Satellite communication technology has emerged as a key solution to address the challenges of data transmission in remote areas.By overcoming the limitations of traditional terrestrial communication networks,it enable...Satellite communication technology has emerged as a key solution to address the challenges of data transmission in remote areas.By overcoming the limitations of traditional terrestrial communication networks,it enables long-distance data transmission anytime and anywhere,ensuring the timely and accurate delivery of water level data,which is particularly crucial for fishway water level monitoring.To enhance the effectiveness of fishway water level monitoring,this study proposes a multi-task learning model,AS-SOMTF,designed for real-time and comprehensive prediction.The model integrates auxiliary sequences with primary input sequences to capture complex relationships and dependencies,thereby improving representational capacity.In addition,a novel timeseries embedding algorithm,AS-SOM,is introduced,which combines generative inference and pooling operations to optimize prediction efficiency for long sequences.This innovation not only ensures the timely transmission of water level data but also enhances the accuracy of real-time monitoring.Compared with traditional models such as Transformer and Long Short-Term Memory(LSTM)networks,the proposed model achieves improvements of 3.8%and 1.4%in prediction accuracy,respectively.These advancements provide more precise technical support for water level forecasting and resource management in the Diqing Tibetan Autonomous Prefecture of the Lancang River,contributing to ecosystem protection and improved operational safety.展开更多
Convective storms and lightning are among the most important weather phenomena that are challenging to forecast.In this study,a novel multi-task learning(MTL)encoder-decoder U-net neural network was developed to forec...Convective storms and lightning are among the most important weather phenomena that are challenging to forecast.In this study,a novel multi-task learning(MTL)encoder-decoder U-net neural network was developed to forecast convective storms and lightning with lead times for up to 90 min,using GOES-16 geostationary satellite infrared brightness temperatures(IRBTs),lightning flashes from Geostationary Lightning Mapper(GLM),and vertically integrated liquid(VIL)from Next Generation Weather Radar(NEXRAD).To cope with the heavily skewed distribution of lightning data,a spatiotemporal exponent-weighted loss function and log-transformed lightning normalization approach were developed.The effects of MTL,single-task learning(STL),and IRBTs as auxiliary input features on convection and lightning nowcasting were investigated.The results showed that normalizing the heavily skew-distributed lightning data along with a log-transformation dramatically outperforms the min-max normalization method for nowcasting an intense lightning event.The MTL model significantly outperformed the STL model for both lightning nowcasting and VIL nowcasting,particularly for intense lightning events.The MTL also helped delay the lightning forecast performance decay with the lead times.Furthermore,incorporating satellite IRBTs as auxiliary input features substantially improved lightning nowcasting,but produced little difference in VIL forecasting.Finally,the MTL model performed better for forecasting both lightning and the VIL of organized convective storms than for isolated cells.展开更多
Aspect-based sentiment analysis(ABSA)is a fine-grained process.Its fundamental subtasks are aspect termextraction(ATE)and aspect polarity classification(APC),and these subtasks are dependent and closely related.Howeve...Aspect-based sentiment analysis(ABSA)is a fine-grained process.Its fundamental subtasks are aspect termextraction(ATE)and aspect polarity classification(APC),and these subtasks are dependent and closely related.However,most existing works on Arabic ABSA content separately address them,assume that aspect terms are preidentified,or use a pipeline model.Pipeline solutions design different models for each task,and the output from the ATE model is used as the input to the APC model,which may result in error propagation among different steps because APC is affected by ATE error.These methods are impractical for real-world scenarios where the ATE task is the base task for APC,and its result impacts the accuracy of APC.Thus,in this study,we focused on a multi-task learning model for Arabic ATE and APC in which the model is jointly trained on two subtasks simultaneously in a singlemodel.This paper integrates themulti-task model,namely Local Cotext Foucse-Aspect Term Extraction and Polarity classification(LCF-ATEPC)and Arabic Bidirectional Encoder Representation from Transformers(AraBERT)as a shred layer for Arabic contextual text representation.The LCF-ATEPC model is based on a multi-head selfattention and local context focus mechanism(LCF)to capture the interactive information between an aspect and its context.Moreover,data augmentation techniques are proposed based on state-of-the-art augmentation techniques(word embedding substitution with constraints and contextual embedding(AraBERT))to increase the diversity of the training dataset.This paper examined the effect of data augmentation on the multi-task model for Arabic ABSA.Extensive experiments were conducted on the original and combined datasets(merging the original and augmented datasets).Experimental results demonstrate that the proposed Multi-task model outperformed existing APC techniques.Superior results were obtained by AraBERT and LCF-ATEPC with fusion layer(AR-LCF-ATEPC-Fusion)and the proposed data augmentation word embedding-based method(FastText)on the combined dataset.展开更多
Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Re...Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.展开更多
Underground engineering projects such as deep tunnel excavation often encounter rockburst disasters accompanied by numerous microseismic events.Rapid interpretation of microseismic signals is crucial for the timely id...Underground engineering projects such as deep tunnel excavation often encounter rockburst disasters accompanied by numerous microseismic events.Rapid interpretation of microseismic signals is crucial for the timely identification of rockbursts.However,conventional processing encompasses multi-step workflows,including classification,denoising,picking,locating,and computational analysis,coupled with manual intervention,which collectively compromise the reliability of early warnings.To address these challenges,this study innovatively proposes the“microseismic stethoscope"-a multi-task machine learning and deep learning model designed for the automated processing of massive microseismic signals.This model efficiently extracts three key parameters that are necessary for recognizing rockburst disasters:rupture location,microseismic energy,and moment magnitude.Specifically,the model extracts raw waveform features from three dedicated sub-networks:a classifier for source zone classification,and two regressors for microseismic energy and moment magnitude estimation.This model demonstrates superior efficiency compared to traditional processing and semi-automated processing,reducing per-event processing time from 0.71 s to 0.49 s to merely 0.036 s.It concurrently achieves 98%accuracy in source zone classification,with microseismic energy and moment magnitude estimation errors of 0.13 and 0.05,respectively.This model has been well applied and validated in the Daxiagu Tunnel case in Sichuan,China.The application results indicate that the model is as accurate as traditional methods in determining source parameters,and thus can be used to identify potential geomechanical processes of rockburst disasters.By enhancing the signal processing reliability of microseismic events,the proposed model in this study presents a significant advancement in the identification of rockburst disasters.展开更多
Federated Learning(FL)enables joint training over distributed devices without data exchange but is highly vulnerable to attacks by adversaries in the form of model poisoning and malicious update injection.This work pr...Federated Learning(FL)enables joint training over distributed devices without data exchange but is highly vulnerable to attacks by adversaries in the form of model poisoning and malicious update injection.This work proposes Secured-FL,a blockchain-based defensive framework that combines smart contract-based authentication,clustering-driven outlier elimination,and dynamic threshold adjustment to defend against adversarial attacks.The framework was implemented on a private Ethereum network with a Proof-of-Authority consensus algorithm to ensure tamper-resistant and auditable model updates.Large-scale simulation on the Cyber Data dataset,under up to 50%malicious client settings,demonstrates Secured-FL achieves 6%-12%higher accuracy,9%-15%lower latency,and approximately 14%less computational expense compared to the PPSS benchmark framework.Additional tests,including confusion matrices,ROC and Precision-Recall curves,and ablation tests,confirm the interpretability and robustness of the defense.Tests for scalability also show consistent performance up to 500 clients,affirming appropriateness to reasonably large deployments.These results make Secured-FL a feasible,adversarially resilient FL paradigm with promising potential for application in smart cities,medicine,and other mission-critical IoT deployments.展开更多
The rapid advancement of machine learning based tight-binding Hamiltonian(MLTB)methods has opened new avenues for efficient and accurate electronic structure simulations,particularly in large-scale systems and long-ti...The rapid advancement of machine learning based tight-binding Hamiltonian(MLTB)methods has opened new avenues for efficient and accurate electronic structure simulations,particularly in large-scale systems and long-time scenarios.This review begins with a concise overview of traditional tight-binding(TB)models,including both(semi-)empirical and first-principles approaches,establishing the foundation for understanding MLTB developments.We then present a systematic classification of existing MLTB methodologies,grouped into two major categories:direct prediction of TB Hamiltonian elements and inference of empirical parameters.A comparative analysis with other ML-based electronic structure models is also provided,highlighting the advancement of MLTB approaches.Finally,we explore the emerging MLTB application ecosystem,highlighting how the integration of MLTB models with a diverse suite of post-processing tools from linear-scaling solvers to quantum transport frameworks and molecular dynamics interfaces is essential for tackling complex scientific problems across different domains.The continued advancement of this integrated paradigm promises to accelerate materials discovery and open new frontiers in the predictive simulation of complex quantum phenomena.展开更多
BACKGROUND:This study aims to develop and validate a machine learning-based in-hospital mortality predictive model for acute aortic syndrome(AAS)in the emergency department(ED)and to derive a simplifi ed version suita...BACKGROUND:This study aims to develop and validate a machine learning-based in-hospital mortality predictive model for acute aortic syndrome(AAS)in the emergency department(ED)and to derive a simplifi ed version suitable for rapid clinical application.METHODS:In this multi-center retrospective cohort study,AAS patient data from three hospitals were analyzed.The modeling cohort included data from the First Affiliated Hospital of Zhengzhou University and the People’s Hospital of Xinjiang Uygur Autonomous Region,with Peking University Third Hospital data serving as the external test set.Four machine learning algorithms—logistic regression(LR),multilayer perceptron(MLP),Gaussian naive Bayes(GNB),and random forest(RF)—were used to develop predictive models based on 34 early-accessible clinical variables.A simplifi ed model was then derived based on fi ve key variables(Stanford type,pericardial eff usion,asymmetric peripheral arterial pulsation,decreased bowel sounds,and dyspnea)via Least Absolute Shrinkage and Selection Operator(LASSO)regression to improve ED applicability.RESULTS:A total of 929 patients were included in the modeling cohort,and 210 were included in the external test set.Four machine learning models based on 34 clinical variables were developed,achieving internal and external validation AUCs of 0.85-0.90 and 0.73-0.85,respectively.The simplifi ed model incorporating fi ve key variables demonstrated internal and external validation AUCs of 0.71-0.86 and 0.75-0.78,respectively.Both models showed robust calibration and predictive stability across datasets.CONCLUSION:Both kinds of models were built based on machine learning tools,and proved to have certain prediction performance and extrapolation.展开更多
This study aims to develop an accurate and robust machine learning model to predict the carbonation depth of fly ash concrete,overcoming the limitations of traditional predictive methods.Five ensemble-based models,suc...This study aims to develop an accurate and robust machine learning model to predict the carbonation depth of fly ash concrete,overcoming the limitations of traditional predictive methods.Five ensemble-based models,such as adaptive boosting(AdaBoost),categorical boosting(CatBoost),gradient boosting regressor(GBR),hist gradient boosting regressor(HistGBR),and extreme gradient boosting(XGBoost),were developed and optimized using 729 high-quality dataset points incorporating seven input parameters,including cement,CO_(2),exposure time,water-binder ratio,fly ash,curing time,and compressive strength.Several performance evaluation metrics were used to compare the models.The GBR model emerged as the best-performing model,based on high coefficient of determination(R^(2))values and balanced error metrics across both validation and testing datasets.While all models performed exceptionally well on the training data,GBR demonstrated superior generalization capability,with R^(2) values of 0.9438 on the validation set and 0.9310 on the testing set.Furthermore,its low mean squared error(MSE),root mean square error(RMSE),mean absolute error(MAE),and median absolute error(MdAE)confirmed its robustness and accuracy.Moreover,shapley additive explanations(SHAP)analysis enhanced the interpretability of predictions,highlighting the curing time and exposure time as the most critical drivers of carbonation depth.展开更多
Oxide dispersion strengthened(ODS)alloys are extensively used owing to high thermostability and creep strength contributed from uniformly dispersed fine oxides particles.However,the existence of these strengthening pa...Oxide dispersion strengthened(ODS)alloys are extensively used owing to high thermostability and creep strength contributed from uniformly dispersed fine oxides particles.However,the existence of these strengthening particles also deteriorates the processability and it is of great importance to establish accurate processing maps to guide the thermomechanical processes to enhance the formability.In this study,we performed particle swarm optimization-based back propagation artificial neural network model to predict the high temperature flow behavior of 0.25wt%Al2O3 particle-reinforced Cu alloys,and compared the accuracy with that of derived by Arrhenius-type constitutive model and back propagation artificial neural network model.To train these models,we obtained the raw data by fabricating ODS Cu alloys using the internal oxidation and reduction method,and conducting systematic hot compression tests between 400 and800℃with strain rates of 10^(-2)-10 S^(-1).At last,processing maps for ODS Cu alloys were proposed by combining processing parameters,mechanical behavior,microstructure characterization,and the modeling results achieved a coefficient of determination higher than>99%.展开更多
This study presents a physics-informed modelling framework that combines finite element method(FEM)simulations and supervised machine learning(ML)to predict the self-healing performance of microbial concrete.A FEniCS-...This study presents a physics-informed modelling framework that combines finite element method(FEM)simulations and supervised machine learning(ML)to predict the self-healing performance of microbial concrete.A FEniCS-based FEM platform resolves multiphysics phenomena including nutrient diffusion,microbial CaCO_(3) precipitation,and stiffness recovery.These simulations,together with experimental data,are used to train ML models(Random Forest yielding normalized RMSE≈0.10)capable of predicting performance over a wide range of design parameters.Feature importance analysis identifies curing temperature,calcium carbonate precipitation rate,crack width,bacterial strain,and encapsulation method as the most influential parameters.The coupled FEM-ML approach enables sensitivity analysis,design optimization,and prediction beyond the training dataset(consistently exceeding 90%healing efficiency).Experimental validation confirms model robustness in both crack closure and strength recovery.This FEM–ML pipeline thus offers a generalizable,interpretable,and scalable strategy for the design of intelligent,self-adaptive construction materials.展开更多
The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in S...The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources.In this paper,we compare two predominant AI-based approaches for the forensic detection of malicious hate speech:(1)finetuning encoder-only models that have been trained in Spanish and(2)In-Context Learning techniques(Zero-and Few-Shot Learning)with large-scale language models.Our approach goes beyond binary classification,proposing a comprehensive,multidimensional evaluation that labels each text by:(1)type of speech,(2)recipient,(3)level of intensity(ordinal)and(4)targeted group(multi-label).Performance is evaluated using an annotated Spanish corpus,standard metrics such as precision,recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting(Zero-to-Few Shot Retention and Zero-to-Few Shot Gain)are applied.The results indicate that fine-tuned encoder-only models(notably MarIA and BETO variants)consistently deliver the strongest and most reliable performance:in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66%depending on the task.Zero-shot approaches are much less stable and typically yield substantially lower performance(observed F1-scores range approximately 0%–39%),often producing invalid outputs in practice.Few-shot prompting(e.g.,Qwen 38B,Mistral 7B)generally improves stability and recall relative to pure zero-shot,bringing F1-scores into a moderate range of approximately 20%–51%but still falling short of fully fine-tuned models.These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.展开更多
Modern industrial environments require uninterrupted machinery operation to maintain productivity standards while ensuring safety and minimizing costs.Conventional maintenance methods,such as reactive maintenance(i.e....Modern industrial environments require uninterrupted machinery operation to maintain productivity standards while ensuring safety and minimizing costs.Conventional maintenance methods,such as reactive maintenance(i.e.,run to failure)or time-based preventive maintenance(i.e.,scheduled servicing),prove ineffective for complex systems with many Internet of Things(IoT)devices and sensors because they fall short in detecting faults at early stages when it is most crucial.This paper presents a predictive maintenance framework based on a hybrid deep learning model that integrates the capabilities of Long Short-Term Memory(LSTM)Networks and Convolutional Neural Networks(CNNs).The framework integrates spatial feature extraction and temporal sequence modeling to accurately classify the health state of industrial equipment into three categories,including Normal,Require Maintenance,and Failed.The framework uses a modular pipeline that includes IoT-enabled data collection along with secure transmission methods to manage cloud storage and provide real-time fault classification.The FD004 subset of the NASA C-MAPSS dataset,containing multivariate sensor readings from aircraft engines,serves as the training and evaluation data for the model.Experimental results show that the LSTM-CNN model outperforms baseline models such as LSTM-SVM and LSTM-RNN,achieving an overall average accuracy of 86.66%,precision of 86.00%,recall of 86.33%,and F1-score of 86.33%.Contrary to the previous LSTM-CNN-based predictive maintenance models that either provide a binary classification or rely on synthetically balanced data,our paper provides a three-class maintenance state(i.e.,Normal,Require Maintenance,and Failed)along with threshold-based labeling that retains the true nature of the degradation.In addition,our work also provides an IoT-to-cloud-based modular architecture for deployment.It offers Computerized Maintenance Management System(CMMS)integration,making our proposed solution not only technically sound but also practical and innovative.The solution achieves real-world industrial deployment readiness through its reliable performance alongside its scalable system design.展开更多
Accurate short-term forecast of offshore wind fields is still challenging for numerical weather prediction models.Based on three years of 48-hour forecast data from the European Centre for Medium-Range Weather Forecas...Accurate short-term forecast of offshore wind fields is still challenging for numerical weather prediction models.Based on three years of 48-hour forecast data from the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System global model(ECMWF-IFS)over 14 offshore weather stations along the coast of Shandong Province,this study introduces a multi-task learning(MTL)model(TabNet-MTL),which significantly improves the forecast bias of near-surface wind direction and speed simultaneously.TabNet-MTL adopts the feature engineering method,utilizes mean square error as the loss function,and employs the 5-fold cross validation method to ensure the generalization ability of the trained model.It demonstrates superior skills in wind field correction across different forecast lead times over all stations compared to its single-task version(TabNet-STL)and three other popular single-task learning models(Random Forest,LightGBM,and XGBoost).Results show that it significantly reduces root mean square error of the ECMWF-IFS wind speed forecast from 2.20 to 1.25 m s−1,and increases the forecast accuracy of wind direction from 50%to 65%.As an explainable deep learning model,the weather stations and long-term temporal statistics of near-surface wind speed are identified as the most influential variables for TabNet-MTL in constructing its feature engineering.展开更多
As the group-buying model shows significant progress in attracting new users,enhancing user engagement,and increasing platform profitability,providing personalized recommendations for group-buying users has emerged as...As the group-buying model shows significant progress in attracting new users,enhancing user engagement,and increasing platform profitability,providing personalized recommendations for group-buying users has emerged as a new challenge in the field of recommendation systems.This paper introduces a group-buying recommendation model based on multi-head attention mechanisms and multi-task learning,termed the Multi-head Attention Mechanisms and Multi-task Learning Group-Buying Recommendation(MAMGBR)model,specifically designed to optimize group-buying recommendations on e-commerce platforms.The core dataset of this study comes from the Chinese maternal and infant e-commerce platform“Beibei,”encompassing approximately 430,000 successful groupbuying actions and over 120,000 users.Themodel focuses on twomain tasks:recommending items for group organizers(Task Ⅰ)and recommending participants for a given group-buying event(Task Ⅱ).In model evaluation,MAMGBR achieves an MRR@10 of 0.7696 for Task I,marking a 20.23%improvement over baseline models.Furthermore,in Task II,where complex interaction patterns prevail,MAMGBR utilizes auxiliary loss functions to effectively model the multifaceted roles of users,items,and participants,leading to a 24.08%increase in MRR@100 under a 1:99 sample ratio.Experimental results show that compared to benchmark models,such as NGCF and EATNN,MAMGBR’s integration ofmulti-head attentionmechanisms,expert networks,and gating mechanisms enables more accurate modeling of user preferences and social associations within group-buying scenarios,significantly enhancing recommendation accuracy and platform group-buying success rates.展开更多
Accurate multi-energy load forecasting is a prerequisite for on-demand energy supply in integrated energy systems.However,due to differences in response characteristics and load patterns among electrical,heating,and c...Accurate multi-energy load forecasting is a prerequisite for on-demand energy supply in integrated energy systems.However,due to differences in response characteristics and load patterns among electrical,heating,and cooling loads,multi-energy load forecasting faces the challenges of heterogeneous time scales and imbalanced forecasting accuracy across load types.To address these challenges,this paper proposes a multi-task learning model with stacked cross-attention.This model incorporates a time scale alignment module to align the time scales of different loads,and employs Informer encoders as experts to extract load-specific features.Stacked cross-attention as the soft sharing mechanism dynamically fuses expert features at the sequence level,enhancing inter-task collaboration and adaptability.This design improves the overall accuracy of multi-energy load fore-casting with mixed time scales while reducing forecasting imbalance across load types.Comparison results demonstrate that the model with the stacked cross-attention achieves the best forecasting performance and lowers the imbalance index by 79.17%.Furthermore,the experts based on Informer encoders yield over a 30.09%MAPE reduction compared to alternative expert architectures.Compared to the multi-gate mixture-of-experts based models,classical forecasting models,and recent advanced models,the proposed model achieves superior forecasting accuracy,validating its effectiveness and advancement.展开更多
Deep neural networks(DNNs)are effective in solving both forward and inverse problems for nonlinear partial differential equations(PDEs).However,conventional DNNs are not effective in handling problems such as delay di...Deep neural networks(DNNs)are effective in solving both forward and inverse problems for nonlinear partial differential equations(PDEs).However,conventional DNNs are not effective in handling problems such as delay differential equations(DDEs)and delay integrodifferential equations(DIDEs)with constant delays,primarily due to their low regularity at delayinduced breaking points.In this paper,a DNN method that combines multi-task learning(MTL)which is proposed to solve both the forward and inverse problems of DIDEs.The core idea of this approach is to divide the original equation into multiple tasks based on the delay,using auxiliary outputs to represent the integral terms,followed by the use of MTL to seamlessly incorporate the properties at the breaking points into the loss function.Furthermore,given the increased training dificulty associated with multiple tasks and outputs,we employ a sequential training scheme to reduce training complexity and provide reference solutions for subsequent tasks.This approach significantly enhances the approximation accuracy of solving DIDEs with DNNs,as demonstrated by comparisons with traditional DNN methods.We validate the effectiveness of this method through several numerical experiments,test various parameter sharing structures in MTL and compare the testing results of these structures.Finally,this method is implemented to solve the inverse problem of nonlinear DIDE and the results show that the unknown parameters of DIDE can be discovered with sparse or noisy data.展开更多
Cooperative multi-agent reinforcement learning(MARL)is a key technology for enabling cooperation in complex multi-agent systems.It has achieved remarkable progress in areas such as gaming,autonomous driving,and multi-...Cooperative multi-agent reinforcement learning(MARL)is a key technology for enabling cooperation in complex multi-agent systems.It has achieved remarkable progress in areas such as gaming,autonomous driving,and multi-robot control.Empowering cooperative MARL with multi-task decision-making capabilities is expected to further broaden its application scope.In multi-task scenarios,cooperative MARL algorithms need to address 3 types of multi-task problems:reward-related multi-task,arising from different reward functions;multi-domain multi-task,caused by differences in state and action spaces,state transition functions;and scalability-related multi-task,resulting from the dynamic variation in the number of agents.Most existing studies focus on scalability-related multitask problems.However,with the increasing integration between large language models(LLMs)and multi-agent systems,a growing number of LLM-based multi-agent systems have emerged,enabling more complex multi-task cooperation.This paper provides a comprehensive review of the latest advances in this field.By combining multi-task reinforcement learning with cooperative MARL,we categorize and analyze the 3 major types of multi-task problems under multi-agent settings,offering more fine-grained classifications and summarizing key insights for each.In addition,we summarize commonly used benchmarks and discuss future directions of research in this area,which hold promise for further enhancing the multi-task cooperation capabilities of multi-agent systems and expanding their practical applications in the real world.展开更多
Lithium-ion batteries(LIBs)are widely deployed,from grid-scale storage to electric vehicles.LIBs remain stationary most of their service life,where calendar aging degrades capacity.Understanding the mechanisms of LIB ...Lithium-ion batteries(LIBs)are widely deployed,from grid-scale storage to electric vehicles.LIBs remain stationary most of their service life,where calendar aging degrades capacity.Understanding the mechanisms of LIB calendar aging is crucial for extending battery lifespan.However,LIB calendar aging is influenced by multiple factors,including battery material,its state,and storage environment.Calendar aging experiments are also time-consuming,costly,and lack standardized testing conditions.This study employs a data-driven approach to establish a cross-scale database linking materials,side-reaction mechanisms,and calendar aging of LIBs.MELODI(Mechanism-informed,Explainable,Learning-based Optimization for Degradation Identification)is proposed to identify calendar aging mechanisms and quantify the effects of multi-scale factors.Results reveal that cathode material loss drives up to 91.42%of calendar aging degradation in high-nickel(Ni)batteries,while solid electrolyte interphase growth dominates in lithium iron phosphate(LFP)and low-Ni batteries,contributing up to 82.43%of degradation in LFP batteries and 99.10%of decay in low-Ni batteries,respectively.This study systematically quantifies calendar aging in commercial LIBs under varying materials,states of charge,and temperatures.These findings offer quantitative guidance for experimental design or battery use,and implications for emerging applications like aerial robotics,vehicle-to-grid,and embodied intelligence systems.展开更多
基金supported in part by the National Natural Science Foundation of China under Grant 62371181in part by the Changzhou Science and Technology International Cooperation Program under Grant CZ20230029The Institute of Information&Communications Technology Planning&Evaluation(IITP)grant funded by the Korea government(MSIT)(No.RS-2024-00396797,Development of core technology for intelligent O-RAN security platform).
文摘Satellite communication technology has emerged as a key solution to address the challenges of data transmission in remote areas.By overcoming the limitations of traditional terrestrial communication networks,it enables long-distance data transmission anytime and anywhere,ensuring the timely and accurate delivery of water level data,which is particularly crucial for fishway water level monitoring.To enhance the effectiveness of fishway water level monitoring,this study proposes a multi-task learning model,AS-SOMTF,designed for real-time and comprehensive prediction.The model integrates auxiliary sequences with primary input sequences to capture complex relationships and dependencies,thereby improving representational capacity.In addition,a novel timeseries embedding algorithm,AS-SOM,is introduced,which combines generative inference and pooling operations to optimize prediction efficiency for long sequences.This innovation not only ensures the timely transmission of water level data but also enhances the accuracy of real-time monitoring.Compared with traditional models such as Transformer and Long Short-Term Memory(LSTM)networks,the proposed model achieves improvements of 3.8%and 1.4%in prediction accuracy,respectively.These advancements provide more precise technical support for water level forecasting and resource management in the Diqing Tibetan Autonomous Prefecture of the Lancang River,contributing to ecosystem protection and improved operational safety.
基金supported by the Science and Technology Grant No.520120210003,Jibei Electric Power Company of the State Grid Corporation of China。
文摘Convective storms and lightning are among the most important weather phenomena that are challenging to forecast.In this study,a novel multi-task learning(MTL)encoder-decoder U-net neural network was developed to forecast convective storms and lightning with lead times for up to 90 min,using GOES-16 geostationary satellite infrared brightness temperatures(IRBTs),lightning flashes from Geostationary Lightning Mapper(GLM),and vertically integrated liquid(VIL)from Next Generation Weather Radar(NEXRAD).To cope with the heavily skewed distribution of lightning data,a spatiotemporal exponent-weighted loss function and log-transformed lightning normalization approach were developed.The effects of MTL,single-task learning(STL),and IRBTs as auxiliary input features on convection and lightning nowcasting were investigated.The results showed that normalizing the heavily skew-distributed lightning data along with a log-transformation dramatically outperforms the min-max normalization method for nowcasting an intense lightning event.The MTL model significantly outperformed the STL model for both lightning nowcasting and VIL nowcasting,particularly for intense lightning events.The MTL also helped delay the lightning forecast performance decay with the lead times.Furthermore,incorporating satellite IRBTs as auxiliary input features substantially improved lightning nowcasting,but produced little difference in VIL forecasting.Finally,the MTL model performed better for forecasting both lightning and the VIL of organized convective storms than for isolated cells.
文摘Aspect-based sentiment analysis(ABSA)is a fine-grained process.Its fundamental subtasks are aspect termextraction(ATE)and aspect polarity classification(APC),and these subtasks are dependent and closely related.However,most existing works on Arabic ABSA content separately address them,assume that aspect terms are preidentified,or use a pipeline model.Pipeline solutions design different models for each task,and the output from the ATE model is used as the input to the APC model,which may result in error propagation among different steps because APC is affected by ATE error.These methods are impractical for real-world scenarios where the ATE task is the base task for APC,and its result impacts the accuracy of APC.Thus,in this study,we focused on a multi-task learning model for Arabic ATE and APC in which the model is jointly trained on two subtasks simultaneously in a singlemodel.This paper integrates themulti-task model,namely Local Cotext Foucse-Aspect Term Extraction and Polarity classification(LCF-ATEPC)and Arabic Bidirectional Encoder Representation from Transformers(AraBERT)as a shred layer for Arabic contextual text representation.The LCF-ATEPC model is based on a multi-head selfattention and local context focus mechanism(LCF)to capture the interactive information between an aspect and its context.Moreover,data augmentation techniques are proposed based on state-of-the-art augmentation techniques(word embedding substitution with constraints and contextual embedding(AraBERT))to increase the diversity of the training dataset.This paper examined the effect of data augmentation on the multi-task model for Arabic ABSA.Extensive experiments were conducted on the original and combined datasets(merging the original and augmented datasets).Experimental results demonstrate that the proposed Multi-task model outperformed existing APC techniques.Superior results were obtained by AraBERT and LCF-ATEPC with fusion layer(AR-LCF-ATEPC-Fusion)and the proposed data augmentation word embedding-based method(FastText)on the combined dataset.
文摘Knowledge distillation has become a standard technique for compressing large language models into efficient student models,but existing methods often struggle to balance prediction accuracy with explanation quality.Recent approaches such as Distilling Step-by-Step(DSbS)introduce explanation supervision,yet they apply it in a uniform manner that may not fully exploit the different learning dynamics of prediction and explanation.In this work,we propose a task-structured curriculum learning(TSCL)framework that structures training into three sequential phases:(i)prediction-only,to establish stable feature representations;(ii)joint prediction-explanation,to align task outputs with rationale generation;and(iii)explanation-only,to refine the quality of rationales.This design provides a simple but effective modification to DSbS,requiring no architectural changes and adding negligible training cost.We justify the phase scheduling with ablation studies and convergence analysis,showing that an initial prediction-heavy stage followed by a balanced joint phase improves both stability and explanation alignment.Extensive experiments on five datasets(e-SNLI,ANLI,CommonsenseQA,SVAMP,and MedNLI)demonstrate that TSCL consistently outperforms strong baselines,achieving gains of+1.7-2.6 points in accuracy and 0.8-1.2 in ROUGE-L,corresponding to relative error reductions of up to 21%.Beyond lexical metrics,human evaluation and ERASERstyle faithfulness diagnostics confirm that TSCL produces more faithful and informative explanations.Comparative training curves further reveal faster convergence and lower variance across seeds.Efficiency analysis shows less than 3%overhead in wall-clock training time and no additional inference cost,making the approach practical for realworld deployment.This study demonstrates that a simple task-structured curriculum can significantly improve the effectiveness of knowledge distillation.By separating and sequencing objectives,TSCL achieves a better balance between accuracy,stability,and explanation quality.The framework generalizes across domains,including medical NLI,and offers a principled recipe for future applications in multimodal reasoning and reinforcement learning.
基金supported by the National Natural Science Foundation of China(Grant Nos.42130719 and 42177173)the Doctoral Direct Train Project of Chongqing Natural Science Foundation(Grant No.CSTB2023NSCQ-BSX0029).
文摘Underground engineering projects such as deep tunnel excavation often encounter rockburst disasters accompanied by numerous microseismic events.Rapid interpretation of microseismic signals is crucial for the timely identification of rockbursts.However,conventional processing encompasses multi-step workflows,including classification,denoising,picking,locating,and computational analysis,coupled with manual intervention,which collectively compromise the reliability of early warnings.To address these challenges,this study innovatively proposes the“microseismic stethoscope"-a multi-task machine learning and deep learning model designed for the automated processing of massive microseismic signals.This model efficiently extracts three key parameters that are necessary for recognizing rockburst disasters:rupture location,microseismic energy,and moment magnitude.Specifically,the model extracts raw waveform features from three dedicated sub-networks:a classifier for source zone classification,and two regressors for microseismic energy and moment magnitude estimation.This model demonstrates superior efficiency compared to traditional processing and semi-automated processing,reducing per-event processing time from 0.71 s to 0.49 s to merely 0.036 s.It concurrently achieves 98%accuracy in source zone classification,with microseismic energy and moment magnitude estimation errors of 0.13 and 0.05,respectively.This model has been well applied and validated in the Daxiagu Tunnel case in Sichuan,China.The application results indicate that the model is as accurate as traditional methods in determining source parameters,and thus can be used to identify potential geomechanical processes of rockburst disasters.By enhancing the signal processing reliability of microseismic events,the proposed model in this study presents a significant advancement in the identification of rockburst disasters.
文摘Federated Learning(FL)enables joint training over distributed devices without data exchange but is highly vulnerable to attacks by adversaries in the form of model poisoning and malicious update injection.This work proposes Secured-FL,a blockchain-based defensive framework that combines smart contract-based authentication,clustering-driven outlier elimination,and dynamic threshold adjustment to defend against adversarial attacks.The framework was implemented on a private Ethereum network with a Proof-of-Authority consensus algorithm to ensure tamper-resistant and auditable model updates.Large-scale simulation on the Cyber Data dataset,under up to 50%malicious client settings,demonstrates Secured-FL achieves 6%-12%higher accuracy,9%-15%lower latency,and approximately 14%less computational expense compared to the PPSS benchmark framework.Additional tests,including confusion matrices,ROC and Precision-Recall curves,and ablation tests,confirm the interpretability and robustness of the defense.Tests for scalability also show consistent performance up to 500 clients,affirming appropriateness to reasonably large deployments.These results make Secured-FL a feasible,adversarially resilient FL paradigm with promising potential for application in smart cities,medicine,and other mission-critical IoT deployments.
基金supported by the Advanced Materials-National Science and Technology Major Project(Grant No.2025ZD0618401)the National Natural Science Foundation of China(Grant No.12504285)+1 种基金the Natural Science Foundation of Jiangsu Province(Grant No.BK20250472)NFSG grant from BITS-Pilani,Dubai campus。
文摘The rapid advancement of machine learning based tight-binding Hamiltonian(MLTB)methods has opened new avenues for efficient and accurate electronic structure simulations,particularly in large-scale systems and long-time scenarios.This review begins with a concise overview of traditional tight-binding(TB)models,including both(semi-)empirical and first-principles approaches,establishing the foundation for understanding MLTB developments.We then present a systematic classification of existing MLTB methodologies,grouped into two major categories:direct prediction of TB Hamiltonian elements and inference of empirical parameters.A comparative analysis with other ML-based electronic structure models is also provided,highlighting the advancement of MLTB approaches.Finally,we explore the emerging MLTB application ecosystem,highlighting how the integration of MLTB models with a diverse suite of post-processing tools from linear-scaling solvers to quantum transport frameworks and molecular dynamics interfaces is essential for tackling complex scientific problems across different domains.The continued advancement of this integrated paradigm promises to accelerate materials discovery and open new frontiers in the predictive simulation of complex quantum phenomena.
基金supported by the special fund of the National Clinical Key Specialty Construction Program[(2022)301-2305].
文摘BACKGROUND:This study aims to develop and validate a machine learning-based in-hospital mortality predictive model for acute aortic syndrome(AAS)in the emergency department(ED)and to derive a simplifi ed version suitable for rapid clinical application.METHODS:In this multi-center retrospective cohort study,AAS patient data from three hospitals were analyzed.The modeling cohort included data from the First Affiliated Hospital of Zhengzhou University and the People’s Hospital of Xinjiang Uygur Autonomous Region,with Peking University Third Hospital data serving as the external test set.Four machine learning algorithms—logistic regression(LR),multilayer perceptron(MLP),Gaussian naive Bayes(GNB),and random forest(RF)—were used to develop predictive models based on 34 early-accessible clinical variables.A simplifi ed model was then derived based on fi ve key variables(Stanford type,pericardial eff usion,asymmetric peripheral arterial pulsation,decreased bowel sounds,and dyspnea)via Least Absolute Shrinkage and Selection Operator(LASSO)regression to improve ED applicability.RESULTS:A total of 929 patients were included in the modeling cohort,and 210 were included in the external test set.Four machine learning models based on 34 clinical variables were developed,achieving internal and external validation AUCs of 0.85-0.90 and 0.73-0.85,respectively.The simplifi ed model incorporating fi ve key variables demonstrated internal and external validation AUCs of 0.71-0.86 and 0.75-0.78,respectively.Both models showed robust calibration and predictive stability across datasets.CONCLUSION:Both kinds of models were built based on machine learning tools,and proved to have certain prediction performance and extrapolation.
文摘This study aims to develop an accurate and robust machine learning model to predict the carbonation depth of fly ash concrete,overcoming the limitations of traditional predictive methods.Five ensemble-based models,such as adaptive boosting(AdaBoost),categorical boosting(CatBoost),gradient boosting regressor(GBR),hist gradient boosting regressor(HistGBR),and extreme gradient boosting(XGBoost),were developed and optimized using 729 high-quality dataset points incorporating seven input parameters,including cement,CO_(2),exposure time,water-binder ratio,fly ash,curing time,and compressive strength.Several performance evaluation metrics were used to compare the models.The GBR model emerged as the best-performing model,based on high coefficient of determination(R^(2))values and balanced error metrics across both validation and testing datasets.While all models performed exceptionally well on the training data,GBR demonstrated superior generalization capability,with R^(2) values of 0.9438 on the validation set and 0.9310 on the testing set.Furthermore,its low mean squared error(MSE),root mean square error(RMSE),mean absolute error(MAE),and median absolute error(MdAE)confirmed its robustness and accuracy.Moreover,shapley additive explanations(SHAP)analysis enhanced the interpretability of predictions,highlighting the curing time and exposure time as the most critical drivers of carbonation depth.
基金financial support of the National Natural Science Foundation of China(No.52371103)the Fundamental Research Funds for the Central Universities,China(No.2242023K40028)+1 种基金the Open Research Fund of Jiangsu Key Laboratory for Advanced Metallic Materials,China(No.AMM2023B01).financial support of the Research Fund of Shihezi Key Laboratory of AluminumBased Advanced Materials,China(No.2023PT02)financial support of Guangdong Province Science and Technology Major Project,China(No.2021B0301030005)。
文摘Oxide dispersion strengthened(ODS)alloys are extensively used owing to high thermostability and creep strength contributed from uniformly dispersed fine oxides particles.However,the existence of these strengthening particles also deteriorates the processability and it is of great importance to establish accurate processing maps to guide the thermomechanical processes to enhance the formability.In this study,we performed particle swarm optimization-based back propagation artificial neural network model to predict the high temperature flow behavior of 0.25wt%Al2O3 particle-reinforced Cu alloys,and compared the accuracy with that of derived by Arrhenius-type constitutive model and back propagation artificial neural network model.To train these models,we obtained the raw data by fabricating ODS Cu alloys using the internal oxidation and reduction method,and conducting systematic hot compression tests between 400 and800℃with strain rates of 10^(-2)-10 S^(-1).At last,processing maps for ODS Cu alloys were proposed by combining processing parameters,mechanical behavior,microstructure characterization,and the modeling results achieved a coefficient of determination higher than>99%.
基金funding from the European Union’s Horizon 2020 Research and Innovation Programme under the Marie Skłodowska-Curie Grant Agreement No.945478(SASPRO2)supported by the ReBuilt project:Circular and Digital Renewal of Central Europe Construction and Building Sector CE0100390 ReBuiltthe Slovak Research and Development Agency under APVV-23-0383 and the Slovak Grant Agency VEGA No.2/0080/24.
文摘This study presents a physics-informed modelling framework that combines finite element method(FEM)simulations and supervised machine learning(ML)to predict the self-healing performance of microbial concrete.A FEniCS-based FEM platform resolves multiphysics phenomena including nutrient diffusion,microbial CaCO_(3) precipitation,and stiffness recovery.These simulations,together with experimental data,are used to train ML models(Random Forest yielding normalized RMSE≈0.10)capable of predicting performance over a wide range of design parameters.Feature importance analysis identifies curing temperature,calcium carbonate precipitation rate,crack width,bacterial strain,and encapsulation method as the most influential parameters.The coupled FEM-ML approach enables sensitivity analysis,design optimization,and prediction beyond the training dataset(consistently exceeding 90%healing efficiency).Experimental validation confirms model robustness in both crack closure and strength recovery.This FEM–ML pipeline thus offers a generalizable,interpretable,and scalable strategy for the design of intelligent,self-adaptive construction materials.
基金the research project LaTe4PoliticES(PID2022-138099OB-I00)funded by MCIN/AEI/10.13039/501100011033 and the European Fund for Regional Development(ERDF)-a way to make Europe.Tomás Bernal-Beltrán is supported by University of Murcia through the predoctoral programme.
文摘The malicious dissemination of hate speech via compromised accounts,automated bot networks and malware-driven social media campaigns has become a growing cybersecurity concern.Automatically detecting such content in Spanish is challenging due to linguistic complexity and the scarcity of annotated resources.In this paper,we compare two predominant AI-based approaches for the forensic detection of malicious hate speech:(1)finetuning encoder-only models that have been trained in Spanish and(2)In-Context Learning techniques(Zero-and Few-Shot Learning)with large-scale language models.Our approach goes beyond binary classification,proposing a comprehensive,multidimensional evaluation that labels each text by:(1)type of speech,(2)recipient,(3)level of intensity(ordinal)and(4)targeted group(multi-label).Performance is evaluated using an annotated Spanish corpus,standard metrics such as precision,recall and F1-score and stability-oriented metrics to evaluate the stability of the transition from zero-shot to few-shot prompting(Zero-to-Few Shot Retention and Zero-to-Few Shot Gain)are applied.The results indicate that fine-tuned encoder-only models(notably MarIA and BETO variants)consistently deliver the strongest and most reliable performance:in our experiments their macro F1-scores lie roughly in the range of approximately 46%–66%depending on the task.Zero-shot approaches are much less stable and typically yield substantially lower performance(observed F1-scores range approximately 0%–39%),often producing invalid outputs in practice.Few-shot prompting(e.g.,Qwen 38B,Mistral 7B)generally improves stability and recall relative to pure zero-shot,bringing F1-scores into a moderate range of approximately 20%–51%but still falling short of fully fine-tuned models.These findings highlight the importance of supervised adaptation and discuss the potential of both paradigms as components in AI-powered cybersecurity and malware forensics systems designed to identify and mitigate coordinated online hate campaigns.
文摘Modern industrial environments require uninterrupted machinery operation to maintain productivity standards while ensuring safety and minimizing costs.Conventional maintenance methods,such as reactive maintenance(i.e.,run to failure)or time-based preventive maintenance(i.e.,scheduled servicing),prove ineffective for complex systems with many Internet of Things(IoT)devices and sensors because they fall short in detecting faults at early stages when it is most crucial.This paper presents a predictive maintenance framework based on a hybrid deep learning model that integrates the capabilities of Long Short-Term Memory(LSTM)Networks and Convolutional Neural Networks(CNNs).The framework integrates spatial feature extraction and temporal sequence modeling to accurately classify the health state of industrial equipment into three categories,including Normal,Require Maintenance,and Failed.The framework uses a modular pipeline that includes IoT-enabled data collection along with secure transmission methods to manage cloud storage and provide real-time fault classification.The FD004 subset of the NASA C-MAPSS dataset,containing multivariate sensor readings from aircraft engines,serves as the training and evaluation data for the model.Experimental results show that the LSTM-CNN model outperforms baseline models such as LSTM-SVM and LSTM-RNN,achieving an overall average accuracy of 86.66%,precision of 86.00%,recall of 86.33%,and F1-score of 86.33%.Contrary to the previous LSTM-CNN-based predictive maintenance models that either provide a binary classification or rely on synthetically balanced data,our paper provides a three-class maintenance state(i.e.,Normal,Require Maintenance,and Failed)along with threshold-based labeling that retains the true nature of the degradation.In addition,our work also provides an IoT-to-cloud-based modular architecture for deployment.It offers Computerized Maintenance Management System(CMMS)integration,making our proposed solution not only technically sound but also practical and innovative.The solution achieves real-world industrial deployment readiness through its reliable performance alongside its scalable system design.
基金the National Key Research and Development Plan of China[Grant No.2023YFB3002400]the Shanghai 2021 Natural Science Foundation[Grant Nos.21ZR1420400 and 21ZR1419800]+1 种基金the Shanghai 2023 Natural Science Foundation[Grant No.23ZR1463000]the Shandong Provincial Meteorological Bureau Scientific Research Project[Grant No.2023SDBD05].
文摘Accurate short-term forecast of offshore wind fields is still challenging for numerical weather prediction models.Based on three years of 48-hour forecast data from the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System global model(ECMWF-IFS)over 14 offshore weather stations along the coast of Shandong Province,this study introduces a multi-task learning(MTL)model(TabNet-MTL),which significantly improves the forecast bias of near-surface wind direction and speed simultaneously.TabNet-MTL adopts the feature engineering method,utilizes mean square error as the loss function,and employs the 5-fold cross validation method to ensure the generalization ability of the trained model.It demonstrates superior skills in wind field correction across different forecast lead times over all stations compared to its single-task version(TabNet-STL)and three other popular single-task learning models(Random Forest,LightGBM,and XGBoost).Results show that it significantly reduces root mean square error of the ECMWF-IFS wind speed forecast from 2.20 to 1.25 m s−1,and increases the forecast accuracy of wind direction from 50%to 65%.As an explainable deep learning model,the weather stations and long-term temporal statistics of near-surface wind speed are identified as the most influential variables for TabNet-MTL in constructing its feature engineering.
基金supported by the Key Research and Development Program of Heilongjiang Province(No.2022ZX01A35).
文摘As the group-buying model shows significant progress in attracting new users,enhancing user engagement,and increasing platform profitability,providing personalized recommendations for group-buying users has emerged as a new challenge in the field of recommendation systems.This paper introduces a group-buying recommendation model based on multi-head attention mechanisms and multi-task learning,termed the Multi-head Attention Mechanisms and Multi-task Learning Group-Buying Recommendation(MAMGBR)model,specifically designed to optimize group-buying recommendations on e-commerce platforms.The core dataset of this study comes from the Chinese maternal and infant e-commerce platform“Beibei,”encompassing approximately 430,000 successful groupbuying actions and over 120,000 users.Themodel focuses on twomain tasks:recommending items for group organizers(Task Ⅰ)and recommending participants for a given group-buying event(Task Ⅱ).In model evaluation,MAMGBR achieves an MRR@10 of 0.7696 for Task I,marking a 20.23%improvement over baseline models.Furthermore,in Task II,where complex interaction patterns prevail,MAMGBR utilizes auxiliary loss functions to effectively model the multifaceted roles of users,items,and participants,leading to a 24.08%increase in MRR@100 under a 1:99 sample ratio.Experimental results show that compared to benchmark models,such as NGCF and EATNN,MAMGBR’s integration ofmulti-head attentionmechanisms,expert networks,and gating mechanisms enables more accurate modeling of user preferences and social associations within group-buying scenarios,significantly enhancing recommendation accuracy and platform group-buying success rates.
基金supported by National Natural Science Foundation of China(Grant Nos.U22B20112 and 52476001)Beijing Institute of Technology Research Fund Program for Young Scholars.
文摘Accurate multi-energy load forecasting is a prerequisite for on-demand energy supply in integrated energy systems.However,due to differences in response characteristics and load patterns among electrical,heating,and cooling loads,multi-energy load forecasting faces the challenges of heterogeneous time scales and imbalanced forecasting accuracy across load types.To address these challenges,this paper proposes a multi-task learning model with stacked cross-attention.This model incorporates a time scale alignment module to align the time scales of different loads,and employs Informer encoders as experts to extract load-specific features.Stacked cross-attention as the soft sharing mechanism dynamically fuses expert features at the sequence level,enhancing inter-task collaboration and adaptability.This design improves the overall accuracy of multi-energy load fore-casting with mixed time scales while reducing forecasting imbalance across load types.Comparison results demonstrate that the model with the stacked cross-attention achieves the best forecasting performance and lowers the imbalance index by 79.17%.Furthermore,the experts based on Informer encoders yield over a 30.09%MAPE reduction compared to alternative expert architectures.Compared to the multi-gate mixture-of-experts based models,classical forecasting models,and recent advanced models,the proposed model achieves superior forecasting accuracy,validating its effectiveness and advancement.
文摘Deep neural networks(DNNs)are effective in solving both forward and inverse problems for nonlinear partial differential equations(PDEs).However,conventional DNNs are not effective in handling problems such as delay differential equations(DDEs)and delay integrodifferential equations(DIDEs)with constant delays,primarily due to their low regularity at delayinduced breaking points.In this paper,a DNN method that combines multi-task learning(MTL)which is proposed to solve both the forward and inverse problems of DIDEs.The core idea of this approach is to divide the original equation into multiple tasks based on the delay,using auxiliary outputs to represent the integral terms,followed by the use of MTL to seamlessly incorporate the properties at the breaking points into the loss function.Furthermore,given the increased training dificulty associated with multiple tasks and outputs,we employ a sequential training scheme to reduce training complexity and provide reference solutions for subsequent tasks.This approach significantly enhances the approximation accuracy of solving DIDEs with DNNs,as demonstrated by comparisons with traditional DNN methods.We validate the effectiveness of this method through several numerical experiments,test various parameter sharing structures in MTL and compare the testing results of these structures.Finally,this method is implemented to solve the inverse problem of nonlinear DIDE and the results show that the unknown parameters of DIDE can be discovered with sparse or noisy data.
基金The National Natural Science Foundation of China(62136008,62293541)The Beijing Natural Science Foundation(4232056)The Beijing Nova Program(20240484514).
文摘Cooperative multi-agent reinforcement learning(MARL)is a key technology for enabling cooperation in complex multi-agent systems.It has achieved remarkable progress in areas such as gaming,autonomous driving,and multi-robot control.Empowering cooperative MARL with multi-task decision-making capabilities is expected to further broaden its application scope.In multi-task scenarios,cooperative MARL algorithms need to address 3 types of multi-task problems:reward-related multi-task,arising from different reward functions;multi-domain multi-task,caused by differences in state and action spaces,state transition functions;and scalability-related multi-task,resulting from the dynamic variation in the number of agents.Most existing studies focus on scalability-related multitask problems.However,with the increasing integration between large language models(LLMs)and multi-agent systems,a growing number of LLM-based multi-agent systems have emerged,enabling more complex multi-task cooperation.This paper provides a comprehensive review of the latest advances in this field.By combining multi-task reinforcement learning with cooperative MARL,we categorize and analyze the 3 major types of multi-task problems under multi-agent settings,offering more fine-grained classifications and summarizing key insights for each.In addition,we summarize commonly used benchmarks and discuss future directions of research in this area,which hold promise for further enhancing the multi-task cooperation capabilities of multi-agent systems and expanding their practical applications in the real world.
基金supported by the National Key Research and Development Program of China(2024YFE0213000)the Postdoctoral Innovative Talents Support Program(BX20240232)+1 种基金the Natural Science Foundation of China for Young Scholars(72304031)the Fundamental Research Funds for the Central Universities(FRF-TP-22-024A1).
文摘Lithium-ion batteries(LIBs)are widely deployed,from grid-scale storage to electric vehicles.LIBs remain stationary most of their service life,where calendar aging degrades capacity.Understanding the mechanisms of LIB calendar aging is crucial for extending battery lifespan.However,LIB calendar aging is influenced by multiple factors,including battery material,its state,and storage environment.Calendar aging experiments are also time-consuming,costly,and lack standardized testing conditions.This study employs a data-driven approach to establish a cross-scale database linking materials,side-reaction mechanisms,and calendar aging of LIBs.MELODI(Mechanism-informed,Explainable,Learning-based Optimization for Degradation Identification)is proposed to identify calendar aging mechanisms and quantify the effects of multi-scale factors.Results reveal that cathode material loss drives up to 91.42%of calendar aging degradation in high-nickel(Ni)batteries,while solid electrolyte interphase growth dominates in lithium iron phosphate(LFP)and low-Ni batteries,contributing up to 82.43%of degradation in LFP batteries and 99.10%of decay in low-Ni batteries,respectively.This study systematically quantifies calendar aging in commercial LIBs under varying materials,states of charge,and temperatures.These findings offer quantitative guidance for experimental design or battery use,and implications for emerging applications like aerial robotics,vehicle-to-grid,and embodied intelligence systems.
