The turbulence mechanism plays an important part in the mixing process and momentum transfer of turbulence. A three-dimensional Prandtl mixing length tidal model has been developed to simulate tidal flows and water qu...The turbulence mechanism plays an important part in the mixing process and momentum transfer of turbulence. A three-dimensional Prandtl mixing length tidal model has been developed to simulate tidal flows and water quality. The eddy viscosities and diffusivities are computed from the Prandtl mixing length model. In order to model the water quality of an estuary or coastal area many interdependent processes need to be simulated. These may be conveniently separated into three main groups: transport and mixing processes, biochemical interaction of water quality variables and the utilization and re-cycling of nutrients by living matter. The model simulates full oxygen and nutrient balance, primary productivity and the transport, reaction mechanism and fate of pollutants over tidal time-scales. The model is applied to numerical simulation of tidal flows and water quality in Dalian Bay. The model has been calibrated against a limited data set of historical water quality observations and in general demonstrates excellent agreement with all available data.展开更多
To study the rock deformation with three- dimensional model under rolling forces of disc cutter, by car- rying out the circular-grooving test with disc cutter rolling around on the rock, the rock mechanical behavior u...To study the rock deformation with three- dimensional model under rolling forces of disc cutter, by car- rying out the circular-grooving test with disc cutter rolling around on the rock, the rock mechanical behavior under rolling disc cutter is studied, the mechanical model of disc cutter rolling around the groove is established, and the the- ory of single-point and double-angle variables is proposed. Based on this theory, the physics equations and geometric equations of rock mechanical behavior under disc cutters of tunnel boring machine （TBM） are studied, and then the bal- ance equations of interactive forces between disc cutter and rock are established. Accordingly, formulas about normal force, rolling force and side force of a disc cutter are de- rived, and their validity is studied by tests. Therefore, a new method and theory is proposed to study rock- breaking mech- anism of disc cutters.展开更多
Silicon nitride(Si_(3)N_(4))based ceramics are one of the most attractive advanced engineering materials,which have been widely used under high-speed rotational operation or for mechanical contacts across a curved sur...Silicon nitride(Si_(3)N_(4))based ceramics are one of the most attractive advanced engineering materials,which have been widely used under high-speed rotational operation or for mechanical contacts across a curved surface.In the present study,rotationally symmetric texturing of Si_(3)N_(4),with radial grain align-ment,was obtained by centripetal sinter-forging(CSF)of a partially sintered sample.The average values of the included angles between the c-axis of the local Si_(3)N_(4)grain and radial direction were approxi-mately 16.4°and 11.0°,on the section plane perpendicular to the pressing direction,and parallel to both the pressing and radial directions,respectively.The compressive strain in the pressing direction forced the ceramic body to flow towards the central axis,resulting in compressive strain in the tangential di-rection and tensile strain in the radial direction.A fundamental physical model was created to simulate the grain rotation during the 3-dimentional strain reorientation,which revealed the rod-like grain would preferentially rotate toward the center of the sample under the CSF process.In addition,due to the fric-tion between the sample surface and the pressing punch,the increased shear strain could enhance the Si_(3)N_(4)grain alignment.Consequently,ceramics with rod-like grains perpendicular to the curved side sur-face could be anticipated by applying the centripetal forming concept in a controlled manner.展开更多
Large language models(LLMs)represent significant advancements in artificial intelligence.However,their increasing capabilities come with a serious challenge:misalignment,which refers to the deviation of model behavior...Large language models(LLMs)represent significant advancements in artificial intelligence.However,their increasing capabilities come with a serious challenge:misalignment,which refers to the deviation of model behavior from the designers’intentions and human values.This review aims to synthesize the current understanding of the LLM misalignment issue and provide researchers and practitioners with a comprehensive overview.We define the concept of misalignment and elaborate on its various manifestations,including generating harmful content,factual errors(hallucinations),propagating biases,failing to follow instructions,emerging deceptive behaviors,and emergent misalignment.We explore the multifaceted causes of misalignment,systematically analyzing factors from surface-level technical issues(e.g.,training data,objective function design,model scaling)to deeper fundamental challenges(e.g.,difficulties formalizing values,discrepancies between training signals and real intentions).This review covers existing and emerging techniques for detecting and evaluating the degree of misalignment,such as benchmark tests,red-teaming,and formal safety assessments.Subsequently,we examine strategies to mitigate misalignment,focusing on mainstream alignment techniques such as RLHF,Constitutional AI(CAI),instruction fine-tuning,and novel approaches that address scalability and robustness.In particular,we analyze recent advances in misalignment attack research,including system prompt modifications,supervised fine-tuning,self-supervised representation attacks,and model editing,which challenge the robustness of model alignment.We categorize and analyze the surveyed literature,highlighting major findings,persistent limitations,and current contentious points.Finally,we identify key open questions and propose several promising future research directions,including constructing high-quality alignment datasets,exploring novel alignment methods,coordinating diverse values,and delving into the deep philosophical aspects of alignment.This work underscores the complexity and multidimensionality of LLM misalignment issues,calling for interdisciplinary approaches to reliably align LLMs with human values.展开更多
This paper deals with the problem of recreating horizontal alignments of existing railway lines.The main objective is to propose a simple method for automatically obtaining optimized recreated alignments located as cl...This paper deals with the problem of recreating horizontal alignments of existing railway lines.The main objective is to propose a simple method for automatically obtaining optimized recreated alignments located as close as possible to an existing one.Based on a previously defined geometric model,two different constrained optimization problems are formulated.The first problem uses only the information provided by a set of points representing the track centerline while the second one also considers additional data about the existing alignment.The proposed methodology consists of a two-stage process in which both problems are solved consecutively using numerical techniques.The main results obtained applying this methodology are presented to show its performance and to prove its practical usefulness:an academic example used to compare with other methods,and a case study of a railway section located in Parga(Spain)in which the geometry of its horizontal alignment is successfully recovered.展开更多
Multimodal Aspect-Based Sentiment Analysis(MABSA)aims to detect sentiment polarity toward specific aspects by leveraging both textual and visual inputs.However,existing models suffer from weak aspectimage alignment,mo...Multimodal Aspect-Based Sentiment Analysis(MABSA)aims to detect sentiment polarity toward specific aspects by leveraging both textual and visual inputs.However,existing models suffer from weak aspectimage alignment,modality imbalance dominated by textual signals,and limited reasoning for implicit or ambiguous sentiments requiring external knowledge.To address these issues,we propose a unified framework named Gated-Linear Aspect-Aware Multimodal Sentiment Network(GLAMSNet).First of all,an input encoding module is employed to construct modality-specific and aspect-aware representations.Subsequently,we introduce an image–aspect correlation matching module to provide hierarchical supervision for visual-textual alignment.Building upon these components,we further design a Gated-Linear Aspect-Aware Fusion(GLAF)module to enhance aspect-aware representation learning by adaptively filtering irrelevant textual information and refining semantic alignment under aspect guidance.Additionally,an External Language Model Knowledge-Guided mechanism is integrated to incorporate sentimentaware prior knowledge from GPT-4o,enabling robust semantic reasoning especially under noisy or ambiguous inputs.Experimental studies conducted based on Twitter-15 and Twitter-17 datasets demonstrate that the proposed model outperforms most state-of-the-art methods,achieving 79.36%accuracy and 74.72%F1-score,and 74.31%accuracy and 72.01%F1-score,respectively.展开更多
Euler angle error model, rotation vector error model （RVE） and quaternion error model （QE） were qualitatively and quantitatively compared and an in-flight alignment filter algorithm was designed. This algorithm us...Euler angle error model, rotation vector error model （RVE） and quaternion error model （QE） were qualitatively and quantitatively compared and an in-flight alignment filter algorithm was designed. This algorithm used extended Kalman filter （EKF） based on RVE and QE separately avoi- ding the accuracy problem of the Euler angle model and used Rauch-Tung-Striebel（RTS） smoothing method to refine the accuracy recuperating the coning error for simplified RVE. Simulation results show that RVE and QE are more adapt for nonlinear filter estimation than the Euler angle model. The filter algorithm designed has more advantages in convergence speed, accuracy and stability comparing with the algorithm based on the three models separately.展开更多
In this letter, we briefly describe a program of self adapting hidden Markov model (SA HMM) and its application in multiple sequences alignment. Program consists of two stage optimisation algorithm.
A novel model based on structure alignments is proposed for statistical machine translation in this paper. Meta-structure and sequence of meta-structure for a parse tree are defined. During the translation process, a ...A novel model based on structure alignments is proposed for statistical machine translation in this paper. Meta-structure and sequence of meta-structure for a parse tree are defined. During the translation process, a parse tree is decomposed to deal with the structure divergence and the alignments can be constructed at different levels of recombination of meta-structure （RM）. This method can perform the structure mapping across the sub-tree structure between languages. As a result, we get not only the translation for the target language, but sequence of meta-stmctu .re of its parse tree at the same time. Experiments show that the model in the framework of log-linear model has better generative ability and significantly outperforms Pharaoh, a phrase-based system.展开更多
Aligning natural language with operating system(OS)commands allows users to performcomplex computer tasks through simple natural language descriptions.However,due to the complex nature of natural language,it still rem...Aligning natural language with operating system(OS)commands allows users to performcomplex computer tasks through simple natural language descriptions.However,due to the complex nature of natural language,it still remains challenging to achieve precise alignment.In this paper,we present ComAlign,a Chinese benchmark dataset that pairs Chinese natural language descriptions with corresponding OS commands.ComAlign covers a broad range of 82 distinct OS command types with a total of 1811 natural language descriptions.We elaborate on the construction of ComAlign and construct three baselines to evaluate the alignment accuracy on ComAlign.Experimental results show that even advanced large language models struggle with certain ambiguously phrased OS commands.Specifically,the best performing baseline achieves 46.9%alignment accuracy.We demonstrate that ComAlign is collected from realworld application scenarios,making it particularly suitable for developing and benchmarking intelligent OS and agent systems that support user-machine interactions through natural language.展开更多
The nearly analytic discrete(NAD)method is a kind of finite difference method with advantages of high accuracy and stability.Previous studies have investigated the NAD method for simulating wave propagation in the tim...The nearly analytic discrete(NAD)method is a kind of finite difference method with advantages of high accuracy and stability.Previous studies have investigated the NAD method for simulating wave propagation in the time-domain.This study applies the NAD method to solving three-dimensional(3D)acoustic wave equations in the frequency-domain.This forward modeling approach is then used as the“engine”for implementing 3D frequency-domain full waveform inversion(FWI).In the numerical modeling experiments,synthetic examples are first given to show the superiority of the NAD method in forward modeling compared with traditional finite difference methods.Synthetic 3D frequency-domain FWI experiments are then carried out to examine the effectiveness of the proposed methods.The inversion results show that the NAD method is more suitable than traditional methods,in terms of computational cost and stability,for 3D frequency-domain FWI,and represents an effective approach for inversion of subsurface model structures.展开更多
Machine Unlearning(MU)has emerged as a promising technique for aligning large language models(LLMs)with safety requirements to steer them forgetting specific harmful contents.Despite the significant progress in previo...Machine Unlearning(MU)has emerged as a promising technique for aligning large language models(LLMs)with safety requirements to steer them forgetting specific harmful contents.Despite the significant progress in previous studies,we argue that the current evaluation criteria,which solely focus on safety evaluation,are actually impractical and biased,leading to concerns about the true effectiveness of MU techniques.To address this,we propose to comprehensively evaluate LLMs after MU from three aspects:safety,over-safety,and general utility.Specifically,a novel benchmark MUBENCH with 18 related datasets is first constructed,where the safety is measured with both vanilla harmful inputs and 10 types of jailbreak attacks.Furthermore,we examine whether MU introduces side effects,focusing on over-safety and utility-loss.Extensive experiments are performed on 3 popular LLMs with 7 recent MU methods.The results highlight a challenging trilemma in safety alignment without side effects,indicating that there is still considerable room for further exploration.MUBENCH serves as a comprehensive benchmark,fostering future research on MU for safety alignment of LLMs.展开更多
Measuring the business-IT alignment(BITA)of an organization determines its alignment level,provides directions for further improvements,and consequently promotes the organizational performances.Due to the capabilities...Measuring the business-IT alignment(BITA)of an organization determines its alignment level,provides directions for further improvements,and consequently promotes the organizational performances.Due to the capabilities of enterprise architecture(EA)in interrelating different business/IT viewpoints and elements,the development of EA is superior to support BITA measurement.Extant BITA measurement literature is sparse when it concerns EA.The literature tends to explain how EA viewpoints or models correlate with BITA,without discussing where to collect and integrate EA data.To address this gap,this paper attempts to propose a specific BITA measurement process through associating a BITA maturity model with a famous EA framework:DoD Architectural Framework 2.0(DoDAF2.0).The BITA metrics in the maturity model are connected to the meta-models and models of DoDAF2.0.An illustrative ArchiSurance case is conducted to explain the measurement process.Systematically,this paper explores the process of BITA measurement from the viewpoint of EA,which helps to collect the measurement data in an organized way and analyzes the BITA level in the phase of architecture development.展开更多
The fundamental step to get a Statistical Shape Model (SSM) is to align all the training samples to the same spatial modality. In this paper, we propose a new 3D alignment method for organic training samples matching,...The fundamental step to get a Statistical Shape Model (SSM) is to align all the training samples to the same spatial modality. In this paper, we propose a new 3D alignment method for organic training samples matching, whose modalities are orientable and surface figures could be recognized. It is a feature based alignment method which matches two models depending on the distribution of surface curvature. According to the affine transformation on 2D Gaussian map, the distances between the corresponding parts on surface could be minimized. We applied our proposed method on 5 cases left lung training samples alignment and 4 cases liver training samples alignment. The experiment results were performed on the left lung training samples and the liver training samples. The availability of proposed method was confirmed.展开更多
The continuation task (Wang & Wang, 2014) proves to have language learning potentials by many empirical studies (Peng, 2018). This study intends to explore its underlining alignment mechanisms through two continua...The continuation task (Wang & Wang, 2014) proves to have language learning potentials by many empirical studies (Peng, 2018). This study intends to explore its underlining alignment mechanisms through two continuation tasks on English materi-als called Chon and Charles by two Chinese linguistics-major graduates (EFL learners). The results show that alignment is ubiqui-tous both linguistically and thematically, and the reading and writing are tightly coupled. This indicates that the continuation task will significantly benefit second language writing pedagogy if applied appropriately.展开更多
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
Entity alignment(EA)is crucial for knowledge fusion and integration,as it aims to match equivalent entities across different KGs.Recently,many neural-based EA methods have been proposed,focusing on developing various ...Entity alignment(EA)is crucial for knowledge fusion and integration,as it aims to match equivalent entities across different KGs.Recently,many neural-based EA methods have been proposed,focusing on developing various graph representation learning models to match entities in vector spaces.However,most real-world KGs are large-scale and contain rich structural and attribute information about entities,presenting challenges for current approaches designed primarily for small-and medium-sized KGs.To address the challenges of large-scale EA,this paper introduces a simple,effective,and scalable method based on language models.Our approach first leverages the capabilities of language models to encode entities'multi-view information into low-dimensional embeddings,identifying potential aligned entity pairs with high similarity.These candidates are then re-ranked using a global matching algorithm to produce the final alignments.Experimental results show that our method achieves state-of-the-art performance on real-world large-scale EA datasets,with superior accuracy and efficiency compared to existing methods.展开更多
Submerged dikes were frequently built in curved estuarine channels to improve navigational conditions. A three-dimensional numerical model of tidal motions was developed to simulate flows through the submerged dike al...Submerged dikes were frequently built in curved estuarine channels to improve navigational conditions. A three-dimensional numerical model of tidal motions was developed to simulate flows through the submerged dike alignment in curved estuarine reach of complex bathymetry on the basis of satisfactory calibrations. Through the analysis of the morphological characteristics of the curved channel of the Huangpu River mouth in the Yangtze River Delta, a submerged dike alignment project was examined. The effect of the navigation channel improvement project and its influence on the flow regimes, the navigational conditions, the flood control and the operating conditions of harbours within the adjacent water area were evaluated, and the technical feasibility of the navigation channel improvement scheme was demonstrated.展开更多
基金The project is supported by The National Natural Science Foundation of China
文摘The turbulence mechanism plays an important part in the mixing process and momentum transfer of turbulence. A three-dimensional Prandtl mixing length tidal model has been developed to simulate tidal flows and water quality. The eddy viscosities and diffusivities are computed from the Prandtl mixing length model. In order to model the water quality of an estuary or coastal area many interdependent processes need to be simulated. These may be conveniently separated into three main groups: transport and mixing processes, biochemical interaction of water quality variables and the utilization and re-cycling of nutrients by living matter. The model simulates full oxygen and nutrient balance, primary productivity and the transport, reaction mechanism and fate of pollutants over tidal time-scales. The model is applied to numerical simulation of tidal flows and water quality in Dalian Bay. The model has been calibrated against a limited data set of historical water quality observations and in general demonstrates excellent agreement with all available data.
基金supported by the National Natural Science Foundation of China (51075147)
文摘To study the rock deformation with three- dimensional model under rolling forces of disc cutter, by car- rying out the circular-grooving test with disc cutter rolling around on the rock, the rock mechanical behavior under rolling disc cutter is studied, the mechanical model of disc cutter rolling around the groove is established, and the the- ory of single-point and double-angle variables is proposed. Based on this theory, the physics equations and geometric equations of rock mechanical behavior under disc cutters of tunnel boring machine （TBM） are studied, and then the bal- ance equations of interactive forces between disc cutter and rock are established. Accordingly, formulas about normal force, rolling force and side force of a disc cutter are de- rived, and their validity is studied by tests. Therefore, a new method and theory is proposed to study rock- breaking mech- anism of disc cutters.
基金supported by the Shandong Provincial Key Research and Development Program(No.2019JZZY010330)National Natural Science Foundation of China(Nos.52172066,52172064,52072077 and51832002)JWRI International Joint Research Collaborators(No.JIJReC)program.
文摘Silicon nitride(Si_(3)N_(4))based ceramics are one of the most attractive advanced engineering materials,which have been widely used under high-speed rotational operation or for mechanical contacts across a curved surface.In the present study,rotationally symmetric texturing of Si_(3)N_(4),with radial grain align-ment,was obtained by centripetal sinter-forging(CSF)of a partially sintered sample.The average values of the included angles between the c-axis of the local Si_(3)N_(4)grain and radial direction were approxi-mately 16.4°and 11.0°,on the section plane perpendicular to the pressing direction,and parallel to both the pressing and radial directions,respectively.The compressive strain in the pressing direction forced the ceramic body to flow towards the central axis,resulting in compressive strain in the tangential di-rection and tensile strain in the radial direction.A fundamental physical model was created to simulate the grain rotation during the 3-dimentional strain reorientation,which revealed the rod-like grain would preferentially rotate toward the center of the sample under the CSF process.In addition,due to the fric-tion between the sample surface and the pressing punch,the increased shear strain could enhance the Si_(3)N_(4)grain alignment.Consequently,ceramics with rod-like grains perpendicular to the curved side sur-face could be anticipated by applying the centripetal forming concept in a controlled manner.
基金supported by National Natural Science Foundation of China(62462019,62172350)Guangdong Basic and Applied Basic Research Foundation(2023A1515012846)+6 种基金Guangxi Science and Technology Major Program(AA24263010)The Key Research and Development Program of Guangxi(AB24010085)Key Laboratory of Equipment Data Security and Guarantee Technology,Ministry of Education(GDZB2024060500)2024 Higher Education Scientific Research Planning Project(No.24NL0419)Nantong Science and Technology Project(No.JC2023070)the Open Fund of Advanced Cryptography and System Security Key Laboratory of Sichuan Province(GrantNo.SKLACSS-202407)sponsored by the Cultivation of Young andMiddle-aged Academic Leaders in the“Qing Lan Project”of Jiangsu Province and the 2025 Outstanding Teaching Team in the“Qing Lan Project”of Jiangsu Province.
文摘Large language models(LLMs)represent significant advancements in artificial intelligence.However,their increasing capabilities come with a serious challenge:misalignment,which refers to the deviation of model behavior from the designers’intentions and human values.This review aims to synthesize the current understanding of the LLM misalignment issue and provide researchers and practitioners with a comprehensive overview.We define the concept of misalignment and elaborate on its various manifestations,including generating harmful content,factual errors(hallucinations),propagating biases,failing to follow instructions,emerging deceptive behaviors,and emergent misalignment.We explore the multifaceted causes of misalignment,systematically analyzing factors from surface-level technical issues(e.g.,training data,objective function design,model scaling)to deeper fundamental challenges(e.g.,difficulties formalizing values,discrepancies between training signals and real intentions).This review covers existing and emerging techniques for detecting and evaluating the degree of misalignment,such as benchmark tests,red-teaming,and formal safety assessments.Subsequently,we examine strategies to mitigate misalignment,focusing on mainstream alignment techniques such as RLHF,Constitutional AI(CAI),instruction fine-tuning,and novel approaches that address scalability and robustness.In particular,we analyze recent advances in misalignment attack research,including system prompt modifications,supervised fine-tuning,self-supervised representation attacks,and model editing,which challenge the robustness of model alignment.We categorize and analyze the surveyed literature,highlighting major findings,persistent limitations,and current contentious points.Finally,we identify key open questions and propose several promising future research directions,including constructing high-quality alignment datasets,exploring novel alignment methods,coordinating diverse values,and delving into the deep philosophical aspects of alignment.This work underscores the complexity and multidimensionality of LLM misalignment issues,calling for interdisciplinary approaches to reliably align LLMs with human values.
基金founded by project TED2021129324B-I00 of the Ministerio de Ciencia e Innovación(Spain)and NextGenerationEU(European Union)the Collaboration Agreement between Consellería de Educación,Formación Profesional e Universidades(Xunta de Galicia,Spain)and Universidade de Santiago de Compostela(Spain)which regulates the Specialization Campus Campus Terra under Grant number 2022-PU014support given by Xunta de Galicia(Spain)by means of the research projects 2023 GPC GI-2084 ED431B2023/17 and GRC GI-1563-ED431C 2021/15,respectively.
文摘This paper deals with the problem of recreating horizontal alignments of existing railway lines.The main objective is to propose a simple method for automatically obtaining optimized recreated alignments located as close as possible to an existing one.Based on a previously defined geometric model,two different constrained optimization problems are formulated.The first problem uses only the information provided by a set of points representing the track centerline while the second one also considers additional data about the existing alignment.The proposed methodology consists of a two-stage process in which both problems are solved consecutively using numerical techniques.The main results obtained applying this methodology are presented to show its performance and to prove its practical usefulness:an academic example used to compare with other methods,and a case study of a railway section located in Parga(Spain)in which the geometry of its horizontal alignment is successfully recovered.
基金supported in part by the National Nature Science Foundation of China under Grants 62476216 and 62273272in part by the Key Research and Development Program of Shaanxi Province under Grant 2024GX-YBXM-146+1 种基金in part by the Scientific Research Program Funded by Education Department of Shaanxi Provincial Government under Grant 23JP091the Youth Innovation Team of Shaanxi Universities.
文摘Multimodal Aspect-Based Sentiment Analysis(MABSA)aims to detect sentiment polarity toward specific aspects by leveraging both textual and visual inputs.However,existing models suffer from weak aspectimage alignment,modality imbalance dominated by textual signals,and limited reasoning for implicit or ambiguous sentiments requiring external knowledge.To address these issues,we propose a unified framework named Gated-Linear Aspect-Aware Multimodal Sentiment Network(GLAMSNet).First of all,an input encoding module is employed to construct modality-specific and aspect-aware representations.Subsequently,we introduce an image–aspect correlation matching module to provide hierarchical supervision for visual-textual alignment.Building upon these components,we further design a Gated-Linear Aspect-Aware Fusion(GLAF)module to enhance aspect-aware representation learning by adaptively filtering irrelevant textual information and refining semantic alignment under aspect guidance.Additionally,an External Language Model Knowledge-Guided mechanism is integrated to incorporate sentimentaware prior knowledge from GPT-4o,enabling robust semantic reasoning especially under noisy or ambiguous inputs.Experimental studies conducted based on Twitter-15 and Twitter-17 datasets demonstrate that the proposed model outperforms most state-of-the-art methods,achieving 79.36%accuracy and 74.72%F1-score,and 74.31%accuracy and 72.01%F1-score,respectively.
文摘Euler angle error model, rotation vector error model （RVE） and quaternion error model （QE） were qualitatively and quantitatively compared and an in-flight alignment filter algorithm was designed. This algorithm used extended Kalman filter （EKF） based on RVE and QE separately avoi- ding the accuracy problem of the Euler angle model and used Rauch-Tung-Striebel（RTS） smoothing method to refine the accuracy recuperating the coning error for simplified RVE. Simulation results show that RVE and QE are more adapt for nonlinear filter estimation than the Euler angle model. The filter algorithm designed has more advantages in convergence speed, accuracy and stability comparing with the algorithm based on the three models separately.
文摘In this letter, we briefly describe a program of self adapting hidden Markov model (SA HMM) and its application in multiple sequences alignment. Program consists of two stage optimisation algorithm.
基金the National High Technology Research and Development Progran of China(No.200606010108.2006AA01Z150)
文摘A novel model based on structure alignments is proposed for statistical machine translation in this paper. Meta-structure and sequence of meta-structure for a parse tree are defined. During the translation process, a parse tree is decomposed to deal with the structure divergence and the alignments can be constructed at different levels of recombination of meta-structure （RM）. This method can perform the structure mapping across the sub-tree structure between languages. As a result, we get not only the translation for the target language, but sequence of meta-stmctu .re of its parse tree at the same time. Experiments show that the model in the framework of log-linear model has better generative ability and significantly outperforms Pharaoh, a phrase-based system.
基金supported by the National Key Research and Development Program under Grant 2024YFB4506200the Science and Technology Innovation Program of Hunan Province under Grant 2024RC1048the National Key Laboratory Foundation Project under Grant 2024-KJWPDL-14.
文摘Aligning natural language with operating system(OS)commands allows users to performcomplex computer tasks through simple natural language descriptions.However,due to the complex nature of natural language,it still remains challenging to achieve precise alignment.In this paper,we present ComAlign,a Chinese benchmark dataset that pairs Chinese natural language descriptions with corresponding OS commands.ComAlign covers a broad range of 82 distinct OS command types with a total of 1811 natural language descriptions.We elaborate on the construction of ComAlign and construct three baselines to evaluate the alignment accuracy on ComAlign.Experimental results show that even advanced large language models struggle with certain ambiguously phrased OS commands.Specifically,the best performing baseline achieves 46.9%alignment accuracy.We demonstrate that ComAlign is collected from realworld application scenarios,making it particularly suitable for developing and benchmarking intelligent OS and agent systems that support user-machine interactions through natural language.
基金supported by the Joint Fund of Seismological Science(Grant No.U1839206)the National R&D Program on Monitoring,Early Warning and Prevention of Major Natural Disaster(Grant No.2017YFC1500301)+2 种基金supported by IGGCAS Research Start-up Funds(Grant No.E0515402)National Natural Science Foundation of China(Grant No.E1115401)supported by National Natural Science Foundation of China(Grant No.11971258).
文摘The nearly analytic discrete(NAD)method is a kind of finite difference method with advantages of high accuracy and stability.Previous studies have investigated the NAD method for simulating wave propagation in the time-domain.This study applies the NAD method to solving three-dimensional(3D)acoustic wave equations in the frequency-domain.This forward modeling approach is then used as the“engine”for implementing 3D frequency-domain full waveform inversion(FWI).In the numerical modeling experiments,synthetic examples are first given to show the superiority of the NAD method in forward modeling compared with traditional finite difference methods.Synthetic 3D frequency-domain FWI experiments are then carried out to examine the effectiveness of the proposed methods.The inversion results show that the NAD method is more suitable than traditional methods,in terms of computational cost and stability,for 3D frequency-domain FWI,and represents an effective approach for inversion of subsurface model structures.
基金supported by the National Natural Science Foundation of China(Grant No.62176078)the Fundamental Research Funds for the Central Universities(2022FRFK060002).
文摘Machine Unlearning(MU)has emerged as a promising technique for aligning large language models(LLMs)with safety requirements to steer them forgetting specific harmful contents.Despite the significant progress in previous studies,we argue that the current evaluation criteria,which solely focus on safety evaluation,are actually impractical and biased,leading to concerns about the true effectiveness of MU techniques.To address this,we propose to comprehensively evaluate LLMs after MU from three aspects:safety,over-safety,and general utility.Specifically,a novel benchmark MUBENCH with 18 related datasets is first constructed,where the safety is measured with both vanilla harmful inputs and 10 types of jailbreak attacks.Furthermore,we examine whether MU introduces side effects,focusing on over-safety and utility-loss.Extensive experiments are performed on 3 popular LLMs with 7 recent MU methods.The results highlight a challenging trilemma in safety alignment without side effects,indicating that there is still considerable room for further exploration.MUBENCH serves as a comprehensive benchmark,fostering future research on MU for safety alignment of LLMs.
基金supported by the National Natural Science Foundation of China(71571189)the State Key Laboratory of Air Traffic Management System and Technology(SKLATM201806)
文摘Measuring the business-IT alignment(BITA)of an organization determines its alignment level,provides directions for further improvements,and consequently promotes the organizational performances.Due to the capabilities of enterprise architecture(EA)in interrelating different business/IT viewpoints and elements,the development of EA is superior to support BITA measurement.Extant BITA measurement literature is sparse when it concerns EA.The literature tends to explain how EA viewpoints or models correlate with BITA,without discussing where to collect and integrate EA data.To address this gap,this paper attempts to propose a specific BITA measurement process through associating a BITA maturity model with a famous EA framework:DoD Architectural Framework 2.0(DoDAF2.0).The BITA metrics in the maturity model are connected to the meta-models and models of DoDAF2.0.An illustrative ArchiSurance case is conducted to explain the measurement process.Systematically,this paper explores the process of BITA measurement from the viewpoint of EA,which helps to collect the measurement data in an organized way and analyzes the BITA level in the phase of architecture development.
文摘The fundamental step to get a Statistical Shape Model (SSM) is to align all the training samples to the same spatial modality. In this paper, we propose a new 3D alignment method for organic training samples matching, whose modalities are orientable and surface figures could be recognized. It is a feature based alignment method which matches two models depending on the distribution of surface curvature. According to the affine transformation on 2D Gaussian map, the distances between the corresponding parts on surface could be minimized. We applied our proposed method on 5 cases left lung training samples alignment and 4 cases liver training samples alignment. The experiment results were performed on the left lung training samples and the liver training samples. The availability of proposed method was confirmed.
文摘The continuation task (Wang & Wang, 2014) proves to have language learning potentials by many empirical studies (Peng, 2018). This study intends to explore its underlining alignment mechanisms through two continuation tasks on English materi-als called Chon and Charles by two Chinese linguistics-major graduates (EFL learners). The results show that alignment is ubiqui-tous both linguistically and thematically, and the reading and writing are tightly coupled. This indicates that the continuation task will significantly benefit second language writing pedagogy if applied appropriately.
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supported by the National Natural Science Foundation of China(No.62276026)。
文摘Entity alignment(EA)is crucial for knowledge fusion and integration,as it aims to match equivalent entities across different KGs.Recently,many neural-based EA methods have been proposed,focusing on developing various graph representation learning models to match entities in vector spaces.However,most real-world KGs are large-scale and contain rich structural and attribute information about entities,presenting challenges for current approaches designed primarily for small-and medium-sized KGs.To address the challenges of large-scale EA,this paper introduces a simple,effective,and scalable method based on language models.Our approach first leverages the capabilities of language models to encode entities'multi-view information into low-dimensional embeddings,identifying potential aligned entity pairs with high similarity.These candidates are then re-ranked using a global matching algorithm to produce the final alignments.Experimental results show that our method achieves state-of-the-art performance on real-world large-scale EA datasets,with superior accuracy and efficiency compared to existing methods.
文摘Submerged dikes were frequently built in curved estuarine channels to improve navigational conditions. A three-dimensional numerical model of tidal motions was developed to simulate flows through the submerged dike alignment in curved estuarine reach of complex bathymetry on the basis of satisfactory calibrations. Through the analysis of the morphological characteristics of the curved channel of the Huangpu River mouth in the Yangtze River Delta, a submerged dike alignment project was examined. The effect of the navigation channel improvement project and its influence on the flow regimes, the navigational conditions, the flood control and the operating conditions of harbours within the adjacent water area were evaluated, and the technical feasibility of the navigation channel improvement scheme was demonstrated.
