金河田“网@神”——电脑网络机箱

“网@神”机箱是一款不错的产品,它的品质绝对值得信赖,而且“网络+电话”对一些繁忙的网迷说也是一项非常实用的功能。

Learning the parameters of a class of stochastic Lotka-Volterra systems with neural networks

In this paper,we propose a neural network approach to learn the parameters of a class of stochastic Lotka-Volterra systems.Approximations of the mean and covariance matrix of the observational variables are obtained from the Euler-Maruyama discretization of the underlying stochastic differential equations(SDEs),based on which the loss function is built.The stochastic gradient descent method is applied in the neural network training.Numerical experiments demonstrate the effectiveness of our method.

Predictor−corrector inverse design scheme for property−composition prediction of amorphous alloys

In order to develop a generic framework capable of designing novel amorphous alloys with selected target properties,a predictor−corrector inverse design scheme(PCIDS)consisting of a predictor module and a corrector module was presented.A high-precision forward prediction model based on deep neural networks was developed to implement these two parts.Of utmost importance,domain knowledge-guided inverse design networks(DKIDNs)and regular inverse design networks(RIDNs)were also developed.The forward prediction model possesses a coefficient of determination(R^(2))of 0.990 for the shear modulus and 0.986 for the bulk modulus on the testing set.Furthermore,the DKIDNs model exhibits superior performance compared to the RIDNs model.It is finally demonstrated that PCIDS can efficiently predict amorphous alloy compositions with the required target properties.

Pseudo channel:time embedding for motor imagery decoding

Motor imagery(MI)based electroencephalogram(EEG)represents a frontier in enabling direct neural control of external devices and advancing neural rehabilitation.This study introduces a novel time embedding technique,termed traveling-wave based time embedding,utilized as a pseudo channel to enhance the decoding accuracy of MI-EEG signals across various neural network architectures.Unlike traditional neural network methods that fail to account for the temporal dynamics in MI-EEG in individual difference,our approach captures time-related changes for different participants based on a priori knowledge.Through extensive experimentation with multiple participants,we demonstrate that this method not only improves classification accuracy but also exhibits greater adaptability to individual differences compared to position encoding used in Transformer architecture.Significantly,our results reveal that traveling-wave based time embedding crucially enhances decoding accuracy,particularly for participants typically considered“EEG-illiteracy”.As a novel direction in EEG research,the traveling-wave based time embedding not only offers fresh insights for neural network decoding strategies but also expands new avenues for research into attention mechanisms in neuroscience and a deeper understanding of EEG signals.

Classification method of infrasound events based on the MVIDA algorithm and MS-SE-ResNet

The verification of nuclear test ban necessitates the classification and identification of infrasound events.The accurate and eff ective classification of seismic and chemical explosion infrasounds can promote the classification and identification of infrasound events.However,overfitting of the signals of seismic and chemical explosion infrasounds easily occurs during training due to the limited amount of data.Thus,to solve this problem,this paper proposes a classification method based on the mixed virtual infrasound data augmentation(MVIDA)algorithm and multiscale squeeze-and-excitation ResNet(MS-SE-ResNet).In this study,the eff ectiveness of the proposed method is verified through simulation and comparison experiments.The simulation results reveal that the MS-SE-ResNet network can eff ectively determine the separability of chemical explosion and seismic infrasounds in the frequency domain,and the average classification accuracy on the dataset enhanced by the MVIDA algorithm reaches 81.12%.This value is higher than those of the other four types of comparative classification methods.This work also demonstrates the eff ectiveness and stability of the augmentation algorithm and classification network in the classification of few-shot infrasound events.

Embedding-based approximate query for knowledge graph

To solve the low efficiency of approximate queries caused by the large sizes of the knowledge graphs in the real world,an embedding-based approximate query method is proposed.First,the nodes in the query graph are classified according to the degrees of approximation required for different types of nodes.This classification transforms the query problem into three constraints,from which approximate information is extracted.Second,candidates are generated by calculating the similarity between embeddings.Finally,a deep neural network model is designed,incorporating a loss function based on the high-dimensional ellipsoidal diffusion distance.This model identifies the distance between nodes using their embeddings and constructs a score function.k nodes are returned as the query results.The results show that the proposed method can return both exact results and approximate matching results.On datasets DBLP(DataBase systems and Logic Programming)and FUA-S(Flight USA Airports-Sparse),this method exhibits superior performance in terms of precision and recall,returning results in 0.10 and 0.03 s,respectively.This indicates greater efficiency compared to PathSim and other comparative methods.

Gated Neural Network-Based Unsteady Aerodynamic Modeling for Large Angles of Attack

Modeling of unsteady aerodynamic loads at high angles of attack using a small amount of experimental or simulation data to construct predictive models for unknown states can greatly improve the efficiency of aircraft unsteady aerodynamic design and flight dynamics analysis.In this paper,aiming at the problems of poor generalization of traditional aerodynamic models and intelligent models,an intelligent aerodynamic modeling method based on gated neural units is proposed.The time memory characteristics of the gated neural unit is fully utilized,thus the nonlinear flow field characterization ability of the learning and training process is enhanced,and the generalization ability of the whole prediction model is improved.The prediction and verification of the model are carried out under the maneuvering flight condition of NACA0015 airfoil.The results show that the model has good adaptability.In the interpolation prediction,the maximum prediction error of the lift and drag coefficients and the moment coefficient does not exceed 10%,which can basically represent the variation characteristics of the entire flow field.In the construction of extrapolation models,the training model based on the strong nonlinear data has good accuracy for weak nonlinear prediction.Furthermore,the error is larger,even exceeding 20%,which indicates that the extrapolation and generalization capabilities need to be further optimized by integrating physical models.Compared with the conventional state space equation model,the proposed method can improve the extrapolation accuracy and efficiency by 78%and 60%,respectively,which demonstrates the applied potential of this method in aerodynamic modeling.

Rolling bearing fault diagnosis based on data-level and feature-level information fusion

To address the limitation of single acceleration sensor signals in effectively reflecting the health status of rolling bearings,a rolling bearing fault diagnosis method based on the fusion of data-level and feature-level information was proposed.First,according to the impact characteristics of rolling bearing faults,correlation kurtosis rules were designed to guide the weight distribution of multi-sensor signals.These rules were then combined with a weighted fusion method to obtain high-quality data-level fusion signals.Subsequently,a feature-fusion convolutional neural network(FFCNN)that merges the one-dimensional(1D)features extracted from the fused signal with the two-dimensional(2D)features extracted from the wavelet time-frequency spectrum was designed to obtain a comprehensive representation of the health status of rolling bearings.Finally,the fused features were fed into a Softmax classifier to complete the fault diagnosis.The results show that the proposed method exhibits an average test accuracy of over 99.00%on the two rolling bearing fault datasets,outperforming other comparison methods.Thus,the method can be effectively utilized for diagnosing rolling bearing faults.

The enlightenment of artificial intelligence large-scale model on the research of intelligent eye diagnosis in traditional Chinese medicine

Eye diagnosis is a method for inspecting systemic diseases and syndromes by observing the eyes.With the development of intelligent diagnosis in traditional Chinese medicine(TCM);artificial intelligence(AI)can improve the accuracy and efficiency of eye diagnosis.However;the research on intelligent eye diagnosis still faces many challenges;including the lack of standardized and precisely labeled data;multi-modal information analysis;and artificial in-telligence models for syndrome differentiation.The widespread application of AI models in medicine provides new insights and opportunities for the research of eye diagnosis intelli-gence.This study elaborates on the three key technologies of AI models in the intelligent ap-plication of TCM eye diagnosis;and explores the implications for the research of eye diagno-sis intelligence.First;a database concerning eye diagnosis was established based on self-su-pervised learning so as to solve the issues related to the lack of standardized and precisely la-beled data.Next;the cross-modal understanding and generation of deep neural network models to address the problem of lacking multi-modal information analysis.Last;the build-ing of data-driven models for eye diagnosis to tackle the issue of the absence of syndrome dif-ferentiation models.In summary;research on intelligent eye diagnosis has great potential to be applied the surge of AI model applications.

Generalized Mechanism for the Solid Phase Transition of M_(2)O_(3)(M=Al,Ga)Featuring Single Cation Migration and Martensitic Lattice Transformation

Al_(2)O_(3)and Ga_(2)O_(3)exhibit numerous crystal phases with distinct stabilities and materialproperties.However,the phase transitions among thosematerialsare typicallyundesirable in industrial applications,making it imperative to elucidate the transition mechanisms between these phases.The configurational similarities between Al_(2)O_(3)and Ga_(2)O_(3)allow for the replication of phase transition pathways between these materials.In this study,we investigate the potential phase transition pathway of alumina from the 0-phase to the α-phase using stochastic surface walking global optimization based on global neural network potentials,while extending an existing Ga_(2)O_(3)phase transition path.Through this exploration,we identify a novel single-atom migration pseudomartensitic mechanism,which combines martensitic transformation with single-atom diffusion.This discovery offers valuable insights for experimental endeavors aimed at stabilizing alumina in transitional phases.

Stability of second order Hopfield neural networks with time delays

Dynamical behaviors of a class of second order Hopfield neural networks with time delays is investigated.The existence of a unique equilibrium point is proved by using Brouwer's fixed point theorem and the counter proof method,and some sufficient conditions for the global asymptotic stability of the equilibrium point are obtained through the combination of a suitable Lyapunov function and an algebraic inequality technique.

3D Ice Shape Description Method Based on BLSOM Neural Network

When checking the ice shape calculation software,its accuracy is judged based on the proximity between the calculated ice shape and the typical test ice shape.Therefore,determining the typical test ice shape becomes the key task of the icing wind tunnel tests.In the icing wind tunnel test of the tail wing model of a large amphibious aircraft,in order to obtain accurate typical test ice shape,the Romer Absolute Scanner is used to obtain the 3D point cloud data of the ice shape on the tail wing model.Then,the batch-learning self-organizing map(BLSOM)neural network is used to obtain the 2D average ice shape along the model direction based on the 3D point cloud data of the ice shape,while its tolerance band is calculated using the probabilistic statistical method.The results show that the combination of 2D average ice shape and its tolerance band can represent the 3D characteristics of the test ice shape effectively,which can be used as the typical test ice shape for comparative analysis with the calculated ice shape.

Multi-dimension and multi-modal rolling mill vibration prediction model based on multi-level network fusion

Mill vibration is a common problem in rolling production,which directly affects the thickness accuracy of the strip and may even lead to strip fracture accidents in serious cases.The existing vibration prediction models do not consider the features contained in the data,resulting in limited improvement of model accuracy.To address these challenges,this paper proposes a multi-dimensional multi-modal cold rolling vibration time series prediction model(MDMMVPM)based on the deep fusion of multi-level networks.In the model,the long-term and short-term modal features of multi-dimensional data are considered,and the appropriate prediction algorithms are selected for different data features.Based on the established prediction model,the effects of tension and rolling force on mill vibration are analyzed.Taking the 5th stand of a cold mill in a steel mill as the research object,the innovative model is applied to predict the mill vibration for the first time.The experimental results show that the correlation coefficient(R^(2))of the model proposed in this paper is 92.5%,and the root-mean-square error(RMSE)is 0.0011,which significantly improves the modeling accuracy compared with the existing models.The proposed model is also suitable for the hot rolling process,which provides a new method for the prediction of strip rolling vibration.

Oxygen Reduction Reaction Activity of Fe-based Dual-Atom Catalysts with Different Local Configurations via Graph Neural Representation

The performance of proton exchange membrane fuel cells depends heavily on the oxygen reduction reaction(ORR)at the cathode,for which platinum-based catalysts are currently the standard.The high cost and limited availability of platinum have driven the search for alternative catalysts.While FeN4 single-atom catalysts have shown promising potential,their ORR activity needs to be further enhanced.In contrast,dual-atom catalysts(DACs)offer not only higher metal loading but also the ability to break the ORR scaling relations.However,the diverse local structures and tunable coordination environments of DACs create a vast chemical space,making large-scale computational screening challenging.In this study,we developed a graph neural network(GNN)-based framework to predict the ORR activity of Fe-based DACs,effectively addressing the challenges posed by variations in local catalyst structures.Our model,trained on a dataset of 180 catalysts,accurately predicted the Gibbs free energy of ORR intermediates and overpotentials,and identified 32 DACs with superior catalytic activity compared to FeN4 SAC.This approach not only advances the design of high-performance DACs,but also offers a powerful computational tool that can significantly reduce the time and cost of catalyst development,thereby accelerating the commercialization of fuel cell technologies.

Attention-Guided Sparse Adversarial Attacks with Gradient Dropout

Deep neural networks are extremely vulnerable to externalities from intentionally generated adversarial examples which are achieved by overlaying tiny noise on the clean images.However,most existing transfer-based attack methods are chosen to add perturbations on each pixel of the original image with the same weight,resulting in redundant noise in the adversarial examples,which makes them easier to be detected.Given this deliberation,a novel attentionguided sparse adversarial attack strategy with gradient dropout that can be readily incorporated with existing gradient-based methods is introduced to minimize the intensity and the scale of perturbations and ensure the effectiveness of adversarial examples at the same time.Specifically,in the gradient dropout phase,some relatively unimportant gradient information is randomly discarded to limit the intensity of the perturbation.In the attentionguided phase,the influence of each pixel on the model output is evaluated by using a soft mask-refined attention mechanism,and the perturbation of those pixels with smaller influence is limited to restrict the scale of the perturbation.After conducting thorough experiments on the NeurIPS 2017 adversarial dataset and the ILSVRC 2012 validation dataset,the proposed strategy holds the potential to significantly diminish the superfluous noise present in adversarial examples,all while keeping their attack efficacy intact.For instance,in attacks on adversarially trained models,upon the integration of the strategy,the average level of noise injected into images experiences a decline of 8.32%.However,the average attack success rate decreases by only 0.34%.Furthermore,the competence is possessed to substantially elevate the attack success rate by merely introducing a slight degree of perturbation.

A Hybrid Compensation Scheme for the Input Rate-Dependent Hysteresis of the Piezoelectric Ceramic Actuators

A hybrid compensation scheme for piezoelectric ceramic actuators(PEAs)is proposed.In the hybrid compensation scheme,the input rate-dependent hysteresis characteristics of the PEAs are compensated.The feedforward controller is a novel input rate-dependent neural network hysteresis inverse model,while the feedback controller is a proportion integration differentiation(PID)controller.In the proposed inverse model,an input ratedependent auxiliary inverse operator(RAIO)and output of the hysteresis construct the expanded input space(EIS)of the inverse model which transforms the hysteresis inverse with multi-valued mapping into single-valued mapping,and the wiping-out,rate-dependent and continuous properties of the RAIO are analyzed in theories.Based on the EIS method,a hysteresis neural network inverse model,namely the dynamic back propagation neural network(DBPNN)model,is established.Moreover,a hybrid compensation scheme for the PEAs is designed to compensate for the hysteresis.Finally,the proposed method,the conventional PID controller and the hybrid controller with the modified input rate-dependent Prandtl-Ishlinskii(MRPI)model are all applied in the experimental platform.Experimental results show that the proposed method has obvious superiorities in the performance of the system.

High-resolution reconstruction of the ablative RT instability flowfield via convolutional neural networks

High-resolution flow field data has important applications in meteorology,aerospace engineering,high-energy physics and other fields.Experiments and numerical simulations are two main ways to obtain high-resolution flow field data,while the high experiment cost and computing resources for simulation hinder the specificanalysis of flow field evolution.With the development of deep learning technology,convolutional neural networks areused to achieve high-resolution reconstruction of the flow field.In this paper,an ordinary convolutional neuralnetwork and a multi-time-path convolutional neural network are established for the ablative Rayleigh-Taylorinstability.These two methods can reconstruct the high-resolution flow field in just a few seconds,and further greatlyenrich the application of high-resolution reconstruction technology in fluid instability.Compared with the ordinaryconvolutional neural network,the multi-time-path convolutional neural network model has smaller error and canrestore more details of the flow field.The influence of low-resolution flow field data obtained by the two poolingmethods on the convolutional neural networks model is also discussed.

Rapid urban flood forecasting based on cellular automata and deep learning

[Objective]Urban floods are occurring more frequently because of global climate change and urbanization.Accordingly,urban rainstorm and flood forecasting has become a priority in urban hydrology research.However,two-dimensional hydrodynamic models execute calculations slowly,hindering the rapid simulation and forecasting of urban floods.To overcome this limitation and accelerate the speed and improve the accuracy of urban flood simulations and forecasting,numerical simulations and deep learning were combined to develop a more effective urban flood forecasting method.[Methods]Specifically,a cellular automata model was used to simulate the urban flood process and address the need to include a large number of datasets in the deep learning process.Meanwhile,to shorten the time required for urban flood forecasting,a convolutional neural network model was used to establish the mapping relationship between rainfall and inundation depth.[Results]The results show that the relative error of forecasting the maximum inundation depth in flood-prone locations is less than 10%,and the Nash efficiency coefficient of forecasting inundation depth series in flood-prone locations is greater than 0.75.[Conclusion]The result demonstrated that the proposed method could execute highly accurate simulations and quickly produce forecasts,illustrating its superiority as an urban flood forecasting technique.

Adaptive Bayesian inversion of pore water pressures based on artificial neural network : An earth dam case study

Most earth-dam failures are mainly due to seepage,and an accurate assessment of the permeability coefficient provides an indication to avoid a disaster.Parametric uncertainties are encountered in the seepage analysis,and may be reduced by an inverse procedure that calibrates the simulation results to observations on the real system being simulated.This work proposes an adaptive Bayesian inversion method solved using artificial neural network(ANN)based Markov Chain Monte Carlo simulation.The optimized surrogate model achieves a coefficient of determination at 0.98 by ANN with 247 samples,whereby the computational workload can be greatly reduced.It is also significant to balance the accuracy and efficiency of the ANN model by adaptively updating the sample database.The enrichment samples are obtained from the posterior distribution after iteration,which allows a more accurate and rapid manner to the target posterior.The method was then applied to the hydraulic analysis of an earth dam.After calibrating the global permeability coefficient of the earth dam with the pore water pressure at the downstream unsaturated location,it was validated by the pore water pressure monitoring values at the upstream saturated location.In addition,the uncertainty in the permeability coefficient was reduced,from 0.5 to 0.05.It is shown that the provision of adequate prior information is valuable for improving the efficiency of the Bayesian inversion.

Infrasound Event Classification Fusion Model Based on Multiscale SE-CNN and BiLSTM

The classification of infrasound events has considerable importance in improving the capability to identify the types of natural disasters.The traditional infrasound classification mainly relies on machine learning algorithms after artificial feature extraction.However,guaranteeing the effectiveness of the extracted features is difficult.The current trend focuses on using a convolution neural network to automatically extract features for classification.This method can be used to extract signal spatial features automatically through a convolution kernel;however,infrasound signals contain not only spatial information but also temporal information when used as a time series.These extracted temporal features are also crucial.If only a convolution neural network is used,then the time dependence of the infrasound sequence will be missed.Using long short-term memory networks can compensate for the missing time-series features but induces spatial feature information loss of the infrasound signal.A multiscale squeeze excitation–convolution neural network–bidirectional long short-term memory network infrasound event classification fusion model is proposed in this study to address these problems.This model automatically extracted temporal and spatial features,adaptively selected features,and also realized the fusion of the two types of features.Experimental results showed that the classification accuracy of the model was more than 98%,thus verifying the effectiveness and superiority of the proposed model.