In recent years,large vision-language models(VLMs)have achieved significant breakthroughs in cross-modal understanding and generation.However,the safety issues arising from their multimodal interactions become promine...In recent years,large vision-language models(VLMs)have achieved significant breakthroughs in cross-modal understanding and generation.However,the safety issues arising from their multimodal interactions become prominent.VLMs are vulnerable to jailbreak attacks,where attackers craft carefully designed prompts to bypass safety mechanisms,leading them to generate harmful content.To address this,we investigate the alignment between visual inputs and task execution,uncovering locality defects and attention biases in VLMs.Based on these findings,we propose VOTI,a novel jailbreak framework leveraging visual obfuscation and task induction.VOTI subtly embeds malicious keywords within neutral image layouts to evade detection,and breaks down harmful queries into a sequence of subtasks.This approach disperses malicious intent across modalities,exploiting VLMs’over-reliance on local visual cues and their fragility in multi-step reasoning to bypass global safety mechanisms.Implemented as an automated framework,VOTI integrates large language models as red-team assistants to generate and iteratively optimize jailbreak strategies.Extensive experiments across seven mainstream VLMs demonstrate VOTI’s effectiveness,achieving a 73.46%attack success rate on GPT-4o-mini.These results reveal critical vulnerabilities in VLMs,highlighting the urgent need for improving robust defenses and multimodal alignment.展开更多
Video action recognition(VAR)aims to analyze dynamic behaviors in videos and achieve semantic understanding.VAR faces challenges such as temporal dynamics,action-scene coupling,and the complexity of human interactions...Video action recognition(VAR)aims to analyze dynamic behaviors in videos and achieve semantic understanding.VAR faces challenges such as temporal dynamics,action-scene coupling,and the complexity of human interactions.Existing methods can be categorized into motion-level,event-level,and story-level ones based on spatiotemporal granularity.However,single-modal approaches struggle to capture complex behavioral semantics and human factors.Therefore,in recent years,vision-language models(VLMs)have been introduced into this field,providing new research perspectives for VAR.In this paper,we systematically review spatiotemporal hierarchical methods in VAR and explore how the introduction of large models has advanced the field.Additionally,we propose the concept of“Factor”to identify and integrate key information from both visual and textual modalities,enhancing multimodal alignment.We also summarize various multimodal alignment methods and provide in-depth analysis and insights into future research directions.展开更多
The application of visual-language large models in the field of medical health has gradually become a research focus.The models combine the capability for image understanding and natural language processing,and can si...The application of visual-language large models in the field of medical health has gradually become a research focus.The models combine the capability for image understanding and natural language processing,and can simultaneously process multi-modality data such as medical images and medical reports.These models can not only recognize images,but also understand the semantic relationship between images and texts,effectively realize the integration of medical information,and provide strong support for clinical decision-making and disease diagnosis.The visual-language large model has good performance for specific medical tasks,and also shows strong potential and high intelligence in the general task models.This paper provides a comprehensive review of the visual-language large model in the field of medical health.Specifically,this paper first introduces the basic theoretical basis and technical principles.Then,this paper introduces the specific application scenarios in the field of medical health,including modality fusion,semi-supervised learning,weakly supervised learning,unsupervised learning,cross-domain model and general models.Finally,the challenges including insufficient data,interpretability,and practical deployment are discussed.According to the existing challenges,four potential future development directions are given.展开更多
The advent of large vision-language models(LVLMs)represents a remarkable advance in the quest for artificial general intelligence.However,the models’effectiveness in both specialized and general tasks warrants furthe...The advent of large vision-language models(LVLMs)represents a remarkable advance in the quest for artificial general intelligence.However,the models’effectiveness in both specialized and general tasks warrants further investigation.This paper endeavors to evaluate the competency of popular LVLMs in specialized and general tasks,respectively,aiming to offer a comprehensive understanding of these novel models.To gauge their effectiveness in specialized tasks,we employ six challenging tasks in three different application scenarios:natural,healthcare,and industrial.These six tasks include salient/camouflaged/transparent object detection,as well as polyp detection,skin lesion detection,and industrial anomaly detection.We examine the performance of three recent open-source LVLMs,including MiniGPT-v2,LLaVA-1.5,and Shikra,on both visual recognition and localization in these tasks.Moreover,we conduct empirical investigations utilizing the aforementioned LVLMs together with GPT-4V,assessing their multi-modal understanding capabilities in general tasks including object counting,absurd question answering,affordance reasoning,attribute recognition,and spatial relation reasoning.Our investigations reveal that these LVLMs demonstrate limited proficiency not only in specialized tasks but also in general tasks.We delve deep into this inadequacy and uncover several potential factors,including limited cognition in specialized tasks,object hallucination,text-to-image interference,and decreased robustness in complex problems.We hope that this study can provide useful insights for the future development of LVLMs,helping researchers improve LVLMs for both general and specialized applications.展开更多
In their recent paper Pereira et al.(2025)claim that validation is overlooked in mapping and modelling of ecosystem services(ES).They state that“many studies lack critical evaluation of the results and no validation ...In their recent paper Pereira et al.(2025)claim that validation is overlooked in mapping and modelling of ecosystem services(ES).They state that“many studies lack critical evaluation of the results and no validation is provided”and that“the validation step is largely overlooked”.This assertion may have been true several years ago,for example,when Ochoa and Urbina-Cardona(2017)made a similar observation.However,there has been much work on ES model validation over the last decade.展开更多
Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interact...Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interactions to predict future items of interest.However,many current methods rely on unique user and item IDs,limiting their ability to represent users and items effectively,especially in zero-shot learning scenarios where training data is scarce.With the rapid development of Large Language Models(LLMs),researchers are exploring their potential to enhance recommendation systems.However,there is a semantic gap between the linguistic semantics of LLMs and the collaborative semantics of recommendation systems,where items are typically indexed by IDs.Moreover,most research focuses on item representations,neglecting personalized user modeling.To address these issues,we propose a sequential recommendation framework using LLMs,called CIT-Rec,a model that integrates Collaborative semantics for user representation and Image and Text information for item representation to enhance Recommendations.Specifically,by aligning intuitive image information with text containing semantic features,we can more accurately represent items,improving item representation quality.We focus not only on item representations but also on user representations.To more precisely capture users’personalized preferences,we use traditional sequential recommendation models to train on users’historical interaction data,effectively capturing behavioral patterns.Finally,by combining LLMs and traditional sequential recommendation models,we allow the LLM to understand linguistic semantics while capturing collaborative semantics.Extensive evaluations on real-world datasets show that our model outperforms baseline methods,effectively combining user interaction history with item visual and textual modalities to provide personalized recommendations.展开更多
Climate model prediction has been improved by enhancing model resolution as well as the implementation of sophisticated physical parameterization and refinement of data assimilation systems[section 6.1 in Wang et al.(...Climate model prediction has been improved by enhancing model resolution as well as the implementation of sophisticated physical parameterization and refinement of data assimilation systems[section 6.1 in Wang et al.(2025)].In relation to seasonal forecasting and climate projection in the East Asian summer monsoon season,proper simulation of the seasonal migration of rain bands by models is a challenging and limiting factor[section 7.1 in Wang et al.(2025)].展开更多
The collection and annotation of lar ge-scale bird datasets are resource-intensive and time-consuming processes that significantly limit the scalability and accuracy of biodiversity monitoring systems.While self-super...The collection and annotation of lar ge-scale bird datasets are resource-intensive and time-consuming processes that significantly limit the scalability and accuracy of biodiversity monitoring systems.While self-supervised learning(SSL)has emerged as a promising approach for leveraging unannotated data,current SSL methods face two critical challenges in bird species recognition:(1)long-tailed data distributions that result in poor performance on underrepresented species;and(2)domain shift issues caused by data augmentation strategies designed to mitigate class imbalance.Here we present SDNet,a novel SSL-based bird recognition framework that integrates diffusion models with large language models(LLMs)to overcome these limitations.SDNet employs LLMs to generate semantically rich textual descriptions for tail-class species by prompting the models with species taxonomy,morphological attributes,and habitat information,producing detailed natural language priors that capture fine-grained visual characteristics(e.g.,plumage patterns,body proportions,and distinctive markings).These textual descriptions are subsequently used by a conditional diffusion model to synthesize new bird image samples through cross-attention mechanisms that fuse textual embeddings with intermediate visual feature representations during the denoising process,ensuring generated images preserve species-specific morphological details while maintaining photorealistic quality.Additionally,we incorporate a Swin Transformer as the feature extraction backbone whose hierarchical window-based attention mechanism and shifted windowing scheme enable multi-scale local feature extraction that proves particularly effective at capturing finegrained discriminative patterns(such as beak shape and feather texture)while mitigating domain shift between synthetic and original images through consistent feature representations across both data sources.SDNet is validated on both a self-constructed dataset(Bird_BXS)an d a publicly available benchmark(Birds_25),demonstrating substantial improvements over conventional SSL approaches.Our results indicate that the synergistic integration of LLMs,diffusion models,and the Swin Transformer architecture contributes significantly to recognition accuracy,particularly for rare and morphologically similar species.These findings highlight the potential of SDNet for addressing fundamental limitations of existing SSL methods in avian recognition tasks and establishing a new paradigm for efficient self-supervised learning in large-scale ornithological vision applications.展开更多
Federated Learning(FL)enables joint training over distributed devices without data exchange but is highly vulnerable to attacks by adversaries in the form of model poisoning and malicious update injection.This work pr...Federated Learning(FL)enables joint training over distributed devices without data exchange but is highly vulnerable to attacks by adversaries in the form of model poisoning and malicious update injection.This work proposes Secured-FL,a blockchain-based defensive framework that combines smart contract-based authentication,clustering-driven outlier elimination,and dynamic threshold adjustment to defend against adversarial attacks.The framework was implemented on a private Ethereum network with a Proof-of-Authority consensus algorithm to ensure tamper-resistant and auditable model updates.Large-scale simulation on the Cyber Data dataset,under up to 50%malicious client settings,demonstrates Secured-FL achieves 6%-12%higher accuracy,9%-15%lower latency,and approximately 14%less computational expense compared to the PPSS benchmark framework.Additional tests,including confusion matrices,ROC and Precision-Recall curves,and ablation tests,confirm the interpretability and robustness of the defense.Tests for scalability also show consistent performance up to 500 clients,affirming appropriateness to reasonably large deployments.These results make Secured-FL a feasible,adversarially resilient FL paradigm with promising potential for application in smart cities,medicine,and other mission-critical IoT deployments.展开更多
This study traces the development of the Shanghai Typhoon Model(SHTM)from a traditional physics-based regional model toward a data-driven,machine-learning typhoon forecasting system.After upgrading its initial and bou...This study traces the development of the Shanghai Typhoon Model(SHTM)from a traditional physics-based regional model toward a data-driven,machine-learning typhoon forecasting system.After upgrading its initial and boundary conditions,SHTM now leverages large-scale constraints from machine-learning weather prediction(MLWP)models,resulting in an ML–physics hybrid framework.During Typhoon Danas(2025),the hybrid SHTM achieves substantially lower track errors than both the advanced ECMWF Integrated Forecasting System(IFS)and leading MLWP models such as PanGu and FuXi.Furthermore,the hybrid SHTM consistently maintains mean track errors below 200 km up to a forecast lead time of 108 hours,representing a significant advancement in forecast accuracy.In addition,this study highlights the technical roadmap for transitioning from a physics-based typhoon model to a fully data-driven ML typhoon forecast system.It also emphasizes that advances in the physical modeling framework provide a critical foundation for further improving the performance of future data-driven ML typhoon models.展开更多
The rapid advancement of machine learning based tight-binding Hamiltonian(MLTB)methods has opened new avenues for efficient and accurate electronic structure simulations,particularly in large-scale systems and long-ti...The rapid advancement of machine learning based tight-binding Hamiltonian(MLTB)methods has opened new avenues for efficient and accurate electronic structure simulations,particularly in large-scale systems and long-time scenarios.This review begins with a concise overview of traditional tight-binding(TB)models,including both(semi-)empirical and first-principles approaches,establishing the foundation for understanding MLTB developments.We then present a systematic classification of existing MLTB methodologies,grouped into two major categories:direct prediction of TB Hamiltonian elements and inference of empirical parameters.A comparative analysis with other ML-based electronic structure models is also provided,highlighting the advancement of MLTB approaches.Finally,we explore the emerging MLTB application ecosystem,highlighting how the integration of MLTB models with a diverse suite of post-processing tools from linear-scaling solvers to quantum transport frameworks and molecular dynamics interfaces is essential for tackling complex scientific problems across different domains.The continued advancement of this integrated paradigm promises to accelerate materials discovery and open new frontiers in the predictive simulation of complex quantum phenomena.展开更多
Business Process Modelling(BPM)is essential for analyzing,improving,and automating the flow of information within organizations,but traditional approaches based on manual interpretation are slow,error-prone,and requir...Business Process Modelling(BPM)is essential for analyzing,improving,and automating the flow of information within organizations,but traditional approaches based on manual interpretation are slow,error-prone,and require a high level of expertise.This article proposes an innovative alternative solution that overcomes these limitations by automatically generating comprehensive Business Process Modelling and Notation(BPMN)diagrams solely from verbal descriptions of the processes to be modeled,utilizing Large Language Models(LLMs)and multimodal Artificial Intelligence(AI).Experimental results,based on video recordings of process explanations provided by an expert from an organization(in this case,the Commercial Courts of a public justice administration),demonstrate that the proposed methodology successfully enables the automatic generation of complete and accurate BPMN diagrams,leading to significant improvements in the speed,accuracy,and accessibility of process modeling.This research makes a substantial contribution to the field of business process modeling,as its methodology is groundbreaking in its use of LLMs and multimodal AI capabilities to handle different types of source material(text and video),combining several tools to minimize the number of queries and reduce the complexity of the prompts required for the automatic generation of successful BPMN diagrams.展开更多
Pathological scarring,manifested in the form of hypertrophic scars(HTS)and keloid scars(KS),represents a major clinical challenge due to its aesthetic and functional implications for patients.Understanding the molecul...Pathological scarring,manifested in the form of hypertrophic scars(HTS)and keloid scars(KS),represents a major clinical challenge due to its aesthetic and functional implications for patients.Understanding the molecular mechanisms involved in these types of scars and developing effective treatments requires the use of controlled ex-perimental models,especially animals,to overcome the limitations of clinical studies.The aim of this sistematic review is to critically analyze the animal models used in the last five years(2020-2025)for the study of pathological scars,highlighting their advantages,limitations and applicability in the development of new therapeutic strat-egies.Murine,rabbit and porcine models,as well as alternative models,offer varied perspectives on the formation and treatment of HTS and KS,with an emphasis on histological and molecular correlations with human pathology.By synthesizing recent data,the paper highlights the essential role of preclinical research in optimizing an-tifibrotic treatments and in advancing the translation of data into the clinical sphere.Overall,animal models remain essential for bridging mechanistic insights with clinical translation,supporting the development of more effective and personalized anti-scar therapies.展开更多
Large language models(LLMs)show considerable potential to revolutionize healthcare through their performance across diverse clinical applications.Given the inherent constraints of LLMs and the critical nature of medic...Large language models(LLMs)show considerable potential to revolutionize healthcare through their performance across diverse clinical applications.Given the inherent constraints of LLMs and the critical nature of medical practice,a rigorous and systematic evaluation of their medical competence is imperative.This study presents a comprehensive review of the established methodologies and benchmarks for evaluating the medical competence of LLMs,encompassing a thorough analysis of current assessment practices across medical knowledge,clinical practice competence,and ethical-safety considerations.By integrating clinician competency assessment frameworks into LLMs evaluation,we propose a structured tri-dimensional framework that systematically organizes existing evaluation approaches according to medical theoretical knowledge,clinical practice ability,and ethical-safety considerations.Furthermore,this research provides critical insights into future developmental trajectories while establishing foundational frameworks and standardization protocols for the integration of LLMs into medical practice.展开更多
This study demonstrates a novel integration of large language models,machine learning,and multicriteria decision-making to investigate self-moderation in small online communities,a topic under-explored compared to use...This study demonstrates a novel integration of large language models,machine learning,and multicriteria decision-making to investigate self-moderation in small online communities,a topic under-explored compared to user behavior and platform-driven moderation on social media.The proposed methodological framework(1)utilizes large language models for social media post analysis and categorization,(2)employs k-means clustering for content characterization,and(3)incorporates the TODIM(Tomada de Decisão Interativa Multicritério)method to determine moderation strategies based on expert judgments.In general,the fully integrated framework leverages the strengths of these intelligent systems in a more systematic evaluation of large-scale decision problems.When applied in social media moderation,this approach promotes nuanced and context-sensitive self-moderation by taking into account factors such as cultural background and geographic location.The application of this framework is demonstrated within Facebook groups.Eight distinct content clusters encompassing safety,harassment,diversity,and misinformation are identified.Analysis revealed a preference for content removal across all clusters,suggesting a cautious approach towards potentially harmful content.However,the framework also highlights the use of other moderation actions,like account suspension,depending on the content category.These findings contribute to the growing body of research on self-moderation and offer valuable insights for creating safer and more inclusive online spaces within smaller communities.展开更多
Objectives:The five-year survival rate for pancreatic cancer is notably low,posing a significant challenge to patient health.The primary treatments are radiotherapy and chemotherapy,sometimes combined with targeted th...Objectives:The five-year survival rate for pancreatic cancer is notably low,posing a significant challenge to patient health.The primary treatments are radiotherapy and chemotherapy,sometimes combined with targeted therapy;however,their clinical benefits are limited.Therefore,developing new models to evaluate the therapeutic potential of novel molecules is essential.Fingolimod and Dimethyl Fumarate(DMF),currently used to treat multiple sclerosis,have recently been shown to have anti-cancer effects in several preclinical tumor models.This study aims to evaluate the therapeutic potential of Fingolimod and DMF in pancreatic cancer by investigating their respective in vitro cytotoxicity and in vivo antitumor effects.Methods:In this study,we evaluated for the first time these two drugs in pancreatic preclinical models in vitro using 3D spheroid tumor models and in vivo,which are compared to two standard-of-care consisting of Gemcitabine and Erlotinib.Results:In vitro,both Fingolimod and DMF induced cytotoxicity in spheroids from two pancreatic cell lines.Additionally,Fingolimod and DMF displayed anticancer effects in two subcutaneous xenograft models using PANC-1 and CFPAC-1 cells.Conclusions:Although the responses observed with Fingolimod and DMF were similar to those of Gemcitabine and Erlotinib,these findings indicate a potential emerging interest in Fingolimod and DMF for the treatment of pancreatic cancer.However,further work is still necessary to fully characterize how these drugs affect tumor progression.展开更多
Background:Assess ChatGPT and Bard's effectiveness in the initial identification of articles for Otolaryngology—Head and Neck Surgery systematic literature reviews.Methods:Three PRISMA-based systematic reviews(Ja...Background:Assess ChatGPT and Bard's effectiveness in the initial identification of articles for Otolaryngology—Head and Neck Surgery systematic literature reviews.Methods:Three PRISMA-based systematic reviews(Jabbour et al.2017,Wong et al.2018,and Wu et al.2021)were replicated using ChatGPTv3.5 and Bard.Outputs(author,title,publication year,and journal)were compared to the original references and cross-referenced with medical databases for authenticity and recall.Results:Several themes emerged when comparing Bard and ChatGPT across the three reviews.Bard generated more outputs and had greater recall in Wong et al.'s review,with a broader date range in Jabbour et al.'s review.In Wu et al.'s review,ChatGPT-2 had higher recall and identified more authentic outputs than Bard-2.Conclusion:Large language models(LLMs)failed to fully replicate peer-reviewed methodologies,producing outputs with inaccuracies but identifying relevant,especially recent,articles missed by the references.While human-led PRISMA-based reviews remain the gold standard,refining LLMs for literature reviews shows potential.展开更多
The incidence of benign airway stenosis(BAS)is on the rise,and current treatment options are associated with a significant risk of restenosis.Therefore,there is an urgent need to explore new and effective prevention a...The incidence of benign airway stenosis(BAS)is on the rise,and current treatment options are associated with a significant risk of restenosis.Therefore,there is an urgent need to explore new and effective prevention and treatment methods.Animal models serve as essential tools for investigating disease mechanisms and assessing novel therapeutic strategies,and the scientific rigor of their construction and validation significantly impacts the reliability of research findings.This paper systematically reviews the research progress and evaluation systems of BAS animal models over the past decade,aiming to provide a robust foundation for the optimized construction of BAS models,intervention studies,and clinical translation.This effort is intended to facilitate the innovation and advancement in BAS prevention and treatment strategies.展开更多
War rehearsals have become increasingly important in national security due to the growing complexity of international affairs.However,traditional rehearsal methods,such as military chess simulations,are inefficient an...War rehearsals have become increasingly important in national security due to the growing complexity of international affairs.However,traditional rehearsal methods,such as military chess simulations,are inefficient and inflexible,with particularly pronounced limitations in command and decision-making.The overwhelming volume of information and high decision complexity hinder the realization of autonomous and agile command and control.To address this challenge,an intelligent warfare simulation framework named Command-Agent is proposed,which deeply integrates large language models(LLMs)with digital twin battlefields.By constructing a highly realistic battlefield environment through real-time simulation and multi-source data fusion,the natural language interaction capabilities of LLMs are leveraged to lower the command threshold and to enable autonomous command through the Observe-Orient-Decide-Act(OODA)feedback loop.Within the Command-Agent framework,a multimodel collaborative architecture is further adopted to decouple the decision-generation and command-execution functions of LLMs.By combining specialized models such as Deep Seek-R1 and MCTool,the limitations of single-model capabilities are overcome.MCTool is a lightweight execution model fine-tuned for military Function Calling tasks.The framework also introduces a Vector Knowledge Base to mitigate hallucinations commonly exhibited by LLMs.Experimental results demonstrate that Command-Agent not only enables natural language-driven simulation and control but also deeply understands commander intent.Leveraging the multi-model collaborative architecture,during red-blue UAV confrontations involving 2 to 8 UAVs,the integrated score is improved by an average of 41.8%compared to the single-agent system(MCTool),accompanied by a 161.8%optimization in the battle loss ratio.Furthermore,when compared with multi-agent systems lacking the knowledge base,the inclusion of the Vector Knowledge Base further improves overall performance by 16.8%.In comparison with the general model(Qwen2.5-7B),the fine-tuned MCTool leads by 5%in execution efficiency.Therefore,the proposed Command-Agent introduces a novel perspective to the military command system and offers a feasible solution for intelligent battlefield decision-making.展开更多
Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Re...Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.展开更多
文摘In recent years,large vision-language models(VLMs)have achieved significant breakthroughs in cross-modal understanding and generation.However,the safety issues arising from their multimodal interactions become prominent.VLMs are vulnerable to jailbreak attacks,where attackers craft carefully designed prompts to bypass safety mechanisms,leading them to generate harmful content.To address this,we investigate the alignment between visual inputs and task execution,uncovering locality defects and attention biases in VLMs.Based on these findings,we propose VOTI,a novel jailbreak framework leveraging visual obfuscation and task induction.VOTI subtly embeds malicious keywords within neutral image layouts to evade detection,and breaks down harmful queries into a sequence of subtasks.This approach disperses malicious intent across modalities,exploiting VLMs’over-reliance on local visual cues and their fragility in multi-step reasoning to bypass global safety mechanisms.Implemented as an automated framework,VOTI integrates large language models as red-team assistants to generate and iteratively optimize jailbreak strategies.Extensive experiments across seven mainstream VLMs demonstrate VOTI’s effectiveness,achieving a 73.46%attack success rate on GPT-4o-mini.These results reveal critical vulnerabilities in VLMs,highlighting the urgent need for improving robust defenses and multimodal alignment.
基金supported by the Zhejiang Provincial Natural Science Foundation of China(No.LQ23F030001)the National Natural Science Foundation of China(No.62406280)+5 种基金the Autism Research Special Fund of Zhejiang Foundation for Disabled Persons(No.2023008)the Liaoning Province Higher Education Innovative Talents Program Support Project(No.LR2019058)the Liaoning Province Joint Open Fund for Key Scientific and Technological Innovation Bases(No.2021-KF-12-05)the Central Guidance on Local Science and Technology Development Fund of Liaoning Province(No.2023JH6/100100066)the Key Laboratory for Biomedical Engineering of Ministry of Education,Zhejiang University,Chinain part by the Open Research Fund of the State Key Laboratory of Cognitive Neuroscience and Learning.
文摘Video action recognition(VAR)aims to analyze dynamic behaviors in videos and achieve semantic understanding.VAR faces challenges such as temporal dynamics,action-scene coupling,and the complexity of human interactions.Existing methods can be categorized into motion-level,event-level,and story-level ones based on spatiotemporal granularity.However,single-modal approaches struggle to capture complex behavioral semantics and human factors.Therefore,in recent years,vision-language models(VLMs)have been introduced into this field,providing new research perspectives for VAR.In this paper,we systematically review spatiotemporal hierarchical methods in VAR and explore how the introduction of large models has advanced the field.Additionally,we propose the concept of“Factor”to identify and integrate key information from both visual and textual modalities,enhancing multimodal alignment.We also summarize various multimodal alignment methods and provide in-depth analysis and insights into future research directions.
基金The Natural Science Foundation of Hebei Province(F2024501044).
文摘The application of visual-language large models in the field of medical health has gradually become a research focus.The models combine the capability for image understanding and natural language processing,and can simultaneously process multi-modality data such as medical images and medical reports.These models can not only recognize images,but also understand the semantic relationship between images and texts,effectively realize the integration of medical information,and provide strong support for clinical decision-making and disease diagnosis.The visual-language large model has good performance for specific medical tasks,and also shows strong potential and high intelligence in the general task models.This paper provides a comprehensive review of the visual-language large model in the field of medical health.Specifically,this paper first introduces the basic theoretical basis and technical principles.Then,this paper introduces the specific application scenarios in the field of medical health,including modality fusion,semi-supervised learning,weakly supervised learning,unsupervised learning,cross-domain model and general models.Finally,the challenges including insufficient data,interpretability,and practical deployment are discussed.According to the existing challenges,four potential future development directions are given.
基金supported by the National Natural Science Foundation of China(No.62176169)the Fundamental Research Funds for the Central Universities(Nankai University,070-63243150).
文摘The advent of large vision-language models(LVLMs)represents a remarkable advance in the quest for artificial general intelligence.However,the models’effectiveness in both specialized and general tasks warrants further investigation.This paper endeavors to evaluate the competency of popular LVLMs in specialized and general tasks,respectively,aiming to offer a comprehensive understanding of these novel models.To gauge their effectiveness in specialized tasks,we employ six challenging tasks in three different application scenarios:natural,healthcare,and industrial.These six tasks include salient/camouflaged/transparent object detection,as well as polyp detection,skin lesion detection,and industrial anomaly detection.We examine the performance of three recent open-source LVLMs,including MiniGPT-v2,LLaVA-1.5,and Shikra,on both visual recognition and localization in these tasks.Moreover,we conduct empirical investigations utilizing the aforementioned LVLMs together with GPT-4V,assessing their multi-modal understanding capabilities in general tasks including object counting,absurd question answering,affordance reasoning,attribute recognition,and spatial relation reasoning.Our investigations reveal that these LVLMs demonstrate limited proficiency not only in specialized tasks but also in general tasks.We delve deep into this inadequacy and uncover several potential factors,including limited cognition in specialized tasks,object hallucination,text-to-image interference,and decreased robustness in complex problems.We hope that this study can provide useful insights for the future development of LVLMs,helping researchers improve LVLMs for both general and specialized applications.
文摘In their recent paper Pereira et al.(2025)claim that validation is overlooked in mapping and modelling of ecosystem services(ES).They state that“many studies lack critical evaluation of the results and no validation is provided”and that“the validation step is largely overlooked”.This assertion may have been true several years ago,for example,when Ochoa and Urbina-Cardona(2017)made a similar observation.However,there has been much work on ES model validation over the last decade.
基金supported by the National Key R&D Program of China[2022YFF0902703]the State Administration for Market Regulation Science and Technology Plan Project(2024MK033).
文摘Recommendation systems are key to boosting user engagement,satisfaction,and retention,particularly on media platforms where personalized content is vital.Sequential recommendation systems learn from user-item interactions to predict future items of interest.However,many current methods rely on unique user and item IDs,limiting their ability to represent users and items effectively,especially in zero-shot learning scenarios where training data is scarce.With the rapid development of Large Language Models(LLMs),researchers are exploring their potential to enhance recommendation systems.However,there is a semantic gap between the linguistic semantics of LLMs and the collaborative semantics of recommendation systems,where items are typically indexed by IDs.Moreover,most research focuses on item representations,neglecting personalized user modeling.To address these issues,we propose a sequential recommendation framework using LLMs,called CIT-Rec,a model that integrates Collaborative semantics for user representation and Image and Text information for item representation to enhance Recommendations.Specifically,by aligning intuitive image information with text containing semantic features,we can more accurately represent items,improving item representation quality.We focus not only on item representations but also on user representations.To more precisely capture users’personalized preferences,we use traditional sequential recommendation models to train on users’historical interaction data,effectively capturing behavioral patterns.Finally,by combining LLMs and traditional sequential recommendation models,we allow the LLM to understand linguistic semantics while capturing collaborative semantics.Extensive evaluations on real-world datasets show that our model outperforms baseline methods,effectively combining user interaction history with item visual and textual modalities to provide personalized recommendations.
文摘Climate model prediction has been improved by enhancing model resolution as well as the implementation of sophisticated physical parameterization and refinement of data assimilation systems[section 6.1 in Wang et al.(2025)].In relation to seasonal forecasting and climate projection in the East Asian summer monsoon season,proper simulation of the seasonal migration of rain bands by models is a challenging and limiting factor[section 7.1 in Wang et al.(2025)].
基金supported by the National Natural Science Foundation of China(32471964)。
文摘The collection and annotation of lar ge-scale bird datasets are resource-intensive and time-consuming processes that significantly limit the scalability and accuracy of biodiversity monitoring systems.While self-supervised learning(SSL)has emerged as a promising approach for leveraging unannotated data,current SSL methods face two critical challenges in bird species recognition:(1)long-tailed data distributions that result in poor performance on underrepresented species;and(2)domain shift issues caused by data augmentation strategies designed to mitigate class imbalance.Here we present SDNet,a novel SSL-based bird recognition framework that integrates diffusion models with large language models(LLMs)to overcome these limitations.SDNet employs LLMs to generate semantically rich textual descriptions for tail-class species by prompting the models with species taxonomy,morphological attributes,and habitat information,producing detailed natural language priors that capture fine-grained visual characteristics(e.g.,plumage patterns,body proportions,and distinctive markings).These textual descriptions are subsequently used by a conditional diffusion model to synthesize new bird image samples through cross-attention mechanisms that fuse textual embeddings with intermediate visual feature representations during the denoising process,ensuring generated images preserve species-specific morphological details while maintaining photorealistic quality.Additionally,we incorporate a Swin Transformer as the feature extraction backbone whose hierarchical window-based attention mechanism and shifted windowing scheme enable multi-scale local feature extraction that proves particularly effective at capturing finegrained discriminative patterns(such as beak shape and feather texture)while mitigating domain shift between synthetic and original images through consistent feature representations across both data sources.SDNet is validated on both a self-constructed dataset(Bird_BXS)an d a publicly available benchmark(Birds_25),demonstrating substantial improvements over conventional SSL approaches.Our results indicate that the synergistic integration of LLMs,diffusion models,and the Swin Transformer architecture contributes significantly to recognition accuracy,particularly for rare and morphologically similar species.These findings highlight the potential of SDNet for addressing fundamental limitations of existing SSL methods in avian recognition tasks and establishing a new paradigm for efficient self-supervised learning in large-scale ornithological vision applications.
文摘Federated Learning(FL)enables joint training over distributed devices without data exchange but is highly vulnerable to attacks by adversaries in the form of model poisoning and malicious update injection.This work proposes Secured-FL,a blockchain-based defensive framework that combines smart contract-based authentication,clustering-driven outlier elimination,and dynamic threshold adjustment to defend against adversarial attacks.The framework was implemented on a private Ethereum network with a Proof-of-Authority consensus algorithm to ensure tamper-resistant and auditable model updates.Large-scale simulation on the Cyber Data dataset,under up to 50%malicious client settings,demonstrates Secured-FL achieves 6%-12%higher accuracy,9%-15%lower latency,and approximately 14%less computational expense compared to the PPSS benchmark framework.Additional tests,including confusion matrices,ROC and Precision-Recall curves,and ablation tests,confirm the interpretability and robustness of the defense.Tests for scalability also show consistent performance up to 500 clients,affirming appropriateness to reasonably large deployments.These results make Secured-FL a feasible,adversarially resilient FL paradigm with promising potential for application in smart cities,medicine,and other mission-critical IoT deployments.
基金supported by the Special Project-Original Exploration(Grant No.42450163)the National Youth Science Foundation of China Project(Grant No.4240050560)the Research and Development of Key Technologies for Artificial Intelligence Regional Typhoon Forecasting Model project.
文摘This study traces the development of the Shanghai Typhoon Model(SHTM)from a traditional physics-based regional model toward a data-driven,machine-learning typhoon forecasting system.After upgrading its initial and boundary conditions,SHTM now leverages large-scale constraints from machine-learning weather prediction(MLWP)models,resulting in an ML–physics hybrid framework.During Typhoon Danas(2025),the hybrid SHTM achieves substantially lower track errors than both the advanced ECMWF Integrated Forecasting System(IFS)and leading MLWP models such as PanGu and FuXi.Furthermore,the hybrid SHTM consistently maintains mean track errors below 200 km up to a forecast lead time of 108 hours,representing a significant advancement in forecast accuracy.In addition,this study highlights the technical roadmap for transitioning from a physics-based typhoon model to a fully data-driven ML typhoon forecast system.It also emphasizes that advances in the physical modeling framework provide a critical foundation for further improving the performance of future data-driven ML typhoon models.
基金supported by the Advanced Materials-National Science and Technology Major Project(Grant No.2025ZD0618401)the National Natural Science Foundation of China(Grant No.12504285)+1 种基金the Natural Science Foundation of Jiangsu Province(Grant No.BK20250472)NFSG grant from BITS-Pilani,Dubai campus。
文摘The rapid advancement of machine learning based tight-binding Hamiltonian(MLTB)methods has opened new avenues for efficient and accurate electronic structure simulations,particularly in large-scale systems and long-time scenarios.This review begins with a concise overview of traditional tight-binding(TB)models,including both(semi-)empirical and first-principles approaches,establishing the foundation for understanding MLTB developments.We then present a systematic classification of existing MLTB methodologies,grouped into two major categories:direct prediction of TB Hamiltonian elements and inference of empirical parameters.A comparative analysis with other ML-based electronic structure models is also provided,highlighting the advancement of MLTB approaches.Finally,we explore the emerging MLTB application ecosystem,highlighting how the integration of MLTB models with a diverse suite of post-processing tools from linear-scaling solvers to quantum transport frameworks and molecular dynamics interfaces is essential for tackling complex scientific problems across different domains.The continued advancement of this integrated paradigm promises to accelerate materials discovery and open new frontiers in the predictive simulation of complex quantum phenomena.
基金funded by Fundación CajaCanarias and Fundación Bancaria“la Caixa”,grant number 2023DIG11.
文摘Business Process Modelling(BPM)is essential for analyzing,improving,and automating the flow of information within organizations,but traditional approaches based on manual interpretation are slow,error-prone,and require a high level of expertise.This article proposes an innovative alternative solution that overcomes these limitations by automatically generating comprehensive Business Process Modelling and Notation(BPMN)diagrams solely from verbal descriptions of the processes to be modeled,utilizing Large Language Models(LLMs)and multimodal Artificial Intelligence(AI).Experimental results,based on video recordings of process explanations provided by an expert from an organization(in this case,the Commercial Courts of a public justice administration),demonstrate that the proposed methodology successfully enables the automatic generation of complete and accurate BPMN diagrams,leading to significant improvements in the speed,accuracy,and accessibility of process modeling.This research makes a substantial contribution to the field of business process modeling,as its methodology is groundbreaking in its use of LLMs and multimodal AI capabilities to handle different types of source material(text and video),combining several tools to minimize the number of queries and reduce the complexity of the prompts required for the automatic generation of successful BPMN diagrams.
基金Ministry of Research,Innovation and Digitization,CCCDI-UEFISCDI,Grant/Award Number:PN-IV-P7-7.1-PED-2024-1578,within PNCDI Ⅳ.
文摘Pathological scarring,manifested in the form of hypertrophic scars(HTS)and keloid scars(KS),represents a major clinical challenge due to its aesthetic and functional implications for patients.Understanding the molecular mechanisms involved in these types of scars and developing effective treatments requires the use of controlled ex-perimental models,especially animals,to overcome the limitations of clinical studies.The aim of this sistematic review is to critically analyze the animal models used in the last five years(2020-2025)for the study of pathological scars,highlighting their advantages,limitations and applicability in the development of new therapeutic strat-egies.Murine,rabbit and porcine models,as well as alternative models,offer varied perspectives on the formation and treatment of HTS and KS,with an emphasis on histological and molecular correlations with human pathology.By synthesizing recent data,the paper highlights the essential role of preclinical research in optimizing an-tifibrotic treatments and in advancing the translation of data into the clinical sphere.Overall,animal models remain essential for bridging mechanistic insights with clinical translation,supporting the development of more effective and personalized anti-scar therapies.
基金Guangzhou Science and Technology Program,Grant/Award Numbers:2025B03J0110,2024A03J1074,2024A03J0927。
文摘Large language models(LLMs)show considerable potential to revolutionize healthcare through their performance across diverse clinical applications.Given the inherent constraints of LLMs and the critical nature of medical practice,a rigorous and systematic evaluation of their medical competence is imperative.This study presents a comprehensive review of the established methodologies and benchmarks for evaluating the medical competence of LLMs,encompassing a thorough analysis of current assessment practices across medical knowledge,clinical practice competence,and ethical-safety considerations.By integrating clinician competency assessment frameworks into LLMs evaluation,we propose a structured tri-dimensional framework that systematically organizes existing evaluation approaches according to medical theoretical knowledge,clinical practice ability,and ethical-safety considerations.Furthermore,this research provides critical insights into future developmental trajectories while establishing foundational frameworks and standardization protocols for the integration of LLMs into medical practice.
基金funded by the Office of the Vice-President for Research and Development of Cebu Technological University.
文摘This study demonstrates a novel integration of large language models,machine learning,and multicriteria decision-making to investigate self-moderation in small online communities,a topic under-explored compared to user behavior and platform-driven moderation on social media.The proposed methodological framework(1)utilizes large language models for social media post analysis and categorization,(2)employs k-means clustering for content characterization,and(3)incorporates the TODIM(Tomada de Decisão Interativa Multicritério)method to determine moderation strategies based on expert judgments.In general,the fully integrated framework leverages the strengths of these intelligent systems in a more systematic evaluation of large-scale decision problems.When applied in social media moderation,this approach promotes nuanced and context-sensitive self-moderation by taking into account factors such as cultural background and geographic location.The application of this framework is demonstrated within Facebook groups.Eight distinct content clusters encompassing safety,harassment,diversity,and misinformation are identified.Analysis revealed a preference for content removal across all clusters,suggesting a cautious approach towards potentially harmful content.However,the framework also highlights the use of other moderation actions,like account suspension,depending on the content category.These findings contribute to the growing body of research on self-moderation and offer valuable insights for creating safer and more inclusive online spaces within smaller communities.
基金supported by Porsolt SAS,https://www.porsolt.com/.
文摘Objectives:The five-year survival rate for pancreatic cancer is notably low,posing a significant challenge to patient health.The primary treatments are radiotherapy and chemotherapy,sometimes combined with targeted therapy;however,their clinical benefits are limited.Therefore,developing new models to evaluate the therapeutic potential of novel molecules is essential.Fingolimod and Dimethyl Fumarate(DMF),currently used to treat multiple sclerosis,have recently been shown to have anti-cancer effects in several preclinical tumor models.This study aims to evaluate the therapeutic potential of Fingolimod and DMF in pancreatic cancer by investigating their respective in vitro cytotoxicity and in vivo antitumor effects.Methods:In this study,we evaluated for the first time these two drugs in pancreatic preclinical models in vitro using 3D spheroid tumor models and in vivo,which are compared to two standard-of-care consisting of Gemcitabine and Erlotinib.Results:In vitro,both Fingolimod and DMF induced cytotoxicity in spheroids from two pancreatic cell lines.Additionally,Fingolimod and DMF displayed anticancer effects in two subcutaneous xenograft models using PANC-1 and CFPAC-1 cells.Conclusions:Although the responses observed with Fingolimod and DMF were similar to those of Gemcitabine and Erlotinib,these findings indicate a potential emerging interest in Fingolimod and DMF for the treatment of pancreatic cancer.However,further work is still necessary to fully characterize how these drugs affect tumor progression.
文摘Background:Assess ChatGPT and Bard's effectiveness in the initial identification of articles for Otolaryngology—Head and Neck Surgery systematic literature reviews.Methods:Three PRISMA-based systematic reviews(Jabbour et al.2017,Wong et al.2018,and Wu et al.2021)were replicated using ChatGPTv3.5 and Bard.Outputs(author,title,publication year,and journal)were compared to the original references and cross-referenced with medical databases for authenticity and recall.Results:Several themes emerged when comparing Bard and ChatGPT across the three reviews.Bard generated more outputs and had greater recall in Wong et al.'s review,with a broader date range in Jabbour et al.'s review.In Wu et al.'s review,ChatGPT-2 had higher recall and identified more authentic outputs than Bard-2.Conclusion:Large language models(LLMs)failed to fully replicate peer-reviewed methodologies,producing outputs with inaccuracies but identifying relevant,especially recent,articles missed by the references.While human-led PRISMA-based reviews remain the gold standard,refining LLMs for literature reviews shows potential.
基金National Natural Science Foundation of China,Grant/Award Number:82000102 and 82270112。
文摘The incidence of benign airway stenosis(BAS)is on the rise,and current treatment options are associated with a significant risk of restenosis.Therefore,there is an urgent need to explore new and effective prevention and treatment methods.Animal models serve as essential tools for investigating disease mechanisms and assessing novel therapeutic strategies,and the scientific rigor of their construction and validation significantly impacts the reliability of research findings.This paper systematically reviews the research progress and evaluation systems of BAS animal models over the past decade,aiming to provide a robust foundation for the optimized construction of BAS models,intervention studies,and clinical translation.This effort is intended to facilitate the innovation and advancement in BAS prevention and treatment strategies.
文摘War rehearsals have become increasingly important in national security due to the growing complexity of international affairs.However,traditional rehearsal methods,such as military chess simulations,are inefficient and inflexible,with particularly pronounced limitations in command and decision-making.The overwhelming volume of information and high decision complexity hinder the realization of autonomous and agile command and control.To address this challenge,an intelligent warfare simulation framework named Command-Agent is proposed,which deeply integrates large language models(LLMs)with digital twin battlefields.By constructing a highly realistic battlefield environment through real-time simulation and multi-source data fusion,the natural language interaction capabilities of LLMs are leveraged to lower the command threshold and to enable autonomous command through the Observe-Orient-Decide-Act(OODA)feedback loop.Within the Command-Agent framework,a multimodel collaborative architecture is further adopted to decouple the decision-generation and command-execution functions of LLMs.By combining specialized models such as Deep Seek-R1 and MCTool,the limitations of single-model capabilities are overcome.MCTool is a lightweight execution model fine-tuned for military Function Calling tasks.The framework also introduces a Vector Knowledge Base to mitigate hallucinations commonly exhibited by LLMs.Experimental results demonstrate that Command-Agent not only enables natural language-driven simulation and control but also deeply understands commander intent.Leveraging the multi-model collaborative architecture,during red-blue UAV confrontations involving 2 to 8 UAVs,the integrated score is improved by an average of 41.8%compared to the single-agent system(MCTool),accompanied by a 161.8%optimization in the battle loss ratio.Furthermore,when compared with multi-agent systems lacking the knowledge base,the inclusion of the Vector Knowledge Base further improves overall performance by 16.8%.In comparison with the general model(Qwen2.5-7B),the fine-tuned MCTool leads by 5%in execution efficiency.Therefore,the proposed Command-Agent introduces a novel perspective to the military command system and offers a feasible solution for intelligent battlefield decision-making.
文摘Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.
