The complex interactions and conflicting performance demands in multi-component composites pose significant challenges for achieving balanced multi-property optimization through conventional trial-and-error approaches...The complex interactions and conflicting performance demands in multi-component composites pose significant challenges for achieving balanced multi-property optimization through conventional trial-and-error approaches.Machine learning(ML)offers a promising solution,markedly improving materials discovery efficiency.However,the high dimensionality of feature spaces in such systems has long impeded effective ML-driven feature representation and inverse design.To overcome this,we present an Intelligent Screening System(ISS)framework to accelerate the discovery of optimal formulations balancing four key properties in 15-component PTFE-based copper-clad laminate composites(PTFE-CCLCs).ISS adopts modular descriptors based on the physical information of component volume fractions,thereby simplifying the feature representation.By leveraging the inverse prediction capability of ML models and constructing a performance-driven virtual candidate database,ISS significantly reduced the computational complexity associated with high-dimensional spaces.Experimental validation confirmed that ISSoptimized formulations exhibited superior synergy,notably resolving the trade-off between thermal conductivity and peel strength,and outperform many commercial counterparts.Despite limited data and inherent process variability,ISS achieved an average prediction accuracy of 76.5%,with thermal conductivity predictions exceeding 90%,demonstrating robust reliability.This work provides an innovative,efficient strategy for multifunctional optimization and accelerated discovery in ultra-complex composite systems,highlighting the integration of ML and advanced materials design.展开更多
To counteract small sample size,severe class imbalance and high feature redundancy in 90-day mRS prediction after stroke,this study proposes a four-stage pipeline-“ADASYN re-sampling→clinical+statistical feature scr...To counteract small sample size,severe class imbalance and high feature redundancy in 90-day mRS prediction after stroke,this study proposes a four-stage pipeline-“ADASYN re-sampling→clinical+statistical feature screening→dimensionality reduction→5-fold cross-validation”-and benchmark composite deep-learning architectures.ADASYN first balances the minority classes in the original feature space.Next,a tri-level filter(clinical domain knowledge,variance threshold,mutual information)removes clinically meaningless or redundant variables,after which PCA compresses the remaining features while preserving critical neurological signatures(e.g.,brain-herniation history).Four hybrid CNN-RNN models are trained and compared under strict 5-fold cross-validation;the optimal ensemble yields stable,clinically interpretable probabilities that can support individualized rehabilitation planning.展开更多
Humanoid robots hold significant promise for social interaction and emotional companionship.However,their effectiveness hinges on the ability to convey nuanced and authentic emotions.Here,we presented a universal huma...Humanoid robots hold significant promise for social interaction and emotional companionship.However,their effectiveness hinges on the ability to convey nuanced and authentic emotions.Here,we presented a universal humanoid robot head with a facial kinematics model.Using a reinforcement learning framework guided by symmetry assessment,emotion decoupling,and MLLM authenticity evaluation,our system autonomously learns to generate adaptive facial expressions through dynamic landmark adjustments.By transferring the simulation training results to real-world environments,the robot can perform natural and expressive expressions.Another novel feature is the independent regulation of emotion intensity and expression magnitude across emotional categories,which enhances the ability to achieve culturally adaptive and socially resonant robotic expressions significantly.This research advances adaptive humanoid interaction,offering an easier and more efficient pathway toward culturally resonant and psychologically plausible robotic expressions.展开更多
Predicting the behavior of renewable energy systems requires models capable of generating accurate forecasts from limited historical data,a challenge that becomes especially pronounced when commissioning new facil-iti...Predicting the behavior of renewable energy systems requires models capable of generating accurate forecasts from limited historical data,a challenge that becomes especially pronounced when commissioning new facil-ities where operational records are scarce.This review aims to synthesize recent progress in data-efficient deep learning approaches for addressing such“cold-start”forecasting problems.It primarily covers three interrelated domains—solar photovoltaic(PV),wind power,and electrical load forecasting—where data scarcity and operational variability are most critical,while also including representative studies on hydropower and carbon emission prediction to provide a broader systems perspective.To this end,we examined trends from over 150 predominantly peer-reviewed studies published between 2019 and mid-2025,highlighting advances in zero-shot and few-shot meta-learning frameworks that enable rapid model adaptation with minimal labeled data.Moreover,transfer learning approaches combined with spatiotemporal graph neural networks have been employed to transfer knowledge from existing energy assets to new,data-sparse environments,effectively capturing hidden dependencies among geographic features,meteorological dynamics,and grid structures.Synthetic data generation has further proven valuable for expanding training samples and mitigating overfitting in cold-start scenarios.In addition,large language models and explainable artificial intelligence(XAI)—notably conversational XAI systems—have been used to interpret and communicate complex model behaviors in accessible terms,fostering operator trust from the earliest deployment stages.By consolidating methodological advances,unresolved challenges,and open-source resources,this review provides a coherent overview of deep learning strategies that can shorten the data-sparse ramp-up period of new energy infrastructures and accelerate the transition toward resilient,low-carbon electricity grids.展开更多
Computational solid mechanics has become an indispensable approach in engineering,and numerical investigation of fracturing in composites is essential,as composites are widely used in structural applications.Crack evo...Computational solid mechanics has become an indispensable approach in engineering,and numerical investigation of fracturing in composites is essential,as composites are widely used in structural applications.Crack evolution in composites is the path to elucidating the relationship between microstructures and fracture performance,but crack-based finite-element methods are computationally expensive and time-consuming,which limits their application in computation-intensive scenarios.Consequently,this study proposes a deep learning framework called Crack-Net for instant prediction of the dynamic crack growth process,as well as its strain-stress curve.Specifically,Crack-Net introduces an implicit constraint technique,which incorporates the relationship between crack evolution and stress response into the network architecture.This technique substantially reduces data requirements while improving predictive accuracy.The transfer learning technique enables Crack-Net to handle composite materials with reinforcements of different strengths.Trained on high-accuracy fracture development datasets from phase field simulations,the proposed framework is capable of tackling intricate scenarios,involving materials with diverse interfaces,varying initial conditions,and the intricate elastoplastic fracture process.The proposed Crack-Net holds great promise for practical applications in engineering and materials science,in which accurate and efficient fracture prediction is crucial for optimizing material performance and microstructural design.展开更多
Ensuring the consistent mechanical performance of three-dimensional(3D)-printed continuous fiber-reinforced composites is a significant challenge in additive manufacturing.The current reliance on manual monitoring exa...Ensuring the consistent mechanical performance of three-dimensional(3D)-printed continuous fiber-reinforced composites is a significant challenge in additive manufacturing.The current reliance on manual monitoring exacerbates this challenge by rendering the process vulnerable to environmental changes and unexpected factors,resulting in defects and inconsistent product quality,particularly in unmanned long-term operations or printing in extreme environments.To address these issues,we developed a process monitoring and closed-loop feedback control strategy for the 3D printing process.Real-time printing image data were captured and analyzed using a well-trained neural network model,and a real-time control module-enabled closed-loop feedback control of the flow rate was developed.The neural network model,which was based on image processing and artificial intelligence,enabled the recognition of flow rate values with an accuracy of 94.70%.The experimental results showed significant improvements in both the surface performance and mechanical properties of printed composites,with three to six times improvement in tensile strength and elastic modulus,demonstrating the effectiveness of the strategy.This study provides a generalized process monitoring and feedback control method for the 3D printing of continuous fiber-reinforced composites,and offers a potential solution for remote online monitoring and closed-loop adjustment in unmanned or extreme space environments.展开更多
Gas hydrate(GH)is an unconventional resource estimated at 1000-120,000 trillion m^(3)worldwide.Research on GH is ongoing to determine its geological and flow characteristics for commercial produc-tion.After two large-...Gas hydrate(GH)is an unconventional resource estimated at 1000-120,000 trillion m^(3)worldwide.Research on GH is ongoing to determine its geological and flow characteristics for commercial produc-tion.After two large-scale drilling expeditions to study the GH-bearing zone in the Ulleung Basin,the mineral composition of 488 sediment samples was analyzed using X-ray diffraction(XRD).Because the analysis is costly and dependent on experts,a machine learning model was developed to predict the mineral composition using XRD intensity profiles as input data.However,the model’s performance was limited because of improper preprocessing of the intensity profile.Because preprocessing was applied to each feature,the intensity trend was not preserved even though this factor is the most important when analyzing mineral composition.In this study,the profile was preprocessed for each sample using min-max scaling because relative intensity is critical for mineral analysis.For 49 test data among the 488 data,the convolutional neural network(CNN)model improved the average absolute error and coefficient of determination by 41%and 46%,respectively,than those of CNN model with feature-based pre-processing.This study confirms that combining preprocessing for each sample with CNN is the most efficient approach for analyzing XRD data.The developed model can be used for the compositional analysis of sediment samples from the Ulleung Basin and the Korea Plateau.In addition,the overall procedure can be applied to any XRD data of sediments worldwide.展开更多
With the availability of high-performance computing technology and the development of advanced numerical simulation methods, Computational Fluid Dynamics (CFD) is becoming more and more practical and efficient in engi...With the availability of high-performance computing technology and the development of advanced numerical simulation methods, Computational Fluid Dynamics (CFD) is becoming more and more practical and efficient in engineering. As one of the high-precision representative algorithms, the high-order Discontinuous Galerkin Method (DGM) has not only attracted widespread attention from scholars in the CFD research community, but also received strong development. However, when DGM is extended to high-speed aerodynamic flow field calculations, non-physical numerical Gibbs oscillations near shock waves often significantly affect the numerical accuracy and even cause calculation failure. Data driven approaches based on machine learning techniques can be used to learn the characteristics of Gibbs noise, which motivates us to use it in high-speed DG applications. To achieve this goal, labeled data need to be generated in order to train the machine learning models. This paper proposes a new method for denoising modeling of Gibbs phenomenon using a machine learning technique, the zero-shot learning strategy, to eliminate acquiring large amounts of CFD data. The model adopts a graph convolutional network combined with graph attention mechanism to learn the denoising paradigm from synthetic Gibbs noise data and generalize to DGM numerical simulation data. Numerical simulation results show that the Gibbs denoising model proposed in this paper can suppress the numerical oscillation near shock waves in the high-order DGM. Our work automates the extension of DGM to high-speed aerodynamic flow field calculations with higher generalization and lower cost.展开更多
The ability of mobile robots to plan and execute a path is foundational to various path-planning challenges,particularly Coverage Path Planning.While this task has been typically tackled with classical algorithms,thes...The ability of mobile robots to plan and execute a path is foundational to various path-planning challenges,particularly Coverage Path Planning.While this task has been typically tackled with classical algorithms,these often struggle with flexibility and adaptability in unknown environments.On the other hand,recent advances in Reinforcement Learning offer promising approaches,yet a significant gap in the literature remains when it comes to generalization over a large number of parameters.This paper presents a unified,generalized framework for coverage path planning that leverages value-based deep reinforcement learning techniques.The novelty of the framework comes from the design of an observation space that accommodates different map sizes,an action masking scheme that guarantees safety and robustness while also serving as a learning-fromdemonstration technique during training,and a unique reward function that yields value functions that are size-invariant.These are coupled with a curriculum learning-based training strategy and parametric environment randomization,enabling the agent to tackle complete or partial coverage path planning with perfect or incomplete knowledge while generalizing to different map sizes,configurations,sensor payloads,and sub-tasks.Our empirical results show that the algorithm can perform zero-shot learning scenarios at a near-optimal level in environments that follow a similar distribution as during training,outperforming a greedy heuristic by sixfold.Furthermore,in out-of-distribution environments,our method surpasses existing state-of-the-art algorithms in most zero-shot and all few-shot scenarios,paving the way for generalizable and adaptable path-planning algorithms.展开更多
Carbon nanotube-reinforced cement composites have gained significant attention due to their enhanced mechanical properties,particularly in compressive and flexural strength.Despite extensive research,the influence of ...Carbon nanotube-reinforced cement composites have gained significant attention due to their enhanced mechanical properties,particularly in compressive and flexural strength.Despite extensive research,the influence of various parameters on these properties remains inadequately understood,primarily due to the complex interactions within the composites.This study addresses this gap by employingmachine learning techniques to conduct a sensitivity analysis on the compressive and flexural strength of carbon nanotube-reinforced cement composites.It systematically evaluates nine data-preprocessing techniques and benchmarks eleven machine-learning algorithms to reveal tradeoffs between predictive accuracy and computational complexity,which has not previously been explored in carbon nanotube-reinforced cement composite research.In this regard,four main factors are considered in the sensitivity analysis,which are the machine learning model type,the data pre-processing technique,and the effect of the concrete constituent materials on the compressive and flexural strength both globally through feature importance assessment and locally through partial dependence analysis.Accordingly,this research optimizes ninety-nine models representing combinations of eleven machine learning algorithms and nine data preprocessing techniques to accurately predict the mechanical properties of carbon nanotube-reinforced cement composites.Moreover,the study aims to unravel the relationships between different parameters and their impact on the composite’s strength by utilizing feature importance and partial dependence analyses.This research is crucial as it provides a comprehensive understanding of the factors influencing the performance of carbon nanotube-reinforced cement composites,which is vital for their efficient design and application in construction.The use of machine learning in this context not only enhances predictive accuracy but also offers insights that are often challenging to obtain through traditional experimental methods.The findings contribute to the field by highlighting the potential of advanced data-driven approaches in optimizing and understanding advanced composite materials,paving the way for more durable and resilient construction materials.展开更多
BACKGROUND Esophageal cancer carries a poor prognosis with low 5-year survival rates and limited early detection options.The skeletal muscle index at the L3 vertebral level is a well-established prognostic marker in e...BACKGROUND Esophageal cancer carries a poor prognosis with low 5-year survival rates and limited early detection options.The skeletal muscle index at the L3 vertebral level is a well-established prognostic marker in esophageal cancer,but most follow-up computed tomography(CT)scans do not extend to L3 and limiting its utility.Radiomics has emerged as a powerful tool for extracting prognostic information from medical images.AIM To evaluate the influential features for esophageal cancer prognosis by integrating radiomic and body compositionbased indices of skeletal muscle and adipose tissue at the T12 level from both pretreatment and follow-up CT images,in order to assess their value in predicting overall survival(OS).METHODS This retrospective study included 212 esophageal cancer patients who underwent concurrent chemoradiotherapy,with both pretreatment and follow-up chest CT scans available.Body organ analysis(BOA)and radiomic features were extracted from skeletal muscle and adipose tissue at the T12 level using automated tools.Four feature subsets(no-radiomics,pretreatment only,follow-up only,and combined inputs)were developed using logistic regression(LR)with least absolute shrinkage and selection operator for feature selection,followed by Cox regression.Prognostic models-including nomogram,support vector classifier,LR,and extra trees classifier-were constructed to predict 1-,2-,and 3-year OS.RESULTS The model integrating both BOA and radiomics from pretreatment and follow-up CT,combined with clinical data,achieved the best performance for 2-year OS prediction,with an area under the time-dependent receiver operating characteristic curve of 0.91,sensitivity of 0.81,and specificity of 0.88 using the LR model.The most predictive features included both clinical variables,body composition indices,and radiomic features,particularly from follow-up VAT.Follow-up imaging contributed significantly to model performance,reinforcing its value in treatment response evaluation.CONCLUSION This is the first study to demonstrate that BOA indices and their corresponding radiomics at the T12-level from both pretreatment and follow-up CT scans-combined with clinical data-can provide accurate prognostic information for esophageal cancer.This approach offers a practical alternative when L3-level imaging is unavailable and supports the clinical integration of automated T12-based imaging biomarkers.The integration of these imaging features with clinical parameters enhances the prediction of survival outcomes and contributes to non-invasive,personalized treatment planning.展开更多
In thermal protection structures,controlling and optimizing the surface roughness of carbon/phenolic(C/Ph)composites can effectively improve thermal protection performance and ensure the safe operation of carriers in ...In thermal protection structures,controlling and optimizing the surface roughness of carbon/phenolic(C/Ph)composites can effectively improve thermal protection performance and ensure the safe operation of carriers in high-temperature environments.This paper introduces a machine learning(ML)framework to forecast the surface roughness of carbon-phenolic composites under various thermal conditions by employing an ML algorithm derived from historical experimental datasets.Firstly,ablation experiments and collection of surface roughness height data of C/Ph composites under different thermal environments were conducted in an electric arc wind tunnel.Then,an ML model based on Ridge regression is developed for surface roughness prediction.The model involves incorporating feature engineering to choose the most concise and pertinent features,as well as developing an ML model.The ML model considers thermal environment parameters and feature screened by feature engineering as inputs,and predicts the surface height as the output.The results demonstrate that the suggested ML framework effectively anticipates the surface shape and associated surface roughness parameters in various heat flow conditions.Compared with the conventional 3D confocal microscope scanning,the method can obtain the surface topography information of the same area in a much shorter time,thus significantly saving time and cost.展开更多
Determining the optimal ceramic content of the ceramics-in-polymer composite electrolytes and the appropriate stack pressure can effectively improve the interfacial contact of solid-state batteries(SSBs).Based on the ...Determining the optimal ceramic content of the ceramics-in-polymer composite electrolytes and the appropriate stack pressure can effectively improve the interfacial contact of solid-state batteries(SSBs).Based on the contact mechanics model and constructed by the conjugate gradient method,continuous convolution,and fast Fourier transform,this paper analyzes and compares the interfacial contact responses involving the polymers commonly used in SSBs,which provides the original training data for machine learning.A support vector regression model is established to predict the relationship between the content of ceramics and the interfacial resistance.The Bayesian optimization and K-fold cross-validation are introduced to find the optimal combination of hyperparameters,which accelerates the training process and improves the model’s accuracy.We found the relationship between the content of ceramics,the stack pressure,and the interfacial resistance.The results can be taken as a reference for the design of the low-resistance composite electrolytes for solid-state batteries.展开更多
Bifunctional oxide-zeolite-based composites(OXZEO)have emerged as promising materials for the direct conversion of syngas to olefins.However,experimental screening and optimization of reaction parameters remain resour...Bifunctional oxide-zeolite-based composites(OXZEO)have emerged as promising materials for the direct conversion of syngas to olefins.However,experimental screening and optimization of reaction parameters remain resource-intensive.To address this challenge,we implemented a three-stage framework integrating machine learning,Bayesian optimization,and experimental validation,utilizing a carefully curated dataset from the literature.Our ensemble-tree model(R^(2)>0.87)identified Zn-Zr and Cu-Mg binary mixed oxides as the most effective OXZEO systems,with their light olefin space-time yields confirmed by physically mixing with HSAPO-34 through experimental validation.Density functional theory calculations further elucidated the activity trends between Zn-Zr and Cu-Mg mixed oxides.Among 16 catalyst and reaction condition descriptors,the oxide/zeolite ratio,reaction temperature,and pressure emerged as the most significant factors.This interpretable,data-driven framework offers a versatile approach that can be applied to other catalytic processes,providing a powerful tool for experiment design and optimization in catalysis.展开更多
Fine-grained image classification, which aims to distinguish images with subtle distinctions, is a challenging task for two main reasons: lack of sufficient training data for every class and difficulty in learning dis...Fine-grained image classification, which aims to distinguish images with subtle distinctions, is a challenging task for two main reasons: lack of sufficient training data for every class and difficulty in learning discriminative features for representation. In this paper, to address the two issues, we propose a two-phase framework for recognizing images from unseen fine-grained classes, i.e., zeroshot fine-grained classification. In the first feature learning phase, we finetune deep convolutional neural networks using hierarchical semantic structure among fine-grained classes to extract discriminative deep visual features. Meanwhile, a domain adaptation structure is induced into deep convolutional neural networks to avoid domain shift from training data to test data. In the second label inference phase, a semantic directed graph is constructed over attributes of fine-grained classes. Based on this graph, we develop a label propagation algorithm to infer the labels of images in the unseen classes. Experimental results on two benchmark datasets demonstrate that our model outperforms the state-of-the-art zero-shot learning models. In addition, the features obtained by our feature learning model also yield significant gains when they are used by other zero-shot learning models, which shows the flexility of our model in zero-shot finegrained classification.展开更多
With the rapid development of artificial intelligence,magnetocaloric materials as well as other materials are being developed with increased efficiency and enhanced performance.However,most studies do not take phase t...With the rapid development of artificial intelligence,magnetocaloric materials as well as other materials are being developed with increased efficiency and enhanced performance.However,most studies do not take phase transitions into account,and as a result,the predictions are usually not accurate enough.In this context,we have established an explicable relationship between alloy compositions and phase transition by feature imputation.A facile machine learning is proposed to screen candidate NiMn-based Heusler alloys with desired magnetic entropy change and magnetic transition temperature with a high accuracy R^(2)≈0.98.As expected,the measured properties of prepared NiMn-based alloys,including phase transition type,magnetic entropy changes and transition temperature,are all in good agreement with the ML predictions.As well as being the first to demonstrate an explicable relationship between alloy compositions,phase transitions and magnetocaloric properties,our proposed ML model is highly predictive and interpretable,which can provide a strong theoretical foundation for identifying high-performance magnetocaloric materials in the future.展开更多
In medium/high entropy alloys, their mechanical properties are strongly dependent on the chemicalelemental composition. Thus, searching for optimum elemental composition remains a critical issue to maximize the mechan...In medium/high entropy alloys, their mechanical properties are strongly dependent on the chemicalelemental composition. Thus, searching for optimum elemental composition remains a critical issue to maximize the mechanical performance. However, this issue solved by traditional optimization process via "trial and error" or experiences of domain experts is extremely difficult. Here we propose an approach based on high-throughput simulation combined machine learning to obtain medium entropy alloys with high strength and low cost. This method not only obtains a large amount of data quickly and accurately,but also helps us to determine the relationship between the composition and mechanical properties.The results reveal a vital importance of high-throughput simulation combined machine learning to find best mechanical properties in a wide range of elemental compositions for development of alloys with expected performance.展开更多
As an essential function of encrypted Internet traffic analysis,encrypted traffic service classification can support both coarse-grained network service traffic management and security supervision.However,the traditio...As an essential function of encrypted Internet traffic analysis,encrypted traffic service classification can support both coarse-grained network service traffic management and security supervision.However,the traditional plaintext-based Deep Packet Inspection(DPI)method cannot be applied to such a classification.Moreover,machine learning-based existing methods encounter two problems during feature selection:complex feature overcost processing and Transport Layer Security(TLS)version discrepancy.In this paper,we consider differences between encryption network protocol stacks and propose a composite deep learning-based method in multiprotocol environments using a sliding multiple Protocol Data Unit(multiPDU)length sequence as features by fully utilizing the Markov property in a multiPDU length sequence and maintaining suitability with a TLS-1.3 environment.Control experiments show that both Length-Sensitive(LS)composite deep learning model using a capsule neural network and LS-long short time memory achieve satisfactory effectiveness in F1-score and performance.Owing to faster feature extraction,our method is suitable for actual network environments and superior to state-of-the-art methods.展开更多
Mechanical metamaterials such as auxetic materials have attracted great interest due to their unusual properties that are dictated by their architectures.However,these architected materials usually have low stiffness ...Mechanical metamaterials such as auxetic materials have attracted great interest due to their unusual properties that are dictated by their architectures.However,these architected materials usually have low stiffness because of the bending or rotation deformation mechanisms in the microstructures.In this work,a convolutional neural network(CNN)based self-learning multi-objective optimization is performed to design digital composite materials.The CNN models have undergone rigorous training using randomly generated two-phase digital composite materials,along with their corresponding Poisson's ratios and stiffness values.Then the CNN models are used for designing composite material structures with the minimum Poisson's ratio at a given volume fraction constraint.Furthermore,we have designed composite materials with optimized stiffness while exhibiting a desired Poisson's ratio(negative,zero,or positive).The optimized designs have been successfully and efficiently obtained,and their validity has been confirmed through finite element analysis results.This self-learning multi-objective optimization model offers a promising approach for achieving comprehensive multi-objective optimization.展开更多
基金financially supported by the National Key Research and Development Project of China(No.2022YFB3806900)。
文摘The complex interactions and conflicting performance demands in multi-component composites pose significant challenges for achieving balanced multi-property optimization through conventional trial-and-error approaches.Machine learning(ML)offers a promising solution,markedly improving materials discovery efficiency.However,the high dimensionality of feature spaces in such systems has long impeded effective ML-driven feature representation and inverse design.To overcome this,we present an Intelligent Screening System(ISS)framework to accelerate the discovery of optimal formulations balancing four key properties in 15-component PTFE-based copper-clad laminate composites(PTFE-CCLCs).ISS adopts modular descriptors based on the physical information of component volume fractions,thereby simplifying the feature representation.By leveraging the inverse prediction capability of ML models and constructing a performance-driven virtual candidate database,ISS significantly reduced the computational complexity associated with high-dimensional spaces.Experimental validation confirmed that ISSoptimized formulations exhibited superior synergy,notably resolving the trade-off between thermal conductivity and peel strength,and outperform many commercial counterparts.Despite limited data and inherent process variability,ISS achieved an average prediction accuracy of 76.5%,with thermal conductivity predictions exceeding 90%,demonstrating robust reliability.This work provides an innovative,efficient strategy for multifunctional optimization and accelerated discovery in ultra-complex composite systems,highlighting the integration of ML and advanced materials design.
基金Shanghai University of Engineering Science Undergraduate Innovation Training Program(Project No.:cx2521005)。
文摘To counteract small sample size,severe class imbalance and high feature redundancy in 90-day mRS prediction after stroke,this study proposes a four-stage pipeline-“ADASYN re-sampling→clinical+statistical feature screening→dimensionality reduction→5-fold cross-validation”-and benchmark composite deep-learning architectures.ADASYN first balances the minority classes in the original feature space.Next,a tri-level filter(clinical domain knowledge,variance threshold,mutual information)removes clinically meaningless or redundant variables,after which PCA compresses the remaining features while preserving critical neurological signatures(e.g.,brain-herniation history).Four hybrid CNN-RNN models are trained and compared under strict 5-fold cross-validation;the optimal ensemble yields stable,clinically interpretable probabilities that can support individualized rehabilitation planning.
基金supported by the National Natural Science Foundation of China(Grant No.52405041)the Major Program of the Zhejiang Provincial Natural Science Foundation of China(Grant No.LD25E050001)the Key R&D Program of Zhejiang Province(Grant No.2025C01186)。
文摘Humanoid robots hold significant promise for social interaction and emotional companionship.However,their effectiveness hinges on the ability to convey nuanced and authentic emotions.Here,we presented a universal humanoid robot head with a facial kinematics model.Using a reinforcement learning framework guided by symmetry assessment,emotion decoupling,and MLLM authenticity evaluation,our system autonomously learns to generate adaptive facial expressions through dynamic landmark adjustments.By transferring the simulation training results to real-world environments,the robot can perform natural and expressive expressions.Another novel feature is the independent regulation of emotion intensity and expression magnitude across emotional categories,which enhances the ability to achieve culturally adaptive and socially resonant robotic expressions significantly.This research advances adaptive humanoid interaction,offering an easier and more efficient pathway toward culturally resonant and psychologically plausible robotic expressions.
文摘Predicting the behavior of renewable energy systems requires models capable of generating accurate forecasts from limited historical data,a challenge that becomes especially pronounced when commissioning new facil-ities where operational records are scarce.This review aims to synthesize recent progress in data-efficient deep learning approaches for addressing such“cold-start”forecasting problems.It primarily covers three interrelated domains—solar photovoltaic(PV),wind power,and electrical load forecasting—where data scarcity and operational variability are most critical,while also including representative studies on hydropower and carbon emission prediction to provide a broader systems perspective.To this end,we examined trends from over 150 predominantly peer-reviewed studies published between 2019 and mid-2025,highlighting advances in zero-shot and few-shot meta-learning frameworks that enable rapid model adaptation with minimal labeled data.Moreover,transfer learning approaches combined with spatiotemporal graph neural networks have been employed to transfer knowledge from existing energy assets to new,data-sparse environments,effectively capturing hidden dependencies among geographic features,meteorological dynamics,and grid structures.Synthetic data generation has further proven valuable for expanding training samples and mitigating overfitting in cold-start scenarios.In addition,large language models and explainable artificial intelligence(XAI)—notably conversational XAI systems—have been used to interpret and communicate complex model behaviors in accessible terms,fostering operator trust from the earliest deployment stages.By consolidating methodological advances,unresolved challenges,and open-source resources,this review provides a coherent overview of deep learning strategies that can shorten the data-sparse ramp-up period of new energy infrastructures and accelerate the transition toward resilient,low-carbon electricity grids.
基金supported and partially funded by the National Natural Science Foundation of China(52288101)the China Postdoctoral Science Foundation(2024M761535)supported by the High Performance Computing Centers at Eastern Institute of Technology,Ningbo,and Ningbo Institute of Digital Twin.
文摘Computational solid mechanics has become an indispensable approach in engineering,and numerical investigation of fracturing in composites is essential,as composites are widely used in structural applications.Crack evolution in composites is the path to elucidating the relationship between microstructures and fracture performance,but crack-based finite-element methods are computationally expensive and time-consuming,which limits their application in computation-intensive scenarios.Consequently,this study proposes a deep learning framework called Crack-Net for instant prediction of the dynamic crack growth process,as well as its strain-stress curve.Specifically,Crack-Net introduces an implicit constraint technique,which incorporates the relationship between crack evolution and stress response into the network architecture.This technique substantially reduces data requirements while improving predictive accuracy.The transfer learning technique enables Crack-Net to handle composite materials with reinforcements of different strengths.Trained on high-accuracy fracture development datasets from phase field simulations,the proposed framework is capable of tackling intricate scenarios,involving materials with diverse interfaces,varying initial conditions,and the intricate elastoplastic fracture process.The proposed Crack-Net holds great promise for practical applications in engineering and materials science,in which accurate and efficient fracture prediction is crucial for optimizing material performance and microstructural design.
基金supported by National Key Research and Development Program of China(Grant No.2023YFB4604100)National Key Research and Development Program of China(Grant No.2022YFB3806104)+4 种基金Key Research and Development Program in Shaanxi Province(Grant No.2021LLRH-08-17)Young Elite Scientists Sponsorship Program by CAST(No.2023QNRC001)K C Wong Education Foundation of ChinaYouth Innovation Team of Shaanxi Universities of ChinaKey Research and Development Program of Shaanxi Province(Grant 2021LLRH-08-3.1).
文摘Ensuring the consistent mechanical performance of three-dimensional(3D)-printed continuous fiber-reinforced composites is a significant challenge in additive manufacturing.The current reliance on manual monitoring exacerbates this challenge by rendering the process vulnerable to environmental changes and unexpected factors,resulting in defects and inconsistent product quality,particularly in unmanned long-term operations or printing in extreme environments.To address these issues,we developed a process monitoring and closed-loop feedback control strategy for the 3D printing process.Real-time printing image data were captured and analyzed using a well-trained neural network model,and a real-time control module-enabled closed-loop feedback control of the flow rate was developed.The neural network model,which was based on image processing and artificial intelligence,enabled the recognition of flow rate values with an accuracy of 94.70%.The experimental results showed significant improvements in both the surface performance and mechanical properties of printed composites,with three to six times improvement in tensile strength and elastic modulus,demonstrating the effectiveness of the strategy.This study provides a generalized process monitoring and feedback control method for the 3D printing of continuous fiber-reinforced composites,and offers a potential solution for remote online monitoring and closed-loop adjustment in unmanned or extreme space environments.
基金supported by the Gas Hydrate R&D Organization and the Korea Institute of Geoscience and Mineral Resources(KIGAM)(GP2021-010)supported by the National Research Foundation of Korea(NRF)grant funded by the Korean government(MSIT)(No.2021R1C1C1004460)Korea Institute of Energy Technology Evaluation and Planning(KETEP)grant funded by the Korean government(MOTIE)(20214000000500,Training Program of CCUS for Green Growth).
文摘Gas hydrate(GH)is an unconventional resource estimated at 1000-120,000 trillion m^(3)worldwide.Research on GH is ongoing to determine its geological and flow characteristics for commercial produc-tion.After two large-scale drilling expeditions to study the GH-bearing zone in the Ulleung Basin,the mineral composition of 488 sediment samples was analyzed using X-ray diffraction(XRD).Because the analysis is costly and dependent on experts,a machine learning model was developed to predict the mineral composition using XRD intensity profiles as input data.However,the model’s performance was limited because of improper preprocessing of the intensity profile.Because preprocessing was applied to each feature,the intensity trend was not preserved even though this factor is the most important when analyzing mineral composition.In this study,the profile was preprocessed for each sample using min-max scaling because relative intensity is critical for mineral analysis.For 49 test data among the 488 data,the convolutional neural network(CNN)model improved the average absolute error and coefficient of determination by 41%and 46%,respectively,than those of CNN model with feature-based pre-processing.This study confirms that combining preprocessing for each sample with CNN is the most efficient approach for analyzing XRD data.The developed model can be used for the compositional analysis of sediment samples from the Ulleung Basin and the Korea Plateau.In addition,the overall procedure can be applied to any XRD data of sediments worldwide.
基金co-supported by the Aeronautical Science Foundation of China(Nos.2018ZA52002,2019ZA052011).
文摘With the availability of high-performance computing technology and the development of advanced numerical simulation methods, Computational Fluid Dynamics (CFD) is becoming more and more practical and efficient in engineering. As one of the high-precision representative algorithms, the high-order Discontinuous Galerkin Method (DGM) has not only attracted widespread attention from scholars in the CFD research community, but also received strong development. However, when DGM is extended to high-speed aerodynamic flow field calculations, non-physical numerical Gibbs oscillations near shock waves often significantly affect the numerical accuracy and even cause calculation failure. Data driven approaches based on machine learning techniques can be used to learn the characteristics of Gibbs noise, which motivates us to use it in high-speed DG applications. To achieve this goal, labeled data need to be generated in order to train the machine learning models. This paper proposes a new method for denoising modeling of Gibbs phenomenon using a machine learning technique, the zero-shot learning strategy, to eliminate acquiring large amounts of CFD data. The model adopts a graph convolutional network combined with graph attention mechanism to learn the denoising paradigm from synthetic Gibbs noise data and generalize to DGM numerical simulation data. Numerical simulation results show that the Gibbs denoising model proposed in this paper can suppress the numerical oscillation near shock waves in the high-order DGM. Our work automates the extension of DGM to high-speed aerodynamic flow field calculations with higher generalization and lower cost.
基金supported by project RELIABLE(PTDC/EEI-AUT/3522/2020)R&D Unit SYSTEC-Base(UIDB001472020)+1 种基金Programmatic(UIDP001472020)funds-and Associate Laboratory Advanced Production and Intelligent Systems ARISE-LAP01122020funded by national funds through the FCT/MCTES(PIDDAC).
文摘The ability of mobile robots to plan and execute a path is foundational to various path-planning challenges,particularly Coverage Path Planning.While this task has been typically tackled with classical algorithms,these often struggle with flexibility and adaptability in unknown environments.On the other hand,recent advances in Reinforcement Learning offer promising approaches,yet a significant gap in the literature remains when it comes to generalization over a large number of parameters.This paper presents a unified,generalized framework for coverage path planning that leverages value-based deep reinforcement learning techniques.The novelty of the framework comes from the design of an observation space that accommodates different map sizes,an action masking scheme that guarantees safety and robustness while also serving as a learning-fromdemonstration technique during training,and a unique reward function that yields value functions that are size-invariant.These are coupled with a curriculum learning-based training strategy and parametric environment randomization,enabling the agent to tackle complete or partial coverage path planning with perfect or incomplete knowledge while generalizing to different map sizes,configurations,sensor payloads,and sub-tasks.Our empirical results show that the algorithm can perform zero-shot learning scenarios at a near-optimal level in environments that follow a similar distribution as during training,outperforming a greedy heuristic by sixfold.Furthermore,in out-of-distribution environments,our method surpasses existing state-of-the-art algorithms in most zero-shot and all few-shot scenarios,paving the way for generalizable and adaptable path-planning algorithms.
文摘Carbon nanotube-reinforced cement composites have gained significant attention due to their enhanced mechanical properties,particularly in compressive and flexural strength.Despite extensive research,the influence of various parameters on these properties remains inadequately understood,primarily due to the complex interactions within the composites.This study addresses this gap by employingmachine learning techniques to conduct a sensitivity analysis on the compressive and flexural strength of carbon nanotube-reinforced cement composites.It systematically evaluates nine data-preprocessing techniques and benchmarks eleven machine-learning algorithms to reveal tradeoffs between predictive accuracy and computational complexity,which has not previously been explored in carbon nanotube-reinforced cement composite research.In this regard,four main factors are considered in the sensitivity analysis,which are the machine learning model type,the data pre-processing technique,and the effect of the concrete constituent materials on the compressive and flexural strength both globally through feature importance assessment and locally through partial dependence analysis.Accordingly,this research optimizes ninety-nine models representing combinations of eleven machine learning algorithms and nine data preprocessing techniques to accurately predict the mechanical properties of carbon nanotube-reinforced cement composites.Moreover,the study aims to unravel the relationships between different parameters and their impact on the composite’s strength by utilizing feature importance and partial dependence analyses.This research is crucial as it provides a comprehensive understanding of the factors influencing the performance of carbon nanotube-reinforced cement composites,which is vital for their efficient design and application in construction.The use of machine learning in this context not only enhances predictive accuracy but also offers insights that are often challenging to obtain through traditional experimental methods.The findings contribute to the field by highlighting the potential of advanced data-driven approaches in optimizing and understanding advanced composite materials,paving the way for more durable and resilient construction materials.
基金Supported by Taiwan National Science and Technology Council,No.NSTC114-2221-E-035-036Taichung Veterans General Hospital/Feng Chia University Joint Research Program,No.TCVGH-FCU1148207.
文摘BACKGROUND Esophageal cancer carries a poor prognosis with low 5-year survival rates and limited early detection options.The skeletal muscle index at the L3 vertebral level is a well-established prognostic marker in esophageal cancer,but most follow-up computed tomography(CT)scans do not extend to L3 and limiting its utility.Radiomics has emerged as a powerful tool for extracting prognostic information from medical images.AIM To evaluate the influential features for esophageal cancer prognosis by integrating radiomic and body compositionbased indices of skeletal muscle and adipose tissue at the T12 level from both pretreatment and follow-up CT images,in order to assess their value in predicting overall survival(OS).METHODS This retrospective study included 212 esophageal cancer patients who underwent concurrent chemoradiotherapy,with both pretreatment and follow-up chest CT scans available.Body organ analysis(BOA)and radiomic features were extracted from skeletal muscle and adipose tissue at the T12 level using automated tools.Four feature subsets(no-radiomics,pretreatment only,follow-up only,and combined inputs)were developed using logistic regression(LR)with least absolute shrinkage and selection operator for feature selection,followed by Cox regression.Prognostic models-including nomogram,support vector classifier,LR,and extra trees classifier-were constructed to predict 1-,2-,and 3-year OS.RESULTS The model integrating both BOA and radiomics from pretreatment and follow-up CT,combined with clinical data,achieved the best performance for 2-year OS prediction,with an area under the time-dependent receiver operating characteristic curve of 0.91,sensitivity of 0.81,and specificity of 0.88 using the LR model.The most predictive features included both clinical variables,body composition indices,and radiomic features,particularly from follow-up VAT.Follow-up imaging contributed significantly to model performance,reinforcing its value in treatment response evaluation.CONCLUSION This is the first study to demonstrate that BOA indices and their corresponding radiomics at the T12-level from both pretreatment and follow-up CT scans-combined with clinical data-can provide accurate prognostic information for esophageal cancer.This approach offers a practical alternative when L3-level imaging is unavailable and supports the clinical integration of automated T12-based imaging biomarkers.The integration of these imaging features with clinical parameters enhances the prediction of survival outcomes and contributes to non-invasive,personalized treatment planning.
基金supported by the National Natural Science Foundation of China(Grant Nos.U2241240,12172045,and 12221002).
文摘In thermal protection structures,controlling and optimizing the surface roughness of carbon/phenolic(C/Ph)composites can effectively improve thermal protection performance and ensure the safe operation of carriers in high-temperature environments.This paper introduces a machine learning(ML)framework to forecast the surface roughness of carbon-phenolic composites under various thermal conditions by employing an ML algorithm derived from historical experimental datasets.Firstly,ablation experiments and collection of surface roughness height data of C/Ph composites under different thermal environments were conducted in an electric arc wind tunnel.Then,an ML model based on Ridge regression is developed for surface roughness prediction.The model involves incorporating feature engineering to choose the most concise and pertinent features,as well as developing an ML model.The ML model considers thermal environment parameters and feature screened by feature engineering as inputs,and predicts the surface height as the output.The results demonstrate that the suggested ML framework effectively anticipates the surface shape and associated surface roughness parameters in various heat flow conditions.Compared with the conventional 3D confocal microscope scanning,the method can obtain the surface topography information of the same area in a much shorter time,thus significantly saving time and cost.
基金the National Natural Science Foundation of China(12102085)the Postdoctoral Science Foundation of China(2023M730504)the Sichuan Province Regional Innovation and Cooperation Project(2024YFHZ0210).
文摘Determining the optimal ceramic content of the ceramics-in-polymer composite electrolytes and the appropriate stack pressure can effectively improve the interfacial contact of solid-state batteries(SSBs).Based on the contact mechanics model and constructed by the conjugate gradient method,continuous convolution,and fast Fourier transform,this paper analyzes and compares the interfacial contact responses involving the polymers commonly used in SSBs,which provides the original training data for machine learning.A support vector regression model is established to predict the relationship between the content of ceramics and the interfacial resistance.The Bayesian optimization and K-fold cross-validation are introduced to find the optimal combination of hyperparameters,which accelerates the training process and improves the model’s accuracy.We found the relationship between the content of ceramics,the stack pressure,and the interfacial resistance.The results can be taken as a reference for the design of the low-resistance composite electrolytes for solid-state batteries.
基金funded by the KRICT Project (KK2512-10) of the Korea Research Institute of Chemical Technology and the Ministry of Trade, Industry and Energy (MOTIE)the Korea Institute for Advancement of Technology (KIAT) through the Virtual Engineering Platform Program (P0022334)+1 种基金supported by the Carbon Neutral Industrial Strategic Technology Development Program (RS-202300261088) funded by the Ministry of Trade, Industry & Energy (MOTIE, Korea)Further support was provided by research fund of Chungnam National University。
文摘Bifunctional oxide-zeolite-based composites(OXZEO)have emerged as promising materials for the direct conversion of syngas to olefins.However,experimental screening and optimization of reaction parameters remain resource-intensive.To address this challenge,we implemented a three-stage framework integrating machine learning,Bayesian optimization,and experimental validation,utilizing a carefully curated dataset from the literature.Our ensemble-tree model(R^(2)>0.87)identified Zn-Zr and Cu-Mg binary mixed oxides as the most effective OXZEO systems,with their light olefin space-time yields confirmed by physically mixing with HSAPO-34 through experimental validation.Density functional theory calculations further elucidated the activity trends between Zn-Zr and Cu-Mg mixed oxides.Among 16 catalyst and reaction condition descriptors,the oxide/zeolite ratio,reaction temperature,and pressure emerged as the most significant factors.This interpretable,data-driven framework offers a versatile approach that can be applied to other catalytic processes,providing a powerful tool for experiment design and optimization in catalysis.
基金supported by National Basic Research Program of China (973 Program) (No. 2015CB352502)National Nature Science Foundation of China (No. 61573026)Beijing Nature Science Foundation (No. L172037)
文摘Fine-grained image classification, which aims to distinguish images with subtle distinctions, is a challenging task for two main reasons: lack of sufficient training data for every class and difficulty in learning discriminative features for representation. In this paper, to address the two issues, we propose a two-phase framework for recognizing images from unseen fine-grained classes, i.e., zeroshot fine-grained classification. In the first feature learning phase, we finetune deep convolutional neural networks using hierarchical semantic structure among fine-grained classes to extract discriminative deep visual features. Meanwhile, a domain adaptation structure is induced into deep convolutional neural networks to avoid domain shift from training data to test data. In the second label inference phase, a semantic directed graph is constructed over attributes of fine-grained classes. Based on this graph, we develop a label propagation algorithm to infer the labels of images in the unseen classes. Experimental results on two benchmark datasets demonstrate that our model outperforms the state-of-the-art zero-shot learning models. In addition, the features obtained by our feature learning model also yield significant gains when they are used by other zero-shot learning models, which shows the flexility of our model in zero-shot finegrained classification.
基金supported by the National Key R&D Program of China(No.2022YFE0109500)the National Natural Science Foundation of China(Nos.52071255,52301250,52171190 and 12304027)+2 种基金the Key R&D Project of Shaanxi Province(No.2022GXLH-01-07)the Fundamental Research Funds for the Central Universities(China)the World-Class Universities(Disciplines)and the Characteristic Development Guidance Funds for the Central Universities.
文摘With the rapid development of artificial intelligence,magnetocaloric materials as well as other materials are being developed with increased efficiency and enhanced performance.However,most studies do not take phase transitions into account,and as a result,the predictions are usually not accurate enough.In this context,we have established an explicable relationship between alloy compositions and phase transition by feature imputation.A facile machine learning is proposed to screen candidate NiMn-based Heusler alloys with desired magnetic entropy change and magnetic transition temperature with a high accuracy R^(2)≈0.98.As expected,the measured properties of prepared NiMn-based alloys,including phase transition type,magnetic entropy changes and transition temperature,are all in good agreement with the ML predictions.As well as being the first to demonstrate an explicable relationship between alloy compositions,phase transitions and magnetocaloric properties,our proposed ML model is highly predictive and interpretable,which can provide a strong theoretical foundation for identifying high-performance magnetocaloric materials in the future.
基金supported financially by the Foundation for Innovative Research Groups of the National Natural Science Foundation of China (No. 51621004)the National Natural Science Foundation of China (Nos. 51871092, 11772122, 51625404, 51771232+5 种基金51671217)the State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body (No. 71865015)the State Key Laboratory of Powder Metallurgythe National Key Research and Development Program of China (Nos. 2016YFB0700300 and 2016YFB1100103)support of the U.S. Army Research Office Project (Nos. W911NF-13-1-0438 and W911NF-19-2-0049) with the program managers,Drs. M.P. Bakas,S.N. Mathaudhusupport from the National Science Foundation (Nos. DMR-1611180 and 1809640)with the program directors,Drs. J. Yang,J.G. Shiflet,and D. Farkas。
文摘In medium/high entropy alloys, their mechanical properties are strongly dependent on the chemicalelemental composition. Thus, searching for optimum elemental composition remains a critical issue to maximize the mechanical performance. However, this issue solved by traditional optimization process via "trial and error" or experiences of domain experts is extremely difficult. Here we propose an approach based on high-throughput simulation combined machine learning to obtain medium entropy alloys with high strength and low cost. This method not only obtains a large amount of data quickly and accurately,but also helps us to determine the relationship between the composition and mechanical properties.The results reveal a vital importance of high-throughput simulation combined machine learning to find best mechanical properties in a wide range of elemental compositions for development of alloys with expected performance.
基金supported by the General Program of the National Natural Science Foundation of China under Grant No.62172093the National Key R&D Program of China under Grant No.2018YFB1800602+1 种基金2019 Industrial Internet Innovation and Development Project,Ministry of Industry and Information Technology(MIIT)under Grant No.6709010003Ministry of Education-China Mobile Research Fund under Grant No.MCM20180506。
文摘As an essential function of encrypted Internet traffic analysis,encrypted traffic service classification can support both coarse-grained network service traffic management and security supervision.However,the traditional plaintext-based Deep Packet Inspection(DPI)method cannot be applied to such a classification.Moreover,machine learning-based existing methods encounter two problems during feature selection:complex feature overcost processing and Transport Layer Security(TLS)version discrepancy.In this paper,we consider differences between encryption network protocol stacks and propose a composite deep learning-based method in multiprotocol environments using a sliding multiple Protocol Data Unit(multiPDU)length sequence as features by fully utilizing the Markov property in a multiPDU length sequence and maintaining suitability with a TLS-1.3 environment.Control experiments show that both Length-Sensitive(LS)composite deep learning model using a capsule neural network and LS-long short time memory achieve satisfactory effectiveness in F1-score and performance.Owing to faster feature extraction,our method is suitable for actual network environments and superior to state-of-the-art methods.
文摘Mechanical metamaterials such as auxetic materials have attracted great interest due to their unusual properties that are dictated by their architectures.However,these architected materials usually have low stiffness because of the bending or rotation deformation mechanisms in the microstructures.In this work,a convolutional neural network(CNN)based self-learning multi-objective optimization is performed to design digital composite materials.The CNN models have undergone rigorous training using randomly generated two-phase digital composite materials,along with their corresponding Poisson's ratios and stiffness values.Then the CNN models are used for designing composite material structures with the minimum Poisson's ratio at a given volume fraction constraint.Furthermore,we have designed composite materials with optimized stiffness while exhibiting a desired Poisson's ratio(negative,zero,or positive).The optimized designs have been successfully and efficiently obtained,and their validity has been confirmed through finite element analysis results.This self-learning multi-objective optimization model offers a promising approach for achieving comprehensive multi-objective optimization.
