This paper provides a comprehensive overview of the latest stable release of the graphics processing units molecular dynamics(GPUMD)package,GPUMD 4.0.We begin with a brief review of its development history,starting fr...This paper provides a comprehensive overview of the latest stable release of the graphics processing units molecular dynamics(GPUMD)package,GPUMD 4.0.We begin with a brief review of its development history,starting from the initial version.We then discuss the theoretical foundations for the development of the GPUMD package,including the formulations of the interatomic force,virial and heat current for many-body potentials,the development of the highly efficient and flexible neuroevolution potential(NEP)method,the supported integrators and related operations,the various physical properties that can be calculated on the fly,and the GPUMD ecosystem.After presenting these functionalities,we review a range of applications enabled by GPUMD,particularly in combination with the NEP approach.Finally,we outline possible future development directions for GPUMD.展开更多
Ion transport in materials is routinely probed through several experimental techniques,which introduce variability in reported ionic diffusivities and conductivities.The computational prediction of ionic diffusivities...Ion transport in materials is routinely probed through several experimental techniques,which introduce variability in reported ionic diffusivities and conductivities.The computational prediction of ionic diffusivities and conductivities helps in identifying good ionic conductors,and suitable solid electrolytes(SEs),thus establishing firm structure-property relationships.Machine-learned potentials are an attractive strategy to extend the capabilities of accurate ab initio molecular dynamics(AIMD)to longer simulations for larger systems,enabling the study of ion transport at lower temperatures.However,machine-learned potentials being in their infancy,critical assessments of their predicting capabilities are rare.Here,we identified the main factors controlling the quality of a machine-learning potential based on the moment tensor potential formulation,when applied to the properties of ion transport in ionic conductors,such as SEs.Our results underline the importance of high-quality and diverse training sets required to fit moment tensor potentials.We highlight the importance of considering intrinsic defects which may occur in SEs.We demonstrate the limitations posed by short-timescale and high-temperature AIMD simulations to predict the room-temperature properties of materials.展开更多
Objective:To identify potential key genes associated with pre-eclampsia through bioinformatics analysis,construct predictive models using machine-learning algorithms,and evaluate the models'performance in predicti...Objective:To identify potential key genes associated with pre-eclampsia through bioinformatics analysis,construct predictive models using machine-learning algorithms,and evaluate the models'performance in predicting pre-eclampsia.Methods:Gene-expression microarray datasets GSE10588,GSE66273,and GSE30186 related to pre-eclampsia were downloaded from the gene expression omnibus(GEO).Data were normalized using R,and differentially expressed genes(DEGs)were identified.LASSO regression was applied to further filter DEGs.Based on the selected DEGs,six machine-learning models-logistic regression(LR),random forest(RF),support vector machine(SVM),K-nearest neighbors(KNN),neural network(NN),and eXtreme gradient boosting(XGBoost)were built in R,and their performance was validated.Results:From the three datasets,a total of 1,363 genes were extracted.LASSO regression narrowed these to 265 candidate key genes.Multivariate analysis ultimately identified four genes closely associated with pre-eclampsia:EVI5,GCLM,LEP,and SYNPO2L.Using these four key genes,six machine-learning models were constructed.Receiver operating characteristic(ROC)analysis showed that all models achieved AUC>0.9:LR(AUC=0.983,95%CI=0.942-0.998),RF(AUC=0.961,95%CI=0.912-0.987),SVM(AUC=0.936,95%CI=0.879-0.972),KNN(AUC=0.970,95%CI=0.924-0.992),NN(AUC=0.916,95%CI=0.854-0.958),and XGBoost(AUC=0.952,95%CI=0.900-0.982).There was no statistically significant difference among the AUCs of the models(P>0.05).Conclusion:This study identified four key genes linked to preeclampsia through integrated bioinformatics analysis.Predictive models built on these genes can accurately forecast the occurrence of pre-eclampsia,suggesting that the four genes may serve as potential biomarkers for early diagnosis and therapeutic targeting of pre-eclampsia.展开更多
All-solid-state batteries(ASSBs)represent a next-generation energy storage technology,offering enhanced safety,higher energy density,and improved cycling stability compared to conventional liquid-electrolyte-based lit...All-solid-state batteries(ASSBs)represent a next-generation energy storage technology,offering enhanced safety,higher energy density,and improved cycling stability compared to conventional liquid-electrolyte-based lithium-ion batteries.Understanding and optimizing the complex chemistries and interfaces that underpin ASSB performance present significant challenges from both experimental and modeling perspectives.In particular,atomistic simulations face difficulties in capturing the complex structure,disorder,and dynamic evolution of materials and interfaces under practically relevant conditions.While established methods such as density functional theory and classical force fields have provided valuable insights,some questions remain difficult to address,particularly those involving large system sizes or long timescales.Recently,machine learning interatomic potentials(MLIPs)have emerged as a transformative tool,enabling atomistic simulations at length and time scales that were previously challenging to access with conventional approaches.By delivering near first-principles accuracy with much greater efficiency,MLIPs open new avenues for large-scale,long-timescale,and high-throughput simulations of solid-state battery materials.In this review,we present a comparative overview of density functional theory,classical force fields,and MLIPs,highlighting their respective strengths and limitations in ASSB research.We then discuss how MLIPs enable simulations that reach longer timescales,larger system sizes,and support high-throughput calculations,providing unique insights into ion transport and interfacial evolution in ASSBs.Finally,we conclude with a summary and outlook on current challenges and future opportunities for expanding MLIP capabilities and accelerating their impact in solid-state battery research.展开更多
Objective: Postoperative complications adversely affected the prognosis in patients with gastric cancer. This study intends to investigate the feasibility of using machine-learning model to predict surgical outcomes i...Objective: Postoperative complications adversely affected the prognosis in patients with gastric cancer. This study intends to investigate the feasibility of using machine-learning model to predict surgical outcomes in patients undergoing gastrectomy.Methods: In this study, cancer patients who underwent gastrectomy at Shanghai Rui Jin Hospital in 2017 were randomly assigned to a development or validation cohort in a 9:1 ratio. A support vector classification(SVC) model to predict surgical outcomes in patients undergoing gastrectomy was developed and further validated.Results: A total of 321 patients with 32 features were collected. The positive and negative outcomes of postoperative complication after gastrectomy appeared in 100(31.2%) and 221(68.8%) patients, respectively. The SVC model was constructed to predict surgical outcomes in patients undergoing gastrectomy. The accuracy of 10-fold cross validation and external verification was 78.17% and 78.12%, respectively. Further, an online web server has been developed to share the SVC model for machine-learning-assisted prediction of surgical outcomes in patients undergoing gastrectomy in the future procedures, which is accessible at the web address:http://47.100.47.97:5005/r_model_prediction.Conclusions: The SVC model was a useful predictor for measuring the risk of postoperative complications after gastrectomy, which may help stratify patients with different overall status for choice of surgical procedure or other treatments. It can be expected that machine-learning models in cancer informatics research are possibly shareable and accessible via web address all over the world.展开更多
The coronavirus 3C-like(3CL)protease,a cysteine protease,plays an important role in viral infection and immune escape.However,there is still a lack of effective tools for determining the cleavage sites of the 3CL prot...The coronavirus 3C-like(3CL)protease,a cysteine protease,plays an important role in viral infection and immune escape.However,there is still a lack of effective tools for determining the cleavage sites of the 3CL protease.This study systematically investigated the diversity of the cleavage sites of the coronavirus 3CL protease on the viral polyprotein,and found that the cleavage motif were highly conserved for viruses in the genera of Alphacoronavirus,Betacoronavirus and Gammacoronavirus.Strong residue preferences were observed at the neighboring positions of the cleavage sites.A random forest(RF)model was built to predict the cleavage sites of the coronavirus 3CL protease based on the representation of residues in cleavage motifs by amino acid indexes,and the model achieved an AUC of 0.96 in cross-validations.The RF model was further tested on an independent test dataset which were composed of cleavage sites on 99 proteins from multiple coronavirus hosts.It achieved an AUC of 0.95 and predicted correctly 80%of the cleavage sites.Then,1,352 human proteins were predicted to be cleaved by the 3CL protease by the RF model.These proteins were enriched in several GO terms related to the cytoskeleton,such as the microtubule,actin and tubulin.Finally,a webserver named 3CLP was built to predict the cleavage sites of the coronavirus 3CL protease based on the RF model.Overall,the study provides an effective tool for identifying cleavage sites of the 3CL protease and provides insights into the molecular mechanism underlying the pathogenicity of coronaviruses.展开更多
Anomaly detection is becoming increasingly significant in industrial cyber security,and different machine-learning algorithms have been generally acknowledged as various effective intrusion detection engines to succes...Anomaly detection is becoming increasingly significant in industrial cyber security,and different machine-learning algorithms have been generally acknowledged as various effective intrusion detection engines to successfully identify cyber attacks.However,different machine-learning algorithms may exhibit their own detection effects even if they analyze the same feature samples.As a sequence,after developing one feature generation approach,the most effective and applicable detection engines should be desperately selected by comparing distinct properties of each machine-learning algorithm.Based on process control features generated by directed function transition diagrams,this paper introduces five different machine-learning algorithms as alternative detection engines to discuss their matching abilities.Furthermore,this paper not only describes some qualitative properties to compare their advantages and disadvantages,but also gives an in-depth and meticulous research on their detection accuracies and consuming time.In the verified experiments,two attack models and four different attack intensities are defined to facilitate all quantitative comparisons,and the impacts of detection accuracy caused by the feature parameter are also comparatively analyzed.All experimental results can clearly explain that SVM(Support Vector Machine)and WNN(Wavelet Neural Network)are suggested as two applicable detection engines under differing cases.展开更多
The composite exciter and the CaO to Na_(2)SO_(4) dosing ratios are known to have a strong impact on the mechanical strength offly-ash concrete.In the present study a hybrid approach relying on experiments and a machi...The composite exciter and the CaO to Na_(2)SO_(4) dosing ratios are known to have a strong impact on the mechanical strength offly-ash concrete.In the present study a hybrid approach relying on experiments and a machine-learn-ing technique has been used to tackle this problem.The tests have shown that the optimal admixture of CaO and Na_(2)SO_(4) alone is 8%.The best 3D mechanical strength offly-ash concrete is achieved at 8%of the compound activator;If the 28-day mechanical strength is considered,then,the best performances are obtained at 4%of the compound activator.Moreover,the 3D mechanical strength offly-ash concrete is better when the dosing ratio of CaO to Na_(2)SO_(4) in the compound activator is 1:1;the maximum strength offly-ash concrete at 28-day can be achieved for a 1:1 ratio of CaO to Na_(2)SO_(4) by considering a 4%compound activator.In this case,the compressive andflexural strengths are 260 MPa and 53.6 MPa,respectively;the mechanical strength offly-ash concrete at 28-day can be improved by a 4:1 ratio of CaO to Na_(2)SO_(4) by considering 8%and 12%compound excitants.It is shown that the predictions based on the aforementioned machine-learning approach are accurate and reliable.展开更多
Chronic kidney disease(CKD)is an increasingly prevalent medical condition associated with high mortality and cardiovascular complications.The intricate interplay between kidney dysfunction and subsequent metabolic dis...Chronic kidney disease(CKD)is an increasingly prevalent medical condition associated with high mortality and cardiovascular complications.The intricate interplay between kidney dysfunction and subsequent metabolic disturbances may provide insights into the underlying mechanisms driving CKD onset and progression.Herein,we proposed a large-scale plasma metabolite identification and quantification system that combines the strengths of targeted and untargeted metabolomics technologies,i.e.,widely-targeted metabolomics(WT-Met)approach.WT-Met method enables large-scale identification and accurate quantification of thousands of metabolites.We collected plasma samples from 21 healthy controls and 62CKD patients,categorized into different stages(22 in stages 1-3,20 in stage 4,and 20 in stage 5).Using LC-MS-based WT-Met approach,we were able to effectively annotate and quantify a total of 1431metabolites from the plasma samples.Focusing on the 539 endogenous metabolites,we identified 399significantly altered metabolites and depicted their changing patterns from healthy controls to end-stage CKD.Furthermore,we employed machine-learning to identify the optimal combination of metabolites for predicting different stages of CKD.We generated a multiclass classifier consisting of 7 metabolites by machine-learning,which exhibited an average AUC of 0.99 for the test set.In general,amino acids,nucleotides,organic acids,and their metabolites emerged as the most significantly altered metabolites.However,their patterns of change varied across different stages of CKD.The 7-metabolite panel demonstrates promising potential as biomarker candidates for CKD.Further exploration of these metabolites can provide valuable insights into their roles in the etiology and progression of CKD.展开更多
Drug addiction can cause abnormal brain activation changes,which are the root cause of drug craving and brain function errors.This study enrolled drug abusers to determine the effects of different drugs on brain activ...Drug addiction can cause abnormal brain activation changes,which are the root cause of drug craving and brain function errors.This study enrolled drug abusers to determine the effects of different drugs on brain activation.A functional near-infrared spectroscopy(fNIRS)device was used for the research.This study was designed with an experimental paradigm that included the induction of resting and drug addiction cravings.We collected the fNIRS data of 30 drug users,including 10 who used heroin,10 who used Methamphetamine,and 10 who used mixed drugs.First,using Statistical Analysis,the study analyzed the activations of eight functional areas of the left and right hemispheres of the prefrontal cortex of drug addicts who respectively used heroin,Methamphetamine,and mixed drugs,including Left/Right-Dorsolateral prefrontal cortex(L/R-DLPFC),Left/Right-Ventrolateral prefrontal cortex(L/R-VLPFC),Left/Right-Fronto-polar prefrontal cortex(L/R-FPC),and Left/Right Orbitofrontal Cortex(L/R-OFC).Second,referencing the degrees of activation of oxyhaemoglobin concentration(HbO2),the study made an analysis and got the specific activation patterns of each group of the addicts.Finally,after taking out data which are related to the addicts who recorded high degrees of activation among the three groups of addicts,and which had the same channel numbers,the paper classified the different drug abusers using the data as the input data for Convolutional Neural Networks(CNNs).The average three-class accuracy is 67.13%.It is of great significance for the analysis of brain function errors and personalized rehabilitation.展开更多
Reliable sales forecasts are important to the garment industry. In recent years, the global climate is warming, the weather changes frequently, and clothing sales are affected by weather fluctuations. The purpose of t...Reliable sales forecasts are important to the garment industry. In recent years, the global climate is warming, the weather changes frequently, and clothing sales are affected by weather fluctuations. The purpose of this study is to investigate whether weather data can improve the accuracy of product sales and to establish a corresponding clothing sales forecasting model. This model uses the basic attributes of clothing product data, historical sales data, and weather data. It is based on a random forest, XGB, and GBDT adopting a stacking strategy. We found that weather information is not useful for basic clothing sales forecasts, but it did improve the accuracy of seasonal clothing sales forecasts. The MSE of the dresses, down jackets, and shirts are reduced by 86.03%, 80.14%, and 41.49% on average. In addition, we found that the stacking strategy model outperformed the voting strategy model, with an average MSE reduction of 49.28%. Clothing managers can use this model to forecast their sales when they make sales plans based on weather information.展开更多
The plastic deformation of semiconductors,a process critical to their mechanical and electronic properties,involves various mechanisms such as dislocation motion and phase transition.Here,we systematically examined th...The plastic deformation of semiconductors,a process critical to their mechanical and electronic properties,involves various mechanisms such as dislocation motion and phase transition.Here,we systematically examined the temperature-dependent Peierls stress for 30°and 90°partial dislocations in cadmium telluride(CdTe),using a combination of molecular statics and molecular dynamics simulations with a machine-learning force field,as well as density functional theory simulations.Our findings reveal that the 0 K Peierls stresses for these partial dislocations in CdTe are relatively low,ranging from 0.52 GPa to 1.46 GPa,due to its significant ionic bonding characteristics.Notably,in the CdTe system containing either a 30°Cd-core or 90°Te-core partial dislocation,a phase transition from the zinc-blende phase to theβ-Sn-like phase is favored over dislocation motion.This suggests a competitive relationship between these two mechanisms,driven by the bonding characteristics within the dislocation core and the relatively low phase transition stress of∼1.00 GPa.Furthermore,we observed a general trend wherein the Peierls stress for partial dislocations in CdTe exhibits a temperature dependence,which decreases with increasing temperature,becoming lower than the phase transition stress at elevated temperatures.Consequently,the dominant deformation mechanism in CdTe shifts from solid-state phase transition at low temperatures to dislocation motion at high temperatures.This investigation uncovers a compelling interplay between dislocation motion and phase transition in the plastic deformation of CdTe,offering profound insights into the mechanical behavior and electronic performance of CdTe and other II-VI semiconductors.展开更多
The outstanding comprehensive mechanical properties of newly developed hybrid lattice structures make them useful in engineering applications for bearing multiple mechanical loads.Additive-manufacturing technologies m...The outstanding comprehensive mechanical properties of newly developed hybrid lattice structures make them useful in engineering applications for bearing multiple mechanical loads.Additive-manufacturing technologies make it possible to fabricate these highly spatially programmable structures and greatly enhance the freedom in their design.However,traditional analytical methods do not sufficiently reflect the actual vibration-damping mechanism of lattice structures and are limited by their high computational cost.In this study,a hybrid lattice structure consisting of various cells was designed based on quasi-static and vibration experiments.Subsequently,a novel parametric design method based on a data-driven approach was developed for hybrid lattices with engineered properties.The response surface method was adopted to define the sensitive optimization target.A prediction model for the lattice geometric parameters and vibration properties was established using a backpropagation neural network.Then,it was integrated into the genetic algorithm to create the optimal hybrid lattice with varying geometric features and the required wide-band vibration-damping characteristics.Validation experiments were conducted,demonstrating that the optimized hybrid lattice can achieve the target properties.In addition,the data-driven parametric design method can reduce computation time and be widely applied to complex structural designs when analytical and empirical solutions are unavailable.展开更多
High/medium entropy alloys(H/MEAs)have shown unique strengthening behavior and mechanical prop-erties because of the presence of massive local chemical orderings.Nevertheless,dynamic interactions between chemical shor...High/medium entropy alloys(H/MEAs)have shown unique strengthening behavior and mechanical prop-erties because of the presence of massive local chemical orderings.Nevertheless,dynamic interactions between chemical short-range orders(CSROs)and dislocations,and the underlying atomic strengthening mechanism remain elusive.In this work,we first developed a novel machine learning-embedded atom method(ML-EAM)potential of the CoNiV system,trained on a comprehensive first-principles dataset,which enables accurate and efficient modeling of CSRO formation and dislocation dynamics.Then,we in-vestigated the strengthening mechanisms of CSROs in CoNiV MEA through machine learning-augmented molecular dynamics(MD)simulations.Hybrid MD/Monte Carlo simulations reveal that CSRO domains possess an L1_(2)(NiCo)_(3) V structure,whose size increases with lowering annealing temperatures.These domains significantly enhance strength by impeding dislocation motion through complex energy path-ways,increasing depinning forces,and reducing mobility.Moreover,the MD simulations combined with theoretical analysis elucidate the competition between CSRO-assisted strengthening(via antiphase bound-ary formation)and solid solution weakening(via reduced atomic misfit volume).Phonon-drag effects are also amplified by CSROs,further resisting dislocation glide.Our results demonstrate that L1_(2)-CSROs strengthen CoNiV MEA primarily through antiphase boundary and phonon-drag contributions,providing new insights for designing high-performance multi-principal-element alloys via tailoring CSROs.展开更多
In recent years,physical unclonable function(PUF)has emerged as a lightweight solution in the Internet of Things security.However,conventional PUFs based on complementary metal oxide semiconductor(CMOS)present challen...In recent years,physical unclonable function(PUF)has emerged as a lightweight solution in the Internet of Things security.However,conventional PUFs based on complementary metal oxide semiconductor(CMOS)present challenges such as insufficient randomness,significant power and area overhead,and vulnerability to environmental factors,leading to reduced reliability.In this study,we realize a strong,highly reliable and reconfigurable PUF with resistance against machine-learning attacks in a 1 kb spinorbit torque magnetic random access memory fabricated using a 180 nm CMOS process.This strong PUF achieves a challenge-response pair capacity of 10^(9) through a computing-in-memory approach.The results demonstrate that the proposed PUF exhibits near-ideal performance metrics:50.07% uniformity,50% diffuseness,49.89% uniqueness,and a bit error rate of 0%,even in a 375 K environment.The reconfigurability of PUF is demonstrated by a reconfigurable Hamming distance of 49.31% and a correlation coefficient of less than 0.2,making it difficult to extract output keys through side-channel analysis.Furthermore,resistance to machine-learning modeling attacks is confirmed by achieving an ideal accuracy prediction of approximately 50% in the test set.展开更多
Liquid metals demonstrate significant potential for applications in thermal management and flexible electronic circuits, necessitating a comprehensive understanding of their transport properties for technological adva...Liquid metals demonstrate significant potential for applications in thermal management and flexible electronic circuits, necessitating a comprehensive understanding of their transport properties for technological advancements. Experimental measurement of these properties presents challenges due to factors like cost, corrosion and impurity control. Consequently, accurate computational simulations become essential for predicting the physical properties of these materials. In this research, molecular dynamics(MD) simulations were employed to model several properties of gallium(Ga), indium(In) and Ga–In alloys, including lattice structural parameters, radial distribution functions(RDF), structure factors, selfdiffusion coefficients and viscosity. Due to the difficulty of traditional interatomic potentials in capturing the short-range interactions directly related to the mechanical behavior of liquid atoms, machine-learning interatomic potentials(MLIPs)have been constructed to precisely describe the liquid metals Ga, In, and Ga–In alloys. This was achieved by utilizing the moment tensor potential(MTP) framework in combination with an active learning strategy. MTP was trained using a comprehensive database generated from DFT and MD simulations, which include a variety of crystal structures, point defects and liquid structures. The calculations of physical properties in this research have shown strong consistency with experimental data, demonstrating that the MTP can accurately describe the interatomic interactions between Ga–Ga, In–In and Ga–In. Our work has established a novel paradigm for investigating the physical properties of various liquid metal systems, offering valuable insights and references for future research.展开更多
Chemical short-range order(SRO),a phenomenon at the atomic scale resulting from inhomogeneities in the local chemical environment,is usually studied using machine learning force field-based molecular dynamics simulati...Chemical short-range order(SRO),a phenomenon at the atomic scale resulting from inhomogeneities in the local chemical environment,is usually studied using machine learning force field-based molecular dynamics simulations due to the limitations of experimental methods.To promote the reliable application of machine potentials in high-entropy alloy simulations,first,this work uses NEP models trained on two different datasets to predict the SRO coefficients of NbMoTaW.The results show that within the same machine learning framework,there are significant differences in the prediction of SRO coefficients for the Nb-Nb atomic pair.Subsequently,this work predicts the SRO coefficients of NbMoTaW using the NEP model and the SNAP model,both of which are trained on the same dataset.The results reveal significant discrepancies in SRO predictions for like-element pairs(e.g.,Nb-Nb and W-W)between the two potentials,despite the identical training data.The findings of this study indicate that discrepancies in the prediction results of SRO coefficients can arise from either the same machine learning framework trained on different datasets or different learning frameworks trained on the same dataset.This reflects possible incompleteness in the current training set's coverage of local chemical environments at the atomic scale.Future research should establish unified evaluation standards to assess the capability of training sets to accurately describe complex atomic-scale behaviors such as SRO.展开更多
Satellite Precipitation Products(SPPs) face challenges in detecting Extreme Precipitation Events(EPEs). Hence, the primary objective of this research is to introduce a novel framework termed Machine-Learning Clusterin...Satellite Precipitation Products(SPPs) face challenges in detecting Extreme Precipitation Events(EPEs). Hence, the primary objective of this research is to introduce a novel framework termed Machine-Learning Clustering-Merging Algorithms(ML-CMAs) to evaluate EPEs using SPPs and Auxiliary Data(AD). Daily precipitation measurements were utilized for training and evaluating EPE estimates over Iran, which is comprised of arid and semi-arid regions. Statistical analysis and evaluation of five SPPs demonstrated that during EPE occurrences, all products face challenges in precipitation estimation, and using these products individually is not recommended. Among the SPPs, Multi-Source Weighted-Ensemble Precipitation(MSWEP) performed best for heavy(>20 mm d–1) and extreme(>40 mm d–1)precipitation events, followed by Global Satellite Mapping of Precipitation(GSMa P), Integrated Multi-Satellite Retrievals for Global Precipitation Measurement(IMERG), Climate Prediction Center morphing technique(CMORPH), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Dynamic Infrared-Rain Rate(PERSIANN-PDIR). The findings indicate that all proposed methods based on ML-CMAs could estimate precipitation rates more accurately than SPPs and improve statistical indices. The seasonal assessment and spatial analysis of statistical metrics of the overall daily precipitation results for all periods and climates revealed that all methods based on ML-CMAs performed well in all seasons and at nearly all measurement stations. Using unsupervised K-means++ classification for clustering EPEs and Deep Neural Network(DNN) and Convolutional Neural Network(CNN) methods for merging the MLCMAs reduced the error rate of SPPs in EPE estimation by approximately 50%. Therefore, incorporating ML-CMAs along with PWV as AD can significantly improve the performance of SPPs in evaluating EPEs over the study region.展开更多
Increasing evidence suggests that non-operative management(NOM)with antibiotics could serve as a safe alternative to surgery for the treatment of uncomplicated acute appendicitis(AA).However,accurately differentiating...Increasing evidence suggests that non-operative management(NOM)with antibiotics could serve as a safe alternative to surgery for the treatment of uncomplicated acute appendicitis(AA).However,accurately differentiating between uncomplicated and complicated AA remains challenging.Our aim was to develop and validate machine-learning-based diagnostic models to differentiate uncomplicated from complicated AA.This was a multicenter cohort trial conducted from January 2021 and December 2022 across five tertiary hospitals.Three distinct diagnostic models were created,namely,the clinical-parameter-based model,the CT-radiomicsbased model,and the clinical-radiomics-fused model.These models were developed using a comprehensive set of eight machinelearning algorithms,which included logistic regression(LR),support vector machine(SVM),random forest(RF),decision tree(DT),gradient boosting(GB),K-nearest neighbors(KNN),Gaussian Naïve Bayes(GNB),and multi-layer perceptron(MLP).The performance and accuracy of these diverse models were compared.All models exhibited excellent diagnostic performance in the training cohort,achieving a maximal AUC of 1.00.For the clinical-parameter model,the GB classifier yielded the optimal AUC of 0.77(95%confidence interval[CI]:0.64-0.90)in the testing cohort,while the LR classifier yielded the optimal AUC of 0.76(95%CI:0.66-0.86)in the validation cohort.For the CT-radiomics-based model,GB classifier achieved the best AUC of 0.74(95%CI:0.60-0.88)in the testing cohort,and SVM yielded an optimal AUC of 0.63(95%CI:0.51-0.75)in the validation cohort.For the clinical-radiomics-fused model,RF classifier yielded an optimal AUC of 0.84(95%CI:0.74-0.95)in the testing cohort and 0.76(95%CI:0.67-0.86)in the validation cohort.An openaccess,user-friendly online tool was developed for clinical application.This multicenter study suggests that the clinical-radiomicsfused model,constructed using RF algorithm,effectively differentiated between complicated and uncomplicated AA.展开更多
This study traces the development of the Shanghai Typhoon Model(SHTM)from a traditional physics-based regional model toward a data-driven,machine-learning typhoon forecasting system.After upgrading its initial and bou...This study traces the development of the Shanghai Typhoon Model(SHTM)from a traditional physics-based regional model toward a data-driven,machine-learning typhoon forecasting system.After upgrading its initial and boundary conditions,SHTM now leverages large-scale constraints from machine-learning weather prediction(MLWP)models,resulting in an ML–physics hybrid framework.During Typhoon Danas(2025),the hybrid SHTM achieves substantially lower track errors than both the advanced ECMWF Integrated Forecasting System(IFS)and leading MLWP models such as PanGu and FuXi.Furthermore,the hybrid SHTM consistently maintains mean track errors below 200 km up to a forecast lead time of 108 hours,representing a significant advancement in forecast accuracy.In addition,this study highlights the technical roadmap for transitioning from a physics-based typhoon model to a fully data-driven ML typhoon forecast system.It also emphasizes that advances in the physical modeling framework provide a critical foundation for further improving the performance of future data-driven ML typhoon models.展开更多
基金supported by the National Science and Technology Advanced Materials Major Program of China(No.2024ZD0606900)W.Ouyang acknowledges the financial support from the National Natural Science Foundation of China(No.12472099)+10 种基金the Fundamental Research Funds for the Central Universities(No.2042025kf0050)S.Pan,Y.Wang,J.Shi,Z.Liang,J.Wang,and J.Sun acknowledge the financial support from the National Natural Science Foundation of China(Nos.12125404,T2495231,and 123B2049)the Basic Research Program of Jiangsu(Nos.BK20233001 and BK20241253)the Jiangsu Funding Program for Excellent Postdoctoral Talent(Nos.2024ZB002 and 2024ZB075)the Postdoctoral Fellowship Program of CPSF(No.GZC20240695)the AI&AI for the Science Program of Nanjing University,and the Fundamental Research Funds for the Central Universities,as well as the computational resources provided by the High Performance Computing Center of Collaborative Innovation Center of Advanced Microstructures and the highperformance supercomputing center of Nanjing University.P.Guan acknowledges the financial support by the National Natural Science Foundation of China(No.T2325004)E.Lindgren,T.Hainer,L.Svensson.,J.Wiktor,E.Berger,and P.Erhart gratefully acknowledge funding from the Swedish Foundation for Strategic Research(GSn15-0008 and FFL21-0129)the Swedish Research Council(Nos.2020-04935 and 2021-05072)the Knut and Alice Wallenberg Foundation(Nos.2023.0032 and 2024.0042)the European Research Council(ERC Starting Grant No.101162195)as well as computational resources provided by the National Academic Infrastructure for Supercomputing in Sweden at NSC,PDC,and C3SE partially funded by the Swedish Research Council through grant agreement no.2022-06725,as well as the Berzelius resource provided by the Knut and Alice Wallenberg Foundation at NSC.T.A-N.and Y.W.have been supported in part by the Academy of Finland through its QTF Center of Excellence program(project no.312298)European Union-NextGenerationEU instrument grant 353298.Computational resources by the CSC IT Center for Finland and the Aalto Science-IT are also gratefully acknowledged.
文摘This paper provides a comprehensive overview of the latest stable release of the graphics processing units molecular dynamics(GPUMD)package,GPUMD 4.0.We begin with a brief review of its development history,starting from the initial version.We then discuss the theoretical foundations for the development of the GPUMD package,including the formulations of the interatomic force,virial and heat current for many-body potentials,the development of the highly efficient and flexible neuroevolution potential(NEP)method,the supported integrators and related operations,the various physical properties that can be calculated on the fly,and the GPUMD ecosystem.After presenting these functionalities,we review a range of applications enabled by GPUMD,particularly in combination with the NEP approach.Finally,we outline possible future development directions for GPUMD.
基金the National Research Foundation under his NRF Fellowship NRFF12-2020-0012support from the Singapore Ministry of Education Academic Fund Tier 1(R-284-000-186-133).
文摘Ion transport in materials is routinely probed through several experimental techniques,which introduce variability in reported ionic diffusivities and conductivities.The computational prediction of ionic diffusivities and conductivities helps in identifying good ionic conductors,and suitable solid electrolytes(SEs),thus establishing firm structure-property relationships.Machine-learned potentials are an attractive strategy to extend the capabilities of accurate ab initio molecular dynamics(AIMD)to longer simulations for larger systems,enabling the study of ion transport at lower temperatures.However,machine-learned potentials being in their infancy,critical assessments of their predicting capabilities are rare.Here,we identified the main factors controlling the quality of a machine-learning potential based on the moment tensor potential formulation,when applied to the properties of ion transport in ionic conductors,such as SEs.Our results underline the importance of high-quality and diverse training sets required to fit moment tensor potentials.We highlight the importance of considering intrinsic defects which may occur in SEs.We demonstrate the limitations posed by short-timescale and high-temperature AIMD simulations to predict the room-temperature properties of materials.
文摘Objective:To identify potential key genes associated with pre-eclampsia through bioinformatics analysis,construct predictive models using machine-learning algorithms,and evaluate the models'performance in predicting pre-eclampsia.Methods:Gene-expression microarray datasets GSE10588,GSE66273,and GSE30186 related to pre-eclampsia were downloaded from the gene expression omnibus(GEO).Data were normalized using R,and differentially expressed genes(DEGs)were identified.LASSO regression was applied to further filter DEGs.Based on the selected DEGs,six machine-learning models-logistic regression(LR),random forest(RF),support vector machine(SVM),K-nearest neighbors(KNN),neural network(NN),and eXtreme gradient boosting(XGBoost)were built in R,and their performance was validated.Results:From the three datasets,a total of 1,363 genes were extracted.LASSO regression narrowed these to 265 candidate key genes.Multivariate analysis ultimately identified four genes closely associated with pre-eclampsia:EVI5,GCLM,LEP,and SYNPO2L.Using these four key genes,six machine-learning models were constructed.Receiver operating characteristic(ROC)analysis showed that all models achieved AUC>0.9:LR(AUC=0.983,95%CI=0.942-0.998),RF(AUC=0.961,95%CI=0.912-0.987),SVM(AUC=0.936,95%CI=0.879-0.972),KNN(AUC=0.970,95%CI=0.924-0.992),NN(AUC=0.916,95%CI=0.854-0.958),and XGBoost(AUC=0.952,95%CI=0.900-0.982).There was no statistically significant difference among the AUCs of the models(P>0.05).Conclusion:This study identified four key genes linked to preeclampsia through integrated bioinformatics analysis.Predictive models built on these genes can accurately forecast the occurrence of pre-eclampsia,suggesting that the four genes may serve as potential biomarkers for early diagnosis and therapeutic targeting of pre-eclampsia.
文摘All-solid-state batteries(ASSBs)represent a next-generation energy storage technology,offering enhanced safety,higher energy density,and improved cycling stability compared to conventional liquid-electrolyte-based lithium-ion batteries.Understanding and optimizing the complex chemistries and interfaces that underpin ASSB performance present significant challenges from both experimental and modeling perspectives.In particular,atomistic simulations face difficulties in capturing the complex structure,disorder,and dynamic evolution of materials and interfaces under practically relevant conditions.While established methods such as density functional theory and classical force fields have provided valuable insights,some questions remain difficult to address,particularly those involving large system sizes or long timescales.Recently,machine learning interatomic potentials(MLIPs)have emerged as a transformative tool,enabling atomistic simulations at length and time scales that were previously challenging to access with conventional approaches.By delivering near first-principles accuracy with much greater efficiency,MLIPs open new avenues for large-scale,long-timescale,and high-throughput simulations of solid-state battery materials.In this review,we present a comparative overview of density functional theory,classical force fields,and MLIPs,highlighting their respective strengths and limitations in ASSB research.We then discuss how MLIPs enable simulations that reach longer timescales,larger system sizes,and support high-throughput calculations,providing unique insights into ion transport and interfacial evolution in ASSBs.Finally,we conclude with a summary and outlook on current challenges and future opportunities for expanding MLIP capabilities and accelerating their impact in solid-state battery research.
文摘Objective: Postoperative complications adversely affected the prognosis in patients with gastric cancer. This study intends to investigate the feasibility of using machine-learning model to predict surgical outcomes in patients undergoing gastrectomy.Methods: In this study, cancer patients who underwent gastrectomy at Shanghai Rui Jin Hospital in 2017 were randomly assigned to a development or validation cohort in a 9:1 ratio. A support vector classification(SVC) model to predict surgical outcomes in patients undergoing gastrectomy was developed and further validated.Results: A total of 321 patients with 32 features were collected. The positive and negative outcomes of postoperative complication after gastrectomy appeared in 100(31.2%) and 221(68.8%) patients, respectively. The SVC model was constructed to predict surgical outcomes in patients undergoing gastrectomy. The accuracy of 10-fold cross validation and external verification was 78.17% and 78.12%, respectively. Further, an online web server has been developed to share the SVC model for machine-learning-assisted prediction of surgical outcomes in patients undergoing gastrectomy in the future procedures, which is accessible at the web address:http://47.100.47.97:5005/r_model_prediction.Conclusions: The SVC model was a useful predictor for measuring the risk of postoperative complications after gastrectomy, which may help stratify patients with different overall status for choice of surgical procedure or other treatments. It can be expected that machine-learning models in cancer informatics research are possibly shareable and accessible via web address all over the world.
基金supported by the National Key Plan for Scientific Research and Development of China(2016YFD0500300)National Natural Science Foundation of China(32170651)Hunan Provincial Natural Science Foundation of China(2020JJ3006)。
文摘The coronavirus 3C-like(3CL)protease,a cysteine protease,plays an important role in viral infection and immune escape.However,there is still a lack of effective tools for determining the cleavage sites of the 3CL protease.This study systematically investigated the diversity of the cleavage sites of the coronavirus 3CL protease on the viral polyprotein,and found that the cleavage motif were highly conserved for viruses in the genera of Alphacoronavirus,Betacoronavirus and Gammacoronavirus.Strong residue preferences were observed at the neighboring positions of the cleavage sites.A random forest(RF)model was built to predict the cleavage sites of the coronavirus 3CL protease based on the representation of residues in cleavage motifs by amino acid indexes,and the model achieved an AUC of 0.96 in cross-validations.The RF model was further tested on an independent test dataset which were composed of cleavage sites on 99 proteins from multiple coronavirus hosts.It achieved an AUC of 0.95 and predicted correctly 80%of the cleavage sites.Then,1,352 human proteins were predicted to be cleaved by the 3CL protease by the RF model.These proteins were enriched in several GO terms related to the cytoskeleton,such as the microtubule,actin and tubulin.Finally,a webserver named 3CLP was built to predict the cleavage sites of the coronavirus 3CL protease based on the RF model.Overall,the study provides an effective tool for identifying cleavage sites of the 3CL protease and provides insights into the molecular mechanism underlying the pathogenicity of coronaviruses.
基金This work is supported by the Scientific Research Project of Educational Department of Liaoning Province(Grant No.LJKZ0082)the Program of Hainan Association for Science and Technology Plans to Youth R&D Innovation(Grant No.QCXM201910)+2 种基金the National Natural Science Foundation of China(Grant Nos.61802092 and 92067110)the Hainan Provincial Natural Science Foundation of China(Grant No.620RC562)2020 Industrial Internet Innovation and Development Project-Industrial Internet Identification Data Interaction Middleware and Resource Pool Service Platform Project,Ministry of Industry and Information Technology of the People’s Republic of China.
文摘Anomaly detection is becoming increasingly significant in industrial cyber security,and different machine-learning algorithms have been generally acknowledged as various effective intrusion detection engines to successfully identify cyber attacks.However,different machine-learning algorithms may exhibit their own detection effects even if they analyze the same feature samples.As a sequence,after developing one feature generation approach,the most effective and applicable detection engines should be desperately selected by comparing distinct properties of each machine-learning algorithm.Based on process control features generated by directed function transition diagrams,this paper introduces five different machine-learning algorithms as alternative detection engines to discuss their matching abilities.Furthermore,this paper not only describes some qualitative properties to compare their advantages and disadvantages,but also gives an in-depth and meticulous research on their detection accuracies and consuming time.In the verified experiments,two attack models and four different attack intensities are defined to facilitate all quantitative comparisons,and the impacts of detection accuracy caused by the feature parameter are also comparatively analyzed.All experimental results can clearly explain that SVM(Support Vector Machine)and WNN(Wavelet Neural Network)are suggested as two applicable detection engines under differing cases.
基金supported by the Scientific Research Fund Project of Yunnan Education Department(Grant Numbers 2023J1974 and 2023J1976)the Yunnan University Professional Degree Graduate Student Practical Innovation Fund Project(Grant Number ZC-22222374)also supported by the Yunnan Provincial Education Department Fund(Grant No.2022Y286).
文摘The composite exciter and the CaO to Na_(2)SO_(4) dosing ratios are known to have a strong impact on the mechanical strength offly-ash concrete.In the present study a hybrid approach relying on experiments and a machine-learn-ing technique has been used to tackle this problem.The tests have shown that the optimal admixture of CaO and Na_(2)SO_(4) alone is 8%.The best 3D mechanical strength offly-ash concrete is achieved at 8%of the compound activator;If the 28-day mechanical strength is considered,then,the best performances are obtained at 4%of the compound activator.Moreover,the 3D mechanical strength offly-ash concrete is better when the dosing ratio of CaO to Na_(2)SO_(4) in the compound activator is 1:1;the maximum strength offly-ash concrete at 28-day can be achieved for a 1:1 ratio of CaO to Na_(2)SO_(4) by considering a 4%compound activator.In this case,the compressive andflexural strengths are 260 MPa and 53.6 MPa,respectively;the mechanical strength offly-ash concrete at 28-day can be improved by a 4:1 ratio of CaO to Na_(2)SO_(4) by considering 8%and 12%compound excitants.It is shown that the predictions based on the aforementioned machine-learning approach are accurate and reliable.
基金supported by the National Key R&D Program of China(Nos.2022YFC3400700,2022YFA0806600)the Key Research and Development Project of Hubei Province(No.2023BCB094)+1 种基金the Interdisciplinary Innovative Talents Foundation from Renmin Hospital of Wuhan University(No.JCRCGW-2022-008)the Key Laboratory of Hubei Province(No.2021KFY005)。
文摘Chronic kidney disease(CKD)is an increasingly prevalent medical condition associated with high mortality and cardiovascular complications.The intricate interplay between kidney dysfunction and subsequent metabolic disturbances may provide insights into the underlying mechanisms driving CKD onset and progression.Herein,we proposed a large-scale plasma metabolite identification and quantification system that combines the strengths of targeted and untargeted metabolomics technologies,i.e.,widely-targeted metabolomics(WT-Met)approach.WT-Met method enables large-scale identification and accurate quantification of thousands of metabolites.We collected plasma samples from 21 healthy controls and 62CKD patients,categorized into different stages(22 in stages 1-3,20 in stage 4,and 20 in stage 5).Using LC-MS-based WT-Met approach,we were able to effectively annotate and quantify a total of 1431metabolites from the plasma samples.Focusing on the 539 endogenous metabolites,we identified 399significantly altered metabolites and depicted their changing patterns from healthy controls to end-stage CKD.Furthermore,we employed machine-learning to identify the optimal combination of metabolites for predicting different stages of CKD.We generated a multiclass classifier consisting of 7 metabolites by machine-learning,which exhibited an average AUC of 0.99 for the test set.In general,amino acids,nucleotides,organic acids,and their metabolites emerged as the most significantly altered metabolites.However,their patterns of change varied across different stages of CKD.The 7-metabolite panel demonstrates promising potential as biomarker candidates for CKD.Further exploration of these metabolites can provide valuable insights into their roles in the etiology and progression of CKD.
基金This work was supported by the National Natural Science Foundation of China(No.61976133)Shanghai Industrial Collaborative Technology Innovation Project(No.2021-cyxt1-kj14)+2 种基金Major Scienti¯c and Technological Innovation Projects of Shan Dong Province(No.2019JZZY021010)Science and Technology Innovation Base Project of Shanghai Science and Technology Commission(19DZ2255200)Defense Industrial Technology Development Program(JCKY2019413D002).
文摘Drug addiction can cause abnormal brain activation changes,which are the root cause of drug craving and brain function errors.This study enrolled drug abusers to determine the effects of different drugs on brain activation.A functional near-infrared spectroscopy(fNIRS)device was used for the research.This study was designed with an experimental paradigm that included the induction of resting and drug addiction cravings.We collected the fNIRS data of 30 drug users,including 10 who used heroin,10 who used Methamphetamine,and 10 who used mixed drugs.First,using Statistical Analysis,the study analyzed the activations of eight functional areas of the left and right hemispheres of the prefrontal cortex of drug addicts who respectively used heroin,Methamphetamine,and mixed drugs,including Left/Right-Dorsolateral prefrontal cortex(L/R-DLPFC),Left/Right-Ventrolateral prefrontal cortex(L/R-VLPFC),Left/Right-Fronto-polar prefrontal cortex(L/R-FPC),and Left/Right Orbitofrontal Cortex(L/R-OFC).Second,referencing the degrees of activation of oxyhaemoglobin concentration(HbO2),the study made an analysis and got the specific activation patterns of each group of the addicts.Finally,after taking out data which are related to the addicts who recorded high degrees of activation among the three groups of addicts,and which had the same channel numbers,the paper classified the different drug abusers using the data as the input data for Convolutional Neural Networks(CNNs).The average three-class accuracy is 67.13%.It is of great significance for the analysis of brain function errors and personalized rehabilitation.
文摘Reliable sales forecasts are important to the garment industry. In recent years, the global climate is warming, the weather changes frequently, and clothing sales are affected by weather fluctuations. The purpose of this study is to investigate whether weather data can improve the accuracy of product sales and to establish a corresponding clothing sales forecasting model. This model uses the basic attributes of clothing product data, historical sales data, and weather data. It is based on a random forest, XGB, and GBDT adopting a stacking strategy. We found that weather information is not useful for basic clothing sales forecasts, but it did improve the accuracy of seasonal clothing sales forecasts. The MSE of the dresses, down jackets, and shirts are reduced by 86.03%, 80.14%, and 41.49% on average. In addition, we found that the stacking strategy model outperformed the voting strategy model, with an average MSE reduction of 49.28%. Clothing managers can use this model to forecast their sales when they make sales plans based on weather information.
基金supported by the National Science Foundation(No.CMMI-2019459).
文摘The plastic deformation of semiconductors,a process critical to their mechanical and electronic properties,involves various mechanisms such as dislocation motion and phase transition.Here,we systematically examined the temperature-dependent Peierls stress for 30°and 90°partial dislocations in cadmium telluride(CdTe),using a combination of molecular statics and molecular dynamics simulations with a machine-learning force field,as well as density functional theory simulations.Our findings reveal that the 0 K Peierls stresses for these partial dislocations in CdTe are relatively low,ranging from 0.52 GPa to 1.46 GPa,due to its significant ionic bonding characteristics.Notably,in the CdTe system containing either a 30°Cd-core or 90°Te-core partial dislocation,a phase transition from the zinc-blende phase to theβ-Sn-like phase is favored over dislocation motion.This suggests a competitive relationship between these two mechanisms,driven by the bonding characteristics within the dislocation core and the relatively low phase transition stress of∼1.00 GPa.Furthermore,we observed a general trend wherein the Peierls stress for partial dislocations in CdTe exhibits a temperature dependence,which decreases with increasing temperature,becoming lower than the phase transition stress at elevated temperatures.Consequently,the dominant deformation mechanism in CdTe shifts from solid-state phase transition at low temperatures to dislocation motion at high temperatures.This investigation uncovers a compelling interplay between dislocation motion and phase transition in the plastic deformation of CdTe,offering profound insights into the mechanical behavior and electronic performance of CdTe and other II-VI semiconductors.
基金supported by National Natural Science Foundation of China(Grant No.52375380)National Key R&D Program of China(Grant No.2022YFB3402200)the Key Project of National Natural Science Foundation of China(Grant No.12032018).
文摘The outstanding comprehensive mechanical properties of newly developed hybrid lattice structures make them useful in engineering applications for bearing multiple mechanical loads.Additive-manufacturing technologies make it possible to fabricate these highly spatially programmable structures and greatly enhance the freedom in their design.However,traditional analytical methods do not sufficiently reflect the actual vibration-damping mechanism of lattice structures and are limited by their high computational cost.In this study,a hybrid lattice structure consisting of various cells was designed based on quasi-static and vibration experiments.Subsequently,a novel parametric design method based on a data-driven approach was developed for hybrid lattices with engineered properties.The response surface method was adopted to define the sensitive optimization target.A prediction model for the lattice geometric parameters and vibration properties was established using a backpropagation neural network.Then,it was integrated into the genetic algorithm to create the optimal hybrid lattice with varying geometric features and the required wide-band vibration-damping characteristics.Validation experiments were conducted,demonstrating that the optimized hybrid lattice can achieve the target properties.In addition,the data-driven parametric design method can reduce computation time and be widely applied to complex structural designs when analytical and empirical solutions are unavailable.
基金financially supported by the National Natural Science Foundation of China(Nos.52071024,52271003 and 52101188)the National Science Fund for Distinguished Young Scholars of China(No.52225103)+2 种基金the Funds for Creative Research Groups of China(No.51921001)the Projects of International Cooperation and Exchanges NSFC(Nos.51961160729 and 52061135207)the Fundamental Research Fund for the Central Universities of China,and the State Key Laboratory for Advanced Metals and Materials.
文摘High/medium entropy alloys(H/MEAs)have shown unique strengthening behavior and mechanical prop-erties because of the presence of massive local chemical orderings.Nevertheless,dynamic interactions between chemical short-range orders(CSROs)and dislocations,and the underlying atomic strengthening mechanism remain elusive.In this work,we first developed a novel machine learning-embedded atom method(ML-EAM)potential of the CoNiV system,trained on a comprehensive first-principles dataset,which enables accurate and efficient modeling of CSRO formation and dislocation dynamics.Then,we in-vestigated the strengthening mechanisms of CSROs in CoNiV MEA through machine learning-augmented molecular dynamics(MD)simulations.Hybrid MD/Monte Carlo simulations reveal that CSRO domains possess an L1_(2)(NiCo)_(3) V structure,whose size increases with lowering annealing temperatures.These domains significantly enhance strength by impeding dislocation motion through complex energy path-ways,increasing depinning forces,and reducing mobility.Moreover,the MD simulations combined with theoretical analysis elucidate the competition between CSRO-assisted strengthening(via antiphase bound-ary formation)and solid solution weakening(via reduced atomic misfit volume).Phonon-drag effects are also amplified by CSROs,further resisting dislocation glide.Our results demonstrate that L1_(2)-CSROs strengthen CoNiV MEA primarily through antiphase boundary and phonon-drag contributions,providing new insights for designing high-performance multi-principal-element alloys via tailoring CSROs.
基金supported by the National Natural Science Foundation of China(92164206,52261145694,T2394474,T2394470,623B2015,62271026,62401026,and 62404013)the National Key Research and Development Program of China(2022YFB4400200)+1 种基金the New Cornerstone Science Foundation through the XPLORER PRIZE,the National Postdoctoral Program for Innovative Talents(BX20220374 and BX20240455)the China Postdoctoral Science Foundation Funded Project(2023M740177 and 2022M720345).
文摘In recent years,physical unclonable function(PUF)has emerged as a lightweight solution in the Internet of Things security.However,conventional PUFs based on complementary metal oxide semiconductor(CMOS)present challenges such as insufficient randomness,significant power and area overhead,and vulnerability to environmental factors,leading to reduced reliability.In this study,we realize a strong,highly reliable and reconfigurable PUF with resistance against machine-learning attacks in a 1 kb spinorbit torque magnetic random access memory fabricated using a 180 nm CMOS process.This strong PUF achieves a challenge-response pair capacity of 10^(9) through a computing-in-memory approach.The results demonstrate that the proposed PUF exhibits near-ideal performance metrics:50.07% uniformity,50% diffuseness,49.89% uniqueness,and a bit error rate of 0%,even in a 375 K environment.The reconfigurability of PUF is demonstrated by a reconfigurable Hamming distance of 49.31% and a correlation coefficient of less than 0.2,making it difficult to extract output keys through side-channel analysis.Furthermore,resistance to machine-learning modeling attacks is confirmed by achieving an ideal accuracy prediction of approximately 50% in the test set.
基金Project supported by the National Natural Science Foundation of China (Grant Nos. 12202159 and 12472216)。
文摘Liquid metals demonstrate significant potential for applications in thermal management and flexible electronic circuits, necessitating a comprehensive understanding of their transport properties for technological advancements. Experimental measurement of these properties presents challenges due to factors like cost, corrosion and impurity control. Consequently, accurate computational simulations become essential for predicting the physical properties of these materials. In this research, molecular dynamics(MD) simulations were employed to model several properties of gallium(Ga), indium(In) and Ga–In alloys, including lattice structural parameters, radial distribution functions(RDF), structure factors, selfdiffusion coefficients and viscosity. Due to the difficulty of traditional interatomic potentials in capturing the short-range interactions directly related to the mechanical behavior of liquid atoms, machine-learning interatomic potentials(MLIPs)have been constructed to precisely describe the liquid metals Ga, In, and Ga–In alloys. This was achieved by utilizing the moment tensor potential(MTP) framework in combination with an active learning strategy. MTP was trained using a comprehensive database generated from DFT and MD simulations, which include a variety of crystal structures, point defects and liquid structures. The calculations of physical properties in this research have shown strong consistency with experimental data, demonstrating that the MTP can accurately describe the interatomic interactions between Ga–Ga, In–In and Ga–In. Our work has established a novel paradigm for investigating the physical properties of various liquid metal systems, offering valuable insights and references for future research.
基金Project supported by the Hunan Provincial Natural Science Foundation(Grant Nos.2024JJ6190 and 2024JK2007-1)。
文摘Chemical short-range order(SRO),a phenomenon at the atomic scale resulting from inhomogeneities in the local chemical environment,is usually studied using machine learning force field-based molecular dynamics simulations due to the limitations of experimental methods.To promote the reliable application of machine potentials in high-entropy alloy simulations,first,this work uses NEP models trained on two different datasets to predict the SRO coefficients of NbMoTaW.The results show that within the same machine learning framework,there are significant differences in the prediction of SRO coefficients for the Nb-Nb atomic pair.Subsequently,this work predicts the SRO coefficients of NbMoTaW using the NEP model and the SNAP model,both of which are trained on the same dataset.The results reveal significant discrepancies in SRO predictions for like-element pairs(e.g.,Nb-Nb and W-W)between the two potentials,despite the identical training data.The findings of this study indicate that discrepancies in the prediction results of SRO coefficients can arise from either the same machine learning framework trained on different datasets or different learning frameworks trained on the same dataset.This reflects possible incompleteness in the current training set's coverage of local chemical environments at the atomic scale.Future research should establish unified evaluation standards to assess the capability of training sets to accurately describe complex atomic-scale behaviors such as SRO.
文摘Satellite Precipitation Products(SPPs) face challenges in detecting Extreme Precipitation Events(EPEs). Hence, the primary objective of this research is to introduce a novel framework termed Machine-Learning Clustering-Merging Algorithms(ML-CMAs) to evaluate EPEs using SPPs and Auxiliary Data(AD). Daily precipitation measurements were utilized for training and evaluating EPE estimates over Iran, which is comprised of arid and semi-arid regions. Statistical analysis and evaluation of five SPPs demonstrated that during EPE occurrences, all products face challenges in precipitation estimation, and using these products individually is not recommended. Among the SPPs, Multi-Source Weighted-Ensemble Precipitation(MSWEP) performed best for heavy(>20 mm d–1) and extreme(>40 mm d–1)precipitation events, followed by Global Satellite Mapping of Precipitation(GSMa P), Integrated Multi-Satellite Retrievals for Global Precipitation Measurement(IMERG), Climate Prediction Center morphing technique(CMORPH), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Dynamic Infrared-Rain Rate(PERSIANN-PDIR). The findings indicate that all proposed methods based on ML-CMAs could estimate precipitation rates more accurately than SPPs and improve statistical indices. The seasonal assessment and spatial analysis of statistical metrics of the overall daily precipitation results for all periods and climates revealed that all methods based on ML-CMAs performed well in all seasons and at nearly all measurement stations. Using unsupervised K-means++ classification for clustering EPEs and Deep Neural Network(DNN) and Convolutional Neural Network(CNN) methods for merging the MLCMAs reduced the error rate of SPPs in EPE estimation by approximately 50%. Therefore, incorporating ML-CMAs along with PWV as AD can significantly improve the performance of SPPs in evaluating EPEs over the study region.
基金funded by the Natural Science Foundation of China for Youth[grant number 82103593]Program of Three Talents and One Team[grant number 4143524].
文摘Increasing evidence suggests that non-operative management(NOM)with antibiotics could serve as a safe alternative to surgery for the treatment of uncomplicated acute appendicitis(AA).However,accurately differentiating between uncomplicated and complicated AA remains challenging.Our aim was to develop and validate machine-learning-based diagnostic models to differentiate uncomplicated from complicated AA.This was a multicenter cohort trial conducted from January 2021 and December 2022 across five tertiary hospitals.Three distinct diagnostic models were created,namely,the clinical-parameter-based model,the CT-radiomicsbased model,and the clinical-radiomics-fused model.These models were developed using a comprehensive set of eight machinelearning algorithms,which included logistic regression(LR),support vector machine(SVM),random forest(RF),decision tree(DT),gradient boosting(GB),K-nearest neighbors(KNN),Gaussian Naïve Bayes(GNB),and multi-layer perceptron(MLP).The performance and accuracy of these diverse models were compared.All models exhibited excellent diagnostic performance in the training cohort,achieving a maximal AUC of 1.00.For the clinical-parameter model,the GB classifier yielded the optimal AUC of 0.77(95%confidence interval[CI]:0.64-0.90)in the testing cohort,while the LR classifier yielded the optimal AUC of 0.76(95%CI:0.66-0.86)in the validation cohort.For the CT-radiomics-based model,GB classifier achieved the best AUC of 0.74(95%CI:0.60-0.88)in the testing cohort,and SVM yielded an optimal AUC of 0.63(95%CI:0.51-0.75)in the validation cohort.For the clinical-radiomics-fused model,RF classifier yielded an optimal AUC of 0.84(95%CI:0.74-0.95)in the testing cohort and 0.76(95%CI:0.67-0.86)in the validation cohort.An openaccess,user-friendly online tool was developed for clinical application.This multicenter study suggests that the clinical-radiomicsfused model,constructed using RF algorithm,effectively differentiated between complicated and uncomplicated AA.
基金supported by the Special Project-Original Exploration(Grant No.42450163)the National Youth Science Foundation of China Project(Grant No.4240050560)the Research and Development of Key Technologies for Artificial Intelligence Regional Typhoon Forecasting Model project.
文摘This study traces the development of the Shanghai Typhoon Model(SHTM)from a traditional physics-based regional model toward a data-driven,machine-learning typhoon forecasting system.After upgrading its initial and boundary conditions,SHTM now leverages large-scale constraints from machine-learning weather prediction(MLWP)models,resulting in an ML–physics hybrid framework.During Typhoon Danas(2025),the hybrid SHTM achieves substantially lower track errors than both the advanced ECMWF Integrated Forecasting System(IFS)and leading MLWP models such as PanGu and FuXi.Furthermore,the hybrid SHTM consistently maintains mean track errors below 200 km up to a forecast lead time of 108 hours,representing a significant advancement in forecast accuracy.In addition,this study highlights the technical roadmap for transitioning from a physics-based typhoon model to a fully data-driven ML typhoon forecast system.It also emphasizes that advances in the physical modeling framework provide a critical foundation for further improving the performance of future data-driven ML typhoon models.
