancer Genome Atlas with label-free pretraining on 32 cancer types and downstream 5-fold evaluation on five cohorts across overall survival prediction, 3-year mortality classification, and 3-year recurrence classification. PRIME achieves the best macro-average performance among all compared methods, reaching 0.653 C-index, 0.689 AUROC, and 0.637 AUROC on the three tasks, respectively, while improving robustness under test-time missingness and supporting parameter-efficient and label-efficient adaptation. These results support missing-aware multimodal pretraining as a practical strategy for prognosis modeling in fragmented clinical data settings.

cs.LG↗
arXiv2026

EpiScreen: Early Epilepsy Detection from Electronic Health Records with Large Language Models

Epilepsy and psychogenic non-epileptic seizures often present with similar seizure-like manifestations but require fundamentally different management strategies. Misdiagnosis is common and can lead to prolonged diagnostic delays, unnecessary treatments, and substantial patient morbidity. Although prolonged video-electroencephalography is the diagnostic gold standard, its high cost and limited accessibility hinder timely diagnosis. Here, we developed a low-cost, effective approach, EpiScreen, for early epilepsy detection by utilizing routinely collected clinical notes from electronic health records. Through fine-tuning large language models on labeled notes, EpiScreen achieved an AUC of up to 0.875 on the MIMIC-IV dataset and 0.980 on a private cohort of the University of Minnesota. In a clinician-AI collaboration setting, EpiScreen-assisted neurologists outperformed unaided experts by up to 10.9%. Overall, this study demonstrates that EpiScreen supports early epilepsy detection, facilitating timely and cost-effective screening that may reduce diagnostic delays and avoid unnecessary interventions, particularly in resource-limited regions.

cs.CL↗
arXiv2026

Contrasting evolutionary pathways of fast- and slow-rotating galaxies in the green valley

We investigate the evolutionary pathways of green valley (GV) galaxies drawn from the SDSS-IV/MaNGA survey. The GV sample is divided into fast- and slow-rotating galaxies based on stellar spin, and their stellar and gas-phase metallicities are compared. Fast-rotating galaxies exhibit systematically higher metallicities than slow-rotating galaxies in both gas and stars. However, the gas-phase difference is significant only at low stellar masses, while the stellar metallicity offset persists across the full mass range. Using a simple yet physically motivated chemical evolution model, optimised to jointly fit gas-phase metallicities and integrated stellar spectra, we reconstruct the star formation and chemical enrichment histories of individual galaxies and constrain gas inflow and outflow parameters. At low stellar masses, fast- and slow-rotating galaxies show similar gas-infall and star formation timescales, but the the slower population experienced stronger outflows which reduce their chemical content in both gas and stars. At high masses, the combination of reduced pristine gas inflow and more efficient gas removal in slow-rotating galaxies produce gas-phase metallicities comparable to fast-rotating galaxies but systematically lower stellar metallicities. These differences suggest distinct evolutionary pathways for GV galaxies. Slow-rotating galaxies likely experienced more mergers, usually associated with strong gas removal processes, leading to their systematically lower metallicities. At low masses, stronger supernova-driven outflows reduce their chemical content while leaving star-formation timescales similar to fast-rotating galaxies. At high masses, merger-triggered AGN feedback may rapidly deplete and suppress gas infall, producing the shorter star-formation timescales seen in slow-rotating galaxies. Alternative environmental and assembly-driven scenarios are also discussed.

astro-ph.GA↗
arXiv2026

MedCL-Bench: Benchmarking stability-efficiency trade-offs and scaling in biomedical continual learning

Medical language models must be updated as evidence and terminology evolve, yet sequential updating can trigger catastrophic forgetting. Although biomedical NLP has many static benchmarks, no unified, task-diverse benchmark exists for evaluating continual learning under standardized protocols, robustness to task order and compute-aware reporting. We introduce MedCL-Bench, which streams ten biomedical NLP datasets spanning five task families and evaluates eleven continual learning strategies across eight task orders, reporting retention, transfer, and GPU-hour cost. Across backbones and task orders, direct sequential fine-tuning on incoming tasks induces catastrophic forgetting, causing update-induced performance regressions on prior tasks. Continual learning methods occupy distinct retention-compute frontiers: parameter-isolation provides the best retention per GPU-hour, replay offers strong protection at higher cost, and regularization yields limited benefit. Forgetting is task-dependent, with multi-label topic classification most vulnerable and constrained-output tasks more robust. MedCL-Bench provides a reproducible framework for auditing model updates before deployment.

cs.AI↗
arXiv2026

HeartAgent: An Autonomous Agent System for Explainable Differential Diagnosis in Cardiology

Heart diseases remain a leading cause of morbidity and mortality worldwide, necessitating accurate and trustworthy differential diagnosis. However, existing artificial intelligence-based diagnostic methods are often limited by insufficient cardiology knowledge, inadequate support for complex reasoning, and poor interpretability. Here we present HeartAgent, a cardiology-specific agent system designed to support a reliable and explainable differential diagnosis. HeartAgent integrates customized tools and curated data resources and orchestrates multiple specialized sub-agents to perform complex reasoning while generating transparent reasoning trajectories and verifiable supporting references. Evaluated on the MIMIC dataset and a private electronic health records cohort, HeartAgent achieved over 36% and 20% improvements over established comparative methods, in top-3 diagnostic accuracy, respectively. Additionally, clinicians assisted by HeartAgent demonstrated gains of 26.9% in diagnostic accuracy and 22.7% in explanatory quality compared with unaided experts. These results demonstrate that HeartAgent provides reliable, explainable, and clinically actionable decision support for cardiovascular care.

cs.CL↗
arXiv2026

To Reason or Not to: Selective Chain-of-Thought in Medical Question Answering

Objective: To improve the efficiency of medical question answering (MedQA) with large language models (LLMs) by avoiding unnecessary reasoning while maintaining accuracy. Methods: We propose Selective Chain-of-Thought (Selective CoT), an inference-time strategy that first predicts whether a question requires reasoning and generates a rationale only when needed. Two open-source LLMs (Llama-3.1-8B and Qwen-2.5-7B) were evaluated on four biomedical QA benchmarks-HeadQA, MedQA-USMLE, MedMCQA, and PubMedQA. Metrics included accuracy, total generated tokens, and inference time. Results: Selective CoT reduced inference time by 13-45% and token usage by 8-47% with minimal accuracy loss ($\leq$4\%). In some model-task pairs, it achieved both higher accuracy and greater efficiency than standard CoT. Compared with fixed-length CoT, Selective CoT reached similar or superior accuracy at substantially lower computational cost. Discussion: Selective CoT dynamically balances reasoning depth and efficiency by invoking explicit reasoning only when beneficial, reducing redundancy on recall-type questions while preserving interpretability. Conclusion: Selective CoT provides a simple, model-agnostic, and cost-effective approach for medical QA, aligning reasoning effort with question complexity to enhance real-world deployability of LLM-based clinical systems.

cs.CL↗
arXiv2026

LongCat-Flash-Thinking-2601 Technical Report

We introduce LongCat-Flash-Thinking-2601, a 560-billion-parameter open-source Mixture-of-Experts (MoE) reasoning model with superior agentic reasoning capability. LongCat-Flash-Thinking-2601 achieves state-of-the-art performance among open-source models on a wide range of agentic benchmarks, including agentic search, agentic tool use, and tool-integrated reasoning. Beyond benchmark performance, the model demonstrates strong generalization to complex tool interactions and robust behavior under noisy real-world environments. Its advanced capability stems from a unified training framework that combines domain-parallel expert training with subsequent fusion, together with an end-to-end co-design of data construction, environments, algorithms, and infrastructure spanning from pre-training to post-training. In particular, the model's strong generalization capability in complex tool-use are driven by our in-depth exploration of environment scaling and principled task construction. To optimize long-tailed, skewed generation and multi-turn agentic interactions, and to enable stable training across over 10,000 environments spanning more than 20 domains, we systematically extend our asynchronous reinforcement learning framework, DORA, for stable and efficient large-scale multi-environment training. Furthermore, recognizing that real-world tasks are inherently noisy, we conduct a systematic analysis and decomposition of real-world noise patterns, and design targeted training procedures to explicitly incorporate such imperfections into the training process, resulting in improved robustness for real-world applications. To further enhance performance on complex reasoning tasks, we introduce a Heavy Thinking mode that enables effective test-time scaling by jointly expanding reasoning depth and width through intensive parallel thinking.

cs.AI↗