Robust cooperative unmanned aerial vehicle(UAV)formation in complex 3D environments is hampered by reward sparsity and inefficient collaboration.To address this,we propose context-aware relational agent learning(CORAL...Robust cooperative unmanned aerial vehicle(UAV)formation in complex 3D environments is hampered by reward sparsity and inefficient collaboration.To address this,we propose context-aware relational agent learning(CORAL),a novel multi-agent deep reinforcement learning framework.CORAL synergistically integrates two modules:(1)a novelty-based intrinsic reward module to drive efficient exploration and(2)an explicit relational learning module that allows agents to predict peer intentions and enhance coordination.Built on a multi-agent Actor-Critic architecture,CORAL enables agents to balance self-interest with group objectives.Comprehensive evaluations in a high-fidelity simulation show that our method significantly outperforms state-of-theart baselines like multi-agent deep deterministic policy gradient(MADDPG)and monotonic value function factorisation for deep multi-agent reinforcement learning(QMIX)in path planning efficiency,collision avoidance,and scalability.展开更多
This study introduces a novel approach to addressing the challenges of high-dimensional variables and strong nonlinearity in reservoir production and layer configuration optimization.For the first time,relational mach...This study introduces a novel approach to addressing the challenges of high-dimensional variables and strong nonlinearity in reservoir production and layer configuration optimization.For the first time,relational machine learning models are applied in reservoir development optimization.Traditional regression-based models often struggle in complex scenarios,but the proposed relational and regression-based composite differential evolution(RRCODE)method combines a Gaussian naive Bayes relational model with a radial basis function network regression model.This integration effectively captures complex relationships in the optimization process,improving both accuracy and convergence speed.Experimental tests on a multi-layer multi-channel reservoir model,the Egg reservoir model,and a real-field reservoir model(the S reservoir)demonstrate that RRCODE significantly reduces water injection and production volumes while increasing economic returns and cumulative oil recovery.Moreover,the surrogate models employed in RRCODE exhibit lightweight characteristics with low computational overhead.These results highlight RRCODE's superior performance in the integrated optimization of reservoir production and layer configurations,offering more efficient and economically viable solutions for oilfield development.展开更多
Statistical relational learning constructs statistical models from relational databases, combining relational learning and statistical learning. Its strong ability and special property make statistical relational lear...Statistical relational learning constructs statistical models from relational databases, combining relational learning and statistical learning. Its strong ability and special property make statistical relational learning become one of the important areas in machine learning research.In this paper,the general concepts and characters of statistical relational learning are presented firstly.Then some major branches of this newly emerging field are discussed,including logic and rule-based approaches,frame and object-oriented approaches,functional programming-based approaches.After that several methods of applying rough set in statistical relational learning are described,such as gRS-ILP and VPRSILP. Finally some applications of statistical relational leaning are briefly introduced and some future directions of statistical relational learning and the application of rough set in this area are pointed out.展开更多
Imaging through scattering media faces a critical challenge:deep-learning-based methods inherently suppress high-frequency speckle information,limiting the recovery of fine textures and edges.To overcome this spectral...Imaging through scattering media faces a critical challenge:deep-learning-based methods inherently suppress high-frequency speckle information,limiting the recovery of fine textures and edges.To overcome this spectral bias,we introduce the concept of the relative speckle frequency domain(RsFD),which redefines high-frequency features as learnable,adaptive components via frequency-domain decomposition.We demonstrate that independently processing generalized high-frequency speckle components enables neural networks to capture latent target details previously obscured in conventional approaches.Leveraging this principle,we design FDUnet,a dualbranch network comprising a low-frequency sub-network(Lnet)for global structure reconstruction and a relative high-frequency sub-network(RHnet)dedicated to enhancing textures and edges.Experiments confirm FDUnet's superiority:it outperforms state-of-the-art methods in both visual fidelity and quantitative metrics by +5.9% to 8.7% in SSIM and+5.4 to 7.9 dB in PSNR across diverse datasets(MNIST,Fashion-MNIST,FERET).These enhancements translate into notable improvements in the preservation of textures and edges,especially exhibiting exceptional robustness to multimode fiber perturbations.This work bridges the gap between physical priors and neural network learning,unlocking new potentials for high-fidelity applications,such as biomedical endoscopic imaging,in dynamic scattering environments.展开更多
Education is a complex system that has evolved over thousands of years to reach its current level.It has many objects and subjects.The education systems of the countries are very diverse.Almost every country has its o...Education is a complex system that has evolved over thousands of years to reach its current level.It has many objects and subjects.The education systems of the countries are very diverse.Almost every country has its own ranking approach,because there is no universally accepted scientific theory of education.The search for effective reform in education continues today,but any reform that is not based on scientific theory cannot solve the problem.There are many problems in the content and management of education.Knowledge assessment is also flawed.No country can build an ideal school.It can be considered that in the last hundred years,education has not developed conceptually in the desired direction.Thus,education aims to train strong personalities,not perfect(wise)people.Although individualistic education may seem beneficial locally,globally it divides humanity and prevents its sustainable and harmonious living.However,in societies made up of perfect people,in principle there will be no division,harmony will exist,because perfect people solve problems not by force,but by reason,prefer cooperation rather than conflict.This means protecting the planet.To make the world a gun-free society,the view of education must change conceptually.This article presents a new philosophical view of teaching knowledge and proposes a new model,criteria,and theory.展开更多
The current infrared image pedestrian detectors have problems with high rates of false positives and false negatives. To solve these problems, we proposed an improved anchor-free fully convolutional one-stage object d...The current infrared image pedestrian detectors have problems with high rates of false positives and false negatives. To solve these problems, we proposed an improved anchor-free fully convolutional one-stage object detection(FCOS) algorithm. Firstly, we introduced the channel attention module squeeze excitation(SE)-Block in the FCOS backbone network, which was used to learn how to model the relative importance between different feature channels, and to achieve the weight recalibration of the features extracted from the convolution neural network, and improve the weight values that are more important for pedestrian target detection. Secondly, soft non-maximum suppression(Soft-NMS) replaced the conventional NMS within the algorithm's post-processing phase, which was used to reduce the probability of missed detection for occluded pedestrians. The experimental results show that our improved FCOS algorithm improves the average precision(AP) by 6.71% on the original dataset and 7.97% on the augmented KAIST pedestrian dataset compared with the original FCOS algorithm. Our improvements effectively meet the real-time requirements and there is no significant decrease in speed compared with the original FCOS algorithm, and decreased the false positives and false negatives for infrared image pedestrian detection.展开更多
The National Institute of Standards and Technology(NIST)has identified natural language policies as the preferred expression of policy and implicitly called for an automated translation of ABAC natural language access...The National Institute of Standards and Technology(NIST)has identified natural language policies as the preferred expression of policy and implicitly called for an automated translation of ABAC natural language access control policy(NLACP)to a machine-readable form.To study the automation process,we consider the hierarchical ABAC model as our reference model since it better reflects the requirements of real-world organizations.Therefore,this paper focuses on the questions of:how can we automatically infer the hierarchical structure of an ABAC model given NLACPs;and,how can we extract and define the set of authorization attributes based on the resulting structure.To address these questions,we propose an approach built upon recent advancements in natural language processing and machine learning techniques.For such a solution,the lack of appropriate data often poses a bottleneck.Therefore,we decouple the primary contributions of this work into:(1)developing a practical framework to extract authorization attributes of hierarchical ABAC system from natural language artifacts,and(2)generating a set of realistic synthetic natural language access control policies(NLACPs)to evaluate the proposed framework.Our experimental results are promising as we achieved-in average-an F1-score of 0.96 when extracting attributes values of subjects,and 0.91 when extracting the values of objects’attributes from natural language access control policies.展开更多
The National Institute of Standards and Technology(NIST)has identified natural language policies as the preferred expression of policy and implicitly called for an automated translation of ABAC natural language access...The National Institute of Standards and Technology(NIST)has identified natural language policies as the preferred expression of policy and implicitly called for an automated translation of ABAC natural language access control policy(NLACP)to a machine-readable form.To study the automation process,we consider the hierarchical ABAC model as our reference model since it better reflects the requirements of real-world organizations.Therefore,this paper focuses on the questions of:how can we automatically infer the hierarchical structure of an ABAC model given NLACPs;and,how can we extract and define the set of authorization attributes based on the resulting structure.To address these questions,we propose an approach built upon recent advancements in natural language processing and machine learning techniques.For such a solution,the lack of appropriate data often poses a bottleneck.Therefore,we decouple the primary contributions of this work into:(1)developing a practical framework to extract authorization attributes of hierarchical ABAC system from natural language artifacts,and(2)generating a set of realistic synthetic natural language access control policies(NLACPs)to evaluate the proposed framework.Our experimental results are promising as we achieved-in average-an F1-score of 0.96 when extracting attributes values of subjects,and 0.91 when extracting the values of objects’attributes from natural language access control policies.展开更多
基金supported by the STI 2030 Major Projects(No.2022ZD0208804)the National Natural Science Foundation of China(No.62473017)。
文摘Robust cooperative unmanned aerial vehicle(UAV)formation in complex 3D environments is hampered by reward sparsity and inefficient collaboration.To address this,we propose context-aware relational agent learning(CORAL),a novel multi-agent deep reinforcement learning framework.CORAL synergistically integrates two modules:(1)a novelty-based intrinsic reward module to drive efficient exploration and(2)an explicit relational learning module that allows agents to predict peer intentions and enhance coordination.Built on a multi-agent Actor-Critic architecture,CORAL enables agents to balance self-interest with group objectives.Comprehensive evaluations in a high-fidelity simulation show that our method significantly outperforms state-of-theart baselines like multi-agent deep deterministic policy gradient(MADDPG)and monotonic value function factorisation for deep multi-agent reinforcement learning(QMIX)in path planning efficiency,collision avoidance,and scalability.
基金supported by the National Natural Science Foundation of China under Grant 52325402,52274057,and 52074340the National Key R&D Program of China under Grant 2023YFB4104200+2 种基金the Major Scientific and Technological Projects of CNOOC under Grant CCL2022RCPS0397RSN111 Project under Grant B08028China Scholarship Council under Grant 202306450108.
文摘This study introduces a novel approach to addressing the challenges of high-dimensional variables and strong nonlinearity in reservoir production and layer configuration optimization.For the first time,relational machine learning models are applied in reservoir development optimization.Traditional regression-based models often struggle in complex scenarios,but the proposed relational and regression-based composite differential evolution(RRCODE)method combines a Gaussian naive Bayes relational model with a radial basis function network regression model.This integration effectively captures complex relationships in the optimization process,improving both accuracy and convergence speed.Experimental tests on a multi-layer multi-channel reservoir model,the Egg reservoir model,and a real-field reservoir model(the S reservoir)demonstrate that RRCODE significantly reduces water injection and production volumes while increasing economic returns and cumulative oil recovery.Moreover,the surrogate models employed in RRCODE exhibit lightweight characteristics with low computational overhead.These results highlight RRCODE's superior performance in the integrated optimization of reservoir production and layer configurations,offering more efficient and economically viable solutions for oilfield development.
文摘Statistical relational learning constructs statistical models from relational databases, combining relational learning and statistical learning. Its strong ability and special property make statistical relational learning become one of the important areas in machine learning research.In this paper,the general concepts and characters of statistical relational learning are presented firstly.Then some major branches of this newly emerging field are discussed,including logic and rule-based approaches,frame and object-oriented approaches,functional programming-based approaches.After that several methods of applying rough set in statistical relational learning are described,such as gRS-ILP and VPRSILP. Finally some applications of statistical relational leaning are briefly introduced and some future directions of statistical relational learning and the application of rough set in this area are pointed out.
基金National Natural Science Foundation of China(62362037)Fundamental Research Funds for the Central Universities(30919011401,30920010001)+3 种基金Natural Science Foundation of Jiangxi Province(20224ACB202011)Jiangsu Province Key Research and Development Project(BE2023817)Hong Kong Research Grant Council(15217721,15125724,C7074-21GF)Hong Kong Polytechnic University(P0045680,P0043485,P0045762,P0049101)。
文摘Imaging through scattering media faces a critical challenge:deep-learning-based methods inherently suppress high-frequency speckle information,limiting the recovery of fine textures and edges.To overcome this spectral bias,we introduce the concept of the relative speckle frequency domain(RsFD),which redefines high-frequency features as learnable,adaptive components via frequency-domain decomposition.We demonstrate that independently processing generalized high-frequency speckle components enables neural networks to capture latent target details previously obscured in conventional approaches.Leveraging this principle,we design FDUnet,a dualbranch network comprising a low-frequency sub-network(Lnet)for global structure reconstruction and a relative high-frequency sub-network(RHnet)dedicated to enhancing textures and edges.Experiments confirm FDUnet's superiority:it outperforms state-of-the-art methods in both visual fidelity and quantitative metrics by +5.9% to 8.7% in SSIM and+5.4 to 7.9 dB in PSNR across diverse datasets(MNIST,Fashion-MNIST,FERET).These enhancements translate into notable improvements in the preservation of textures and edges,especially exhibiting exceptional robustness to multimode fiber perturbations.This work bridges the gap between physical priors and neural network learning,unlocking new potentials for high-fidelity applications,such as biomedical endoscopic imaging,in dynamic scattering environments.
文摘Education is a complex system that has evolved over thousands of years to reach its current level.It has many objects and subjects.The education systems of the countries are very diverse.Almost every country has its own ranking approach,because there is no universally accepted scientific theory of education.The search for effective reform in education continues today,but any reform that is not based on scientific theory cannot solve the problem.There are many problems in the content and management of education.Knowledge assessment is also flawed.No country can build an ideal school.It can be considered that in the last hundred years,education has not developed conceptually in the desired direction.Thus,education aims to train strong personalities,not perfect(wise)people.Although individualistic education may seem beneficial locally,globally it divides humanity and prevents its sustainable and harmonious living.However,in societies made up of perfect people,in principle there will be no division,harmony will exist,because perfect people solve problems not by force,but by reason,prefer cooperation rather than conflict.This means protecting the planet.To make the world a gun-free society,the view of education must change conceptually.This article presents a new philosophical view of teaching knowledge and proposes a new model,criteria,and theory.
基金supported by the Natural Science Fund of Heilongjiang Province(No.PL2024F027)the National Natural Science Foundation of China(No.61601174)。
文摘The current infrared image pedestrian detectors have problems with high rates of false positives and false negatives. To solve these problems, we proposed an improved anchor-free fully convolutional one-stage object detection(FCOS) algorithm. Firstly, we introduced the channel attention module squeeze excitation(SE)-Block in the FCOS backbone network, which was used to learn how to model the relative importance between different feature channels, and to achieve the weight recalibration of the features extracted from the convolution neural network, and improve the weight values that are more important for pedestrian target detection. Secondly, soft non-maximum suppression(Soft-NMS) replaced the conventional NMS within the algorithm's post-processing phase, which was used to reduce the probability of missed detection for occluded pedestrians. The experimental results show that our improved FCOS algorithm improves the average precision(AP) by 6.71% on the original dataset and 7.97% on the augmented KAIST pedestrian dataset compared with the original FCOS algorithm. Our improvements effectively meet the real-time requirements and there is no significant decrease in speed compared with the original FCOS algorithm, and decreased the false positives and false negatives for infrared image pedestrian detection.
文摘The National Institute of Standards and Technology(NIST)has identified natural language policies as the preferred expression of policy and implicitly called for an automated translation of ABAC natural language access control policy(NLACP)to a machine-readable form.To study the automation process,we consider the hierarchical ABAC model as our reference model since it better reflects the requirements of real-world organizations.Therefore,this paper focuses on the questions of:how can we automatically infer the hierarchical structure of an ABAC model given NLACPs;and,how can we extract and define the set of authorization attributes based on the resulting structure.To address these questions,we propose an approach built upon recent advancements in natural language processing and machine learning techniques.For such a solution,the lack of appropriate data often poses a bottleneck.Therefore,we decouple the primary contributions of this work into:(1)developing a practical framework to extract authorization attributes of hierarchical ABAC system from natural language artifacts,and(2)generating a set of realistic synthetic natural language access control policies(NLACPs)to evaluate the proposed framework.Our experimental results are promising as we achieved-in average-an F1-score of 0.96 when extracting attributes values of subjects,and 0.91 when extracting the values of objects’attributes from natural language access control policies.
基金supported by Princess Nourah bint Abdulrahman University,Riyadh,Saudi Arabia.
文摘The National Institute of Standards and Technology(NIST)has identified natural language policies as the preferred expression of policy and implicitly called for an automated translation of ABAC natural language access control policy(NLACP)to a machine-readable form.To study the automation process,we consider the hierarchical ABAC model as our reference model since it better reflects the requirements of real-world organizations.Therefore,this paper focuses on the questions of:how can we automatically infer the hierarchical structure of an ABAC model given NLACPs;and,how can we extract and define the set of authorization attributes based on the resulting structure.To address these questions,we propose an approach built upon recent advancements in natural language processing and machine learning techniques.For such a solution,the lack of appropriate data often poses a bottleneck.Therefore,we decouple the primary contributions of this work into:(1)developing a practical framework to extract authorization attributes of hierarchical ABAC system from natural language artifacts,and(2)generating a set of realistic synthetic natural language access control policies(NLACPs)to evaluate the proposed framework.Our experimental results are promising as we achieved-in average-an F1-score of 0.96 when extracting attributes values of subjects,and 0.91 when extracting the values of objects’attributes from natural language access control policies.
