arXiv subjects

Nan Zhou

Publications and source records attributed to Nan Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Visually Irrelevant Hallucinations in Medical Vision-Language Models

Medical vision-language models (VLMs) show strong performance in radiology report generation (RRG) and medical visual question answering (Med-VQA), producing free-form clinical descriptions and answers from medical images. However, they may generate clinically meaningful statements that lack visual evidence or contradict image content. We refer to these errors as visually irrelevant hallucinations and investigate them at the phrase level. We propose CLEAR (Counterfactual Localized Evaluation of Alignment and Response), a model-agnostic post-hoc framework for phrase-level hallucination risk estimation. CLEAR decomposes generated responses into visually assessable clinical phrases. For each phrase, phrase-conditioned grounding identifies a candidate visual region, which is selectively perturbed by a localized counterfactual intervention. CLEAR characterizes the intervention using two complementary auditing cues: response cues, capturing semantic and clinically meaningful changes in the generated response, and alignment cues, measuring changes in image-phrase compatibility. These cues are normalized and jointly modeled to estimate phrase-level hallucination risk. Across four datasets spanning RRG and Med-VQA, CLEAR outperforms state-of-the-art hallucination detection baselines across multiple medical and general-purpose VLMs. On MIMIC-CXR, CLEAR improves AUPRC from 85.10% to 90.23%. Its effectiveness is further validated on a GT-derived controlled VIH benchmark. These results establish visually irrelevant hallucination as an important fine-grained reliability failure mode in medical VLMs and demonstrate the value of model-agnostic visual-support auditing for more trustworthy multimodal clinical generation.

cs.CV↗

arXiv2026

Commercial building decarbonization across countries: Future emission trajectories and mitigation priorities

With energy-related carbon dioxide emissions remaining at high levels, decarbonizing commercial building operations across key economies is critical for achieving global climate targets. This study assesses future operational carbon trajectories, mitigation potential, and the driving factors of this potential in commercial building operations across 16 countries from 2024 to 2060 by combining a Kaya-based emission model and Monte Carlo uncertainty analysis. The key findings indicate that (1) under the business-as-usual scenario, China's commercial building operational carbon emissions are projected to peak at 1150 (+-120) million tonnes of carbon dioxide (MtCO2) in 2033 (+-5 years), while the remaining countries exhibit differentiated carbon lock-in levels, with the highest level in the United States at 470 (+-89) MtCO2. (2) The total emission reduction potential increases from 352 MtCO2 in 2030 to 1119 MtCO2 in 2050 and stabilizes at 1123 MtCO2 in 2060, with China, the United States, and Japan contributing 84.7%-88.9% of the total potential. (3) By 2060, the mitigation potential from energy intensity reduction is concentrated mainly in China and the United States, reaching 52 and 32 MtCO2, respectively; end-use decomposition further identifies space heating and appliances as key targets for decarbonizing commercial building operations. In summary, this study characterizes future operational carbon trajectories and clarifies the drivers and end-use priorities of mitigation potential in commercial building operations, providing cross-country evidence for differentiated strategies for long-term commercial building decarbonization.

physics.soc-ph↗

arXiv2026

Coactions of weak multiplier Hopf algebras

In this article, we investigate the notion of a coaction for a weak multiplier Hopf algebra. We show that there is a duality for actions and coactions of weak algebraic quantum groups. Then we obtain the well known Blattner-Montgomery duality theorem for weak algebraic quantum groups in the framework of dual pairs. We form a Morita context when the weak multiplier Hopf algebra admits a two-sided faithful integral and show that the coaction is Galois if and only if the map in the Morita context is surjective. As an application, we obtain the result that a grading by any groupoid $G$ can be considered as a $\mathrm{C}G$-comodule algebra. We also construct a Morita context connecting the smash product $A\# \mathrm{C}(G)$ and the coinvariant subalgebra $A^{\mathrm{C}(G)}$. Finally we construct an isomorphism between $(R\# \mathrm{C}(G))\# \mathrm{C}G$ and $R\otimes (\mathrm{C}(G)\# \mathrm{C}G)$ when $R$ is graded by $G$.

math.QA↗

arXiv2026

WorldMemBench: A Comprehensive Benchmark for Memory of Video World Models

As video models enable longer and more realistic interactive experiences, maintaining a world consistent with its history becomes essential. This requires both preserving previously observed content and ensuring coherent world evolution over time. World memory is a critical capability to evaluate alongside visual fidelity and video duration. We introduce WorldMemBench, a benchmark for systematically evaluating world memory in video world models. Across 260 test cases, it assesses static memory through the retention of human identity and appearance, object appearance and geometry, and environmental content and geometry, and dynamic memory through event progression and state persistence. Using controlled camera departures and multi-revisit trajectories, the benchmark evaluates static content against fixed initial references and dynamic events against expected state transitions. It combines appearance and geometry metrics with event-specific assessments, and reports video quality and camera-following accuracy as complementary diagnostics. Evaluation of nine representative models shows that strong static retention does not imply reliable dynamic state evolution, while repeated revisits reveal memory degradation beyond the first return. Human studies support the alignment between our automatic evaluations and human judgments. WorldMemBench provides a systematic testbed for diagnosing memory failures and guiding the development of persistent, temporally coherent video worlds.

cs.CV↗

arXiv2026

Unraveling persistent urban-rural gaps: A long-term provincial analysis of residential heating and cooling loads

Climate change and rising thermal comfort demand make residential heating and cooling central to building-sector decarbonization. This study presents a bottom-up modeling framework to estimate residential heating and cooling loads across 30 Chinese provinces. The model, developed using EnergyPlus simulations of representative building prototypes, captures heating and cooling load patterns in both urban and rural housing over the period 1980-2024. The results indicate that: (1) In 2020, Guangdong recorded the highest cooling loads (76.5 TWh/a urban; 63.0 TWh/a rural). Henan exhibited the highest rural heating load (174.6 TWh/a), while urban heating loads were highest in Liaoning and Shandong. (2) Between 1980 and 2024, average urban cooling loads increased from 12.4 to 15.1 kWh/m2a, whereas rural cooling loads declined from 22.63 to 19.87 kWh/m2a. Urban heating loads decreased from 44.08 to 39.92 kWh/m2a, and rural heating loads declined more markedly from 100.15 to 72.42 kWh/m2a. (3) Urban residential floor area has exceeded rural stock in 22 provinces in recent years, compared with only four provinces in 2000. Moreover, the existence of 12 urban energy-efficiency standards versus a single rural standard reflects a persistent regulatory disparity. Counterfactual and factor-contribution analyses further indicate that the associated envelope-performance gap is a major contributor to urban-rural load-intensity differences, while building form has a more context-dependent influence. The proposed framework provides a replicable basis for region-specific clean heating strategies and differentiated building standards to support carbon neutrality.

physics.soc-ph↗

arXiv2026

iFuzz-Meta: An Interpretable Fuzzy Learning Framework Bridging Top-Down and Bottom-Up Knowledge Integration

Interpretable representation learning remains a key challenge in modern neural computation, particularly when models are expected not only to perform but also to explain their reasoning. This paper introduces iFuzz-Meta, an interpretable fuzzy rule-based learning framework that preserves human-understandable reasoning structures within modern neural architectures. Each fuzzy rule corresponds to a semantic and spatial prototype defined in the original feature space, enabling transparent inference and direct interpretability. Meta-learning is employed as an analytical paradigm to examine how these interpretable rules reorganize across tasks and domains, providing a principled means to link algorithmic adaptation with cognitive representation. A knowledge-guided regularization mechanism further enables a top-down-bottom-up integration, in which theoretical priors act as soft inductive biases while data-driven learning refines and extends them. This dual process ensures that adaptation proceeds along semantically and physiologically meaningful trajectories, rather than arbitrary parameter shifts. Evaluations demonstrate that iFuzz-Meta achieves interpretable reasoning and stable cross-domain generalization, establishing a potential general pathway toward explainable and knowledge-aware fuzzy systems.

cs.LG↗

arXiv2026

Hallucination Detection and Correction in Medical VLMs via Counter-Evidence Verification

Vision-Language models (VLMs) reliability in medical diagnosis is challenged by trust-undermining hallucinations. Existing hallucination detection approaches mainly focus on identifying factual inconsistencies between generated text and reference data. While some studies analyze where models attend in images, they seldom verify whether such attention truly reflects the visual evidence supporting the generated text. To address this gap, we propose Co}unter-Evidence Verification (CoEV), a training-free plug-and-play framework that detects and corrects hallucinations through evidence-based factual consistency verification. CoEV performs bidirectional verification between textual assertions and visual evidence, testing whether each statement is supported by its corresponding evidence region, and assigns each statement into a four-quadrant diagnostic map capturing combinations of text factuality and visual grounding. CoEV detects hallucinated content and serves as a post hoc refinement tool, correcting hallucinations without retraining. Extensive experiments on four medical datasets show that CoEV combats hallucinations in VLMs.For hallucination detection, CoEV consistently outperforms existing methods, improving average PR-AUC and ROC-AUC by 3.0% and 3.9% absolute points respectively, with notable gains of up to 18.5% in specific VQA scenarios. For hallucination correction, it improves Micro-F1 by up to 12.5%, reduces hallucination rates by over 11.9% on medical report generation, and also boosts medical VQA accuracy. These results show that CoEV enables reliable detection and correction of hallucinations, providing clinicians with dependable, evidence-based cues for diagnosis. Code will be released upon acceptance.

cs.CV↗

arXiv2026

SPACE: Source-free Proxy Anchor Concept Erasure for MLLMs

As Multimodal Large Language Models (MLLMs) face growing privacy risks and regulatory constraints, machine unlearning (MU) has emerged as a crucial solution for removing sensitive data while preserving model performance. However, existing MU methods typically rely on visual data of the target concepts, which is often unavailable due to strict data retention policies, thus creating a demand for source-free unlearning approaches that operate without access to the target data. In this work, we propose Source-free Proxy Anchor Concept Erasure (SPACE), the first source-free unlearning framework specialized for MLLMs. SPACE consists of two stages: (1) Text-Guided Proxy Anchor Selection (TPAS), which retrieves semantically aligned proxy anchors from the shared feature space. (2) Dual-Constraint Semantic Isolation (DCSI), which optimizes these anchors to indirectly erase target concepts. DCSI confines updates to the null space of retained knowledge, ensuring structural integrity. We theoretically prove that SPACE strictly bounds the perturbation on retained knowledge and maximizes feature spectral entropy, thereby maintaining the model's performance. Furthermore, extensive experiments across six datasets show that SPACE achieves performance comparable to that of state-of-the-art data-dependent methods, validating its effectiveness in source-free MU scenarios. The source code will be released.

cs.LG↗

arXiv2026

Synergetic capacity planning of private and public EV charging piles via city-scale multiobjective optimization

Rapid electric vehicle (EV) expansion necessitates optimized charging infrastructure to bridge the persistent gaps between vehicle growth and charger availability. This study develops a demand-driven framework for city-scale EV charging demand assessment and charging pile capacity planning. It employs a bottom-up estimation approach to quantify electricity demand and a Harris Hawks Optimization algorithm to solve capacity planning challenges, capturing spatiotemporal demand variations across powertrain types and guiding allocation over 2022-2030 in Chongqing, China. The results show that (1) compared with June 2022, monthly EV electricity consumption tripled to 57.5 gigawatt-hours by the end of 2024, characterized by significant seasonal volatility and a structural shift in which the combined share of plug-in hybrid electric vehicles and extended-range electric vehicles reached 57.6%, necessitating a transition toward technology-specific infrastructure planning; (2) historical evaluations reveal a marked spatial mismatch, with actual deployment heavily concentrated in the urban core while public charging capacity consistently lagging behind demand, whereas the proposed optimized configuration achieved a superior comprehensive performance score of 0.28, compared to 0.65 for actual deployment, in balancing service adequacy across the "Core-Suburban-Exurban" hierarchy; and (3) by 2030, Chongqing is projected to require approximately 1.8 million charging units to sustain a stable 9:1 private-to-public ratio, a synergetic strategy expects to significantly mitigate urban-rural service disparities and enhance overall system resilience and grid compatibility. Ultimately, this study provides a versatile, spatially explicit tool for policymakers to support sustainable and cost-effective EV infrastructure deployment aligned with long-term electrification targets.

physics.gen-ph↗

arXiv2026

What Makes Synthetic Data Effective in Image Segmentation

Driven by rapid advances in large-scale generative models, synthetic data has emerged as a promising solution for visual understanding. While modern diffusion models achieve remarkable photorealistic image synthesis, their potential in complex visual segmentation tasks remains underexplored. In this work, we conduct a systematic analysis of synthetic images from state-of-the-art diffusion models to uncover the factors governing their utility. In particular, synthetic images characterized by dense scene composition and fine instance fidelity demonstrate distinctive benefits, yielding significantly more discriminative spatial representations. Building on these insights, we propose SENSE, a unified framework that leverages flexible and scalable synthetic data to substantially enhance segmentation performance. Notably, SENSE is model-agnostic, compatible with diverse architectures (e.g., DPT and Mask2Former), and scales effectively across models with varying parameter capacities. Extensive experiments on Cityscapes, COCO, and ADE20K validate the effectiveness and generalization capability of our approach. Code is available at https://github.com/zhang0jhon/SENSE.

cs.CV↗

arXiv2026

City-level energy and emission assessment based on 20+ million electric vehicle registrations in China

China, the world's largest electric vehicle (EV) market, plays a pivotal role in global decarbonization of the transport sector. We present the first high-resolution assessment of EV adoption in 295 cities, utilizing more than 20 million registrations of 586 EV models tracked monthly from 2022 to 2024 and projecting transition pathways to 2035. Real-world data reveal that EVs are 30.9-212.8 megajoules per 100 km more energy efficient than internal combustion vehicles, yet their carbon intensities range from 18.2 to 270.4 gCO2/km among provinces. The limited electrification of hybrids means that gasoline still accounts for 44% of EV energy use. Scenario projections suggest that emissions will peak about 2030 at 21.1-30.9 megatonnes of CO2 and decline by 2035 under continued market transition. The findings establish an empirical foundation for accurate emissions accounting, emphasize the need to reduce regional disparities in adoptability, and offer globally relevant insights for road-transport decarbonization.

physics.soc-ph↗

arXiv2026

Breaking the Communication-Accuracy Trade-off: A Sparsified Information Diffusion Framework for Multi-Agent Collaborative Perception

The growing relevance of multi-agent systems has drawn increasing focus on communication-efficient filters for collaborative perception to alleviate the system's communication burden. While the event-triggered (ET) mechanism can improve communication efficiency in collaborative state estimation, an inevitable trade-off exists between estimation accuracy and communication cost in ET filters. This paper proposes a fast and accurate ET diffusion-based filter for real-time multi-agent collaborative target tracking, aiming to reduce the system's data transmission without compromise in tracking performance. The proposed filter achieves improved tracking accuracy, reduced data transmission, and accelerated convergence using an error-minimized ET cubature information filter (CIF) for local estimation, and a correlation-aware diffusion strategy for global fusion. The experimental results confirm the scalability of the proposed EDC-CIF algorithm and demonstrate its efficacy in simultaneously reducing estimation error and computation time while significantly enhancing communication efficiency.

cs.MA↗

arXiv2026

CoVFT: Context-aware Visual Fine-tuning for Multimodal Large Language Models

Multimodal large language models (MLLMs) achieve remarkable progress in cross-modal perception and reasoning, yet a fundamental question remains unresolved: should the vision encoder be fine-tuned or frozen? Despite the success of models such as LLaVA and Qwen-VL, inconsistent design choices and heterogeneous training setups hinder a unified understanding of visual fine-tuning (VFT) in MLLMs. Through a configuration-aligned benchmark, we find that existing VFT methods fail to consistently outperform the frozen baseline across multimodal tasks. Our analysis suggests that this instability arises from visual preference conflicts, where the context-agnostic nature of vision encoders induces divergent parameter updates under diverse multimodal context. To address this issue, we propose the Context-aware Visual Fine-tuning (CoVFT) framework, which explicitly incorporates multimodal context into visual adaptation. By integrating a Context Vector Extraction (CVE) and a Contextual Mixture-of-Experts (CoMoE) module, CoVFT decomposes conflicting optimization signals and enables stable, context-sensitive visual updates. Extensive experiments on 12 multimodal benchmarks demonstrate that CoVFT achieves state-of-the-art performance with superior stability. Notably, fine-tuning a 7B MLLM with CoVFT surpasses the average performance of its 13B counterpart, revealing substantial untapped potential in visual encoder optimization within MLLMs.

cs.CV↗

arXiv2026

Decarbonizing China's private passenger vehicles: A dynamic material flow assessment of metal demands and embodied emissions

The continuous growth of China's private passenger vehicle fleet has intensified material demand and embodied carbon emissions, underscoring the need for effective decarbonization pathways. This study develops a transferable, dynamic material flow analysis framework to assess vehicle stocks, metal flows (steel, aluminum, and copper), and embodied emissions from 2000 to 2070, and to quantify the contributions of demand-side and technology-side efficiency measures. The results reveal that: (1) The vehicle fleet is projected to peak at 327-507 million vehicles by mid-century, with new energy vehicles dominating both in-use stocks and end-of-life flows by the 2040s. (2) Cumulative metal demand is projected to reach 1914-2990 million tonnes over the upcoming five decades, with 879-1320 million tonnes supplied from secondary sources under baseline conditions. Technology-oriented measures substantially enhance recycling performance, enabling secondary steel to fully meet manufacturing demand and allowing aluminum and copper cycles to approach near closure by 2070. (3) Correspondingly, cumulative embodied carbon emissions from vehicle metals by 2070 range from 4958 to 9218 megatonnes of carbon dioxide, with technological upgrading reducing emissions by 1051-1619 megatonnes. In collaborative scenarios, demand management accounts for 64.3% of total emission reductions, while technology-oriented measures become increasingly important over the medium to long term. Overall, the findings demonstrate that unmanaged demand growth can substantially offset technological mitigation gains, highlighting the necessity of integrated demand- and technology-oriented strategies. This study provides a systemic and transferable framework to guide circular economy development and deep decarbonization transitions in vehicle fleets in China and other emerging economies.

physics.soc-ph↗

arXiv2026

Code-A1: Adversarial Evolving of Code LLM and Test LLM via Reinforcement Learning

Reinforcement learning for code generation relies on verifiable rewards from unit test pass rates. Yet high-quality test suites are scarce, existing datasets offer limited coverage, and static rewards fail to adapt as models improve. Recent self-play methods unify code and test generation in a single model, but face a inherent dilemma: white-box access leads to self-collusion where the model produces trivial tests for easy rewards, yet black-box restriction yields generic tests that miss implementation-specific bugs. We introduce Code-A1, an adversarial co-evolution framework that jointly optimizes a Code LLM and a Test LLM with opposing objectives. The Code LLM is rewarded for passing more tests, while the Test LLM is rewarded for exposing more defects. This architectural separation eliminates self-collusion risks and safely enables white-box test generation, where the Test LLM can inspect candidate code to craft targeted adversarial tests. We further introduce a Mistake Book mechanism for experience replay and a composite reward balancing test validity with adversarial difficulty. Experiments on Qwen2.5-Coder models demonstrate that Code-A1 achieves code generation performance matching or exceeding models trained on human-annotated tests, while significantly improving test generation capability.

cs.CL↗

arXiv2026

A multimodal vision foundation model for generalizable knee pathology

Musculoskeletal disorders represent a leading cause of global disability, creating an urgent demand for precise interpretation of medical imaging. Current artificial intelligence (AI) approaches in orthopedics predominantly rely on task-specific, supervised learning paradigms. These methods are inherently fragmented, require extensive annotated datasets, and often lack generalizability across different modalities and clinical scenarios. The development of foundation models in this field has been constrained by the scarcity of large-scale, curated, and open-source musculoskeletal datasets. To address these challenges, we introduce OrthoFoundation, a multimodal vision foundation model optimized for musculoskeletal pathology. We constructed a pre-training dataset of 1.2 million unlabeled knee X-ray and MRI images from internal and public databases. Utilizing a Dinov3 backbone, the model was trained via self-supervised contrastive learning to capture robust radiological representations. OrthoFoundation achieves state-of-the-art (SOTA) performance across 14 downstream tasks. It attained superior accuracy in X-ray osteoarthritis diagnosis and ranked first in MRI structural injury detection. The model demonstrated remarkable label efficiency, matching supervised baselines using only 50% of labeled data. Furthermore, despite being pre-trained on knee images, OrthoFoundation exhibited exceptional cross-anatomy generalization to the hip, shoulder, and ankle. OrthoFoundation represents a significant advancement toward general-purpose AI for musculoskeletal imaging. By learning fundamental, joint-agnostic radiological semantics from large-scale multimodal data, it overcomes the limitations of conventional models, which provides a robust framework for reducing annotation burdens and enhancing diagnostic accuracy in clinical practice.

cs.CV↗

arXiv2025

The Illusion of Clinical Reasoning: A Benchmark Reveals the Pervasive Gap in Vision-Language Models for Clinical Competency

Background: The rapid integration of foundation models into clinical practice and public health necessitates a rigorous evaluation of their true clinical reasoning capabilities beyond narrow examination success. Current benchmarks, typically based on medical licensing exams or curated vignettes, fail to capture the integrated, multimodal reasoning essential for real-world patient care. Methods: We developed the Bones and Joints (B&J) Benchmark, a comprehensive evaluation framework comprising 1,245 questions derived from real-world patient cases in orthopedics and sports medicine. This benchmark assesses models across 7 tasks that mirror the clinical reasoning pathway, including knowledge recall, text and image interpretation, diagnosis generation, treatment planning, and rationale provision. We evaluated eleven vision-language models (VLMs) and six large language models (LLMs), comparing their performance against expert-derived ground truth. Results: Our results demonstrate a pronounced performance gap between task types. While state-of-the-art models achieved high accuracy, exceeding 90%, on structured multiple-choice questions, their performance markedly declined on open-ended tasks requiring multimodal integration, with accuracy scarcely reaching 60%. VLMs demonstrated substantial limitations in interpreting medical images and frequently exhibited severe text-driven hallucinations, often ignoring contradictory visual evidence. Notably, models specifically fine-tuned for medical applications showed no consistent advantage over general-purpose counterparts. Conclusions: Current artificial intelligence models are not yet clinically competent for complex, multimodal reasoning. Their safe deployment should currently be limited to supportive, text-based roles. Future advancement in core clinical tasks awaits fundamental breakthroughs in multimodal integration and visual understanding.

cs.CV↗

arXiv2025

Distinct Suppression Mechanisms of Superconductivity by Magnetic Domains and Spin Fluctuations in EuFe2(As1-xPx)2 superconductors

Using ac composite magnetoelectric technique, we map the phase diagrams of EuFe(As1-xPx)2 to resolve the interplay between superconductivity and ferromagnetism. For samples with Tc TFM, strong short-range spin correlations and phase boundaries within a multiphase coexistence regime near the triple point act as potent pair-breaking centers, leading to pronounced Hc2 suppression.

cond-mat.supr-con↗