Tunnel boring machines(TBMs)have been widely utilised in tunnel construction due to their high efficiency and reliability.Accurately predicting TBM performance can improve project time management,cost control,and risk...Tunnel boring machines(TBMs)have been widely utilised in tunnel construction due to their high efficiency and reliability.Accurately predicting TBM performance can improve project time management,cost control,and risk management.This study aims to use deep learning to develop real-time models for predicting the penetration rate(PR).The models are built using data from the Changsha metro project,and their performances are evaluated using unseen data from the Zhengzhou Metro project.In one-step forecast,the predicted penetration rate follows the trend of the measured penetration rate in both training and testing.The autoregressive integrated moving average(ARIMA)model is compared with the recurrent neural network(RNN)model.The results show that univariate models,which only consider historical penetration rate itself,perform better than multivariate models that take into account multiple geological and operational parameters(GEO and OP).Next,an RNN variant combining time series of penetration rate with the last-step geological and operational parameters is developed,and it performs better than other models.A sensitivity analysis shows that the penetration rate is the most important parameter,while other parameters have a smaller impact on time series forecasting.It is also found that smoothed data are easier to predict with high accuracy.Nevertheless,over-simplified data can lose real characteristics in time series.In conclusion,the RNN variant can accurately predict the next-step penetration rate,and data smoothing is crucial in time series forecasting.This study provides practical guidance for TBM performance forecasting in practical engineering.展开更多
Optical neural networks have significant advantages in terms of power consumption,parallelism,and high computing speed,which has intrigued extensive attention in both academic and engineering communities.It has been c...Optical neural networks have significant advantages in terms of power consumption,parallelism,and high computing speed,which has intrigued extensive attention in both academic and engineering communities.It has been considered as one of the powerful tools in promoting the fields of imaging processing and object recognition.However,the existing optical system architecture cannot be reconstructed to the realization of multi-functional artificial intelligence systems simultaneously.To push the development of this issue,we propose the pluggable diffractive neural networks(P-DNN),a general paradigm resorting to the cascaded metasurfaces,which can be applied to recognize various tasks by switching internal plug-ins.As the proof-of-principle,the recognition functions of six types of handwritten digits and six types of fashions are numerical simulated and experimental demonstrated at near-infrared regimes.Encouragingly,the proposed paradigm not only improves the flexibility of the optical neural networks but paves the new route for achieving high-speed,low-power and versatile artificial intelligence systems.展开更多
Immune changes and inflammatory responses have been identified as central events in the pathological process of spinal co rd injury.They can greatly affect nerve regeneration and functional recovery.However,there is s...Immune changes and inflammatory responses have been identified as central events in the pathological process of spinal co rd injury.They can greatly affect nerve regeneration and functional recovery.However,there is still limited understanding of the peripheral immune inflammato ry response in spinal cord inju ry.In this study.we obtained microRNA expression profiles from the peripheral blood of patients with spinal co rd injury using high-throughput sequencing.We also obtained the mRNA expression profile of spinal cord injury patients from the Gene Expression Omnibus(GEO)database(GSE151371).We identified 54 differentially expressed microRNAs and 1656 diffe rentially expressed genes using bioinformatics approaches.Functional enrichment analysis revealed that various common immune and inflammation-related signaling pathways,such as neutrophil extracellular trap formation pathway,T cell receptor signaling pathway,and nuclear factor-κB signal pathway,we re abnormally activated or inhibited in spinal cord inju ry patient samples.We applied an integrated strategy that combines weighted gene co-expression network analysis,LASSO logistic regression,and SVM-RFE algorithm and identified three biomarke rs associated with spinal cord injury:ANO10,BST1,and ZFP36L2.We verified the expression levels and diagnostic perfo rmance of these three genes in the original training dataset and clinical samples through the receiver operating characteristic curve.Quantitative polymerase chain reaction results showed that ANO20 and BST1 mRNA levels were increased and ZFP36L2 mRNA was decreased in the peripheral blood of spinal cord injury patients.We also constructed a small RNA-mRNA interaction network using Cytoscape.Additionally,we evaluated the proportion of 22 types of immune cells in the peripheral blood of spinal co rd injury patients using the CIBERSORT tool.The proportions of naive B cells,plasma cells,monocytes,and neutrophils were increased while the proportions of memory B cells,CD8^(+)T cells,resting natural killer cells,resting dendritic cells,and eosinophils were markedly decreased in spinal cord injury patients increased compared with healthy subjects,and ANO10,BST1 and ZFP26L2we re closely related to the proportion of certain immune cell types.The findings from this study provide new directions for the development of treatment strategies related to immune inflammation in spinal co rd inju ry and suggest that ANO10,BST2,and ZFP36L2 are potential biomarkers for spinal cord injury.The study was registe red in the Chinese Clinical Trial Registry(registration No.ChiCTR2200066985,December 12,2022).展开更多
Background In early adolescence,youth are highly prone to suicidal behaviours.Identifying modifiable risk factors during this critical phase is a priority to inform effective suicide prevention strategies.Aims To expl...Background In early adolescence,youth are highly prone to suicidal behaviours.Identifying modifiable risk factors during this critical phase is a priority to inform effective suicide prevention strategies.Aims To explore the risk and protective factors of suicidal behaviours(ie,suicidal ideation,plans and attempts)in early adolescence in China using a social-ecological perspective.Methods Using data from the cross-sectional project‘Healthy and Risky Behaviours Among Middle School Students in Anhui Province,China',stratified random cluster sampling was used to select 5724 middle school students who had completed self-report questionnaires in November 2020.Network analysis was employed to examine the correlates of suicidal ideation,plans and attempts at four levels,namely individual(sex,academic performance,serious physical llness/disability,history of self-harm,depression,impulsivity,sleep problems,resilience),family(family economic status,relationship with mother,relationship with father,family violence,childhood abuse,parental mental illness),school(relationship with teachers,relationship with classmates,school-bullying victimisation and perpetration)and social(social support,satisfaction with society).Results In total,37.9%,19.0%and 5.5%of the students reported suicidal ideation,plans and attempts in the past 6 months,respectively.The estimated network revealed that suicidal ideation,plans and attempts were collectively associated with a history of self-harm,sleep problems,childhood abuse,school bullying and victimisation.Centrality analysis indicated that the most influential nodes in the network were history of self-harm and childhood abuse.Notably,the network also showed unique correlates of suicidal ideation(sex,weight=0.60;impulsivity,weight=0.24;family violence,weight=0.17;relationship with teachers,weight=-0.03;school-bullying perpetration,weight=0.22),suicidal plans(social support,weight=-0.15)and suicidal attempts(relationship with mother,weight=-0.10;parental mental llness,weight=0.61).Conclusions This study identified the correlates of suicidal ideation,plans and attempts,and provided practical implications for suicide prevention for young adolescents in China.Firstly,this study highlighted the importance of joint interventions across multiple departments.Secondly,the common risk factors of suicidal ideation,plans and attempts were elucidated.Thirdly,this study proposed target interventions to address the unique influencing factors of suicidal ideation,plans and attempts.展开更多
A Wireless Sensor Network(WSN)is constructed with numerous sensors over geographical regions.The basic challenge experienced while designing WSN is in increasing the network lifetime and use of low energy.As sensor no...A Wireless Sensor Network(WSN)is constructed with numerous sensors over geographical regions.The basic challenge experienced while designing WSN is in increasing the network lifetime and use of low energy.As sensor nodes are resource constrained in nature,novel techniques are essential to improve lifetime of nodes in WSN.Nodes energy is considered as an important resource for sensor node which are battery powered based.In WSN,energy is consumed mainly while data is being transferred among nodes in the network.Several research works are carried out focusing on preserving energy of nodes in the network and made network to live longer.Moreover,this network is threatened by attacks like vampire attack where the network is loaded by fake traffic.Here,Dual Encoding Recurrent Neural network(DERNNet)is proposed for classifying the vampire nodes s node in the network.Moreover,the Grey Wolf Optimization(GWO)algorithm helps for transferring the data by determining best solutions to optimally select the aggregation points;thereby maximizing battery/lifetime of the network nodes.The proposed method is evaluated with three standard approaches namely Knowledge and Intrusion Detection based Secure Atom Search Routing(KIDSASR),Risk-aware Reputation-based Trust(RaRTrust)model and Activation Function-based Trusted Neighbor Selection(AF-TNS)in terms of various parameters.These existing methods may lead to wastage of energy due to vampire attack,which further reduce the lifetime and increase average energy consumed in the network.Hence,the proposed DERNNet method achieves 31.4%of routing overhead,23%of end-to-end delay,78.6%of energy efficiency,94.8%of throughput,28.2%of average latency,92.4%of packet delivery ratio,85.2%of network lifetime,and 94.3%of classification accuracy.展开更多
Phishing attacks present a persistent and evolving threat in the cybersecurity land-scape,necessitating the development of more sophisticated detection methods.Traditional machine learning approaches to phishing detec...Phishing attacks present a persistent and evolving threat in the cybersecurity land-scape,necessitating the development of more sophisticated detection methods.Traditional machine learning approaches to phishing detection have relied heavily on feature engineering and have often fallen short in adapting to the dynamically changing patterns of phishingUniformResource Locator(URLs).Addressing these challenge,we introduce a framework that integrates the sequential data processing strengths of a Recurrent Neural Network(RNN)with the hyperparameter optimization prowess of theWhale Optimization Algorithm(WOA).Ourmodel capitalizes on an extensive Kaggle dataset,featuring over 11,000 URLs,each delineated by 30 attributes.The WOA’s hyperparameter optimization enhances the RNN’s performance,evidenced by a meticulous validation process.The results,encapsulated in precision,recall,and F1-score metrics,surpass baseline models,achieving an overall accuracy of 92%.This study not only demonstrates the RNN’s proficiency in learning complex patterns but also underscores the WOA’s effectiveness in refining machine learning models for the critical task of phishing detection.展开更多
Convolutional neural networks depend on deep network architectures to extract accurate information for image super‐resolution.However,obtained information of these con-volutional neural networks cannot completely exp...Convolutional neural networks depend on deep network architectures to extract accurate information for image super‐resolution.However,obtained information of these con-volutional neural networks cannot completely express predicted high‐quality images for complex scenes.A dynamic network for image super‐resolution(DSRNet)is presented,which contains a residual enhancement block,wide enhancement block,feature refine-ment block and construction block.The residual enhancement block is composed of a residual enhanced architecture to facilitate hierarchical features for image super‐resolution.To enhance robustness of obtained super‐resolution model for complex scenes,a wide enhancement block achieves a dynamic architecture to learn more robust information to enhance applicability of an obtained super‐resolution model for varying scenes.To prevent interference of components in a wide enhancement block,a refine-ment block utilises a stacked architecture to accurately learn obtained features.Also,a residual learning operation is embedded in the refinement block to prevent long‐term dependency problem.Finally,a construction block is responsible for reconstructing high‐quality images.Designed heterogeneous architecture can not only facilitate richer structural information,but also be lightweight,which is suitable for mobile digital devices.Experimental results show that our method is more competitive in terms of performance,recovering time of image super‐resolution and complexity.The code of DSRNet can be obtained at https://github.com/hellloxiaotian/DSRNet.展开更多
Accurately predicting fluid forces acting on the sur-face of a structure is crucial in engineering design.However,this task becomes particularly challenging in turbulent flow,due to the complex and irregular changes i...Accurately predicting fluid forces acting on the sur-face of a structure is crucial in engineering design.However,this task becomes particularly challenging in turbulent flow,due to the complex and irregular changes in the flow field.In this study,we propose a novel deep learning method,named mapping net-work-coordinated stacked gated recurrent units(MSU),for pre-dicting pressure on a circular cylinder from velocity data.Specifi-cally,our coordinated learning strategy is designed to extract the most critical velocity point for prediction,a process that has not been explored before.In our experiments,MSU extracts one point from a velocity field containing 121 points and utilizes this point to accurately predict 100 pressure points on the cylinder.This method significantly reduces the workload of data measure-ment in practical engineering applications.Our experimental results demonstrate that MSU predictions are highly similar to the real turbulent data in both spatio-temporal and individual aspects.Furthermore,the comparison results show that MSU predicts more precise results,even outperforming models that use all velocity field points.Compared with state-of-the-art methods,MSU has an average improvement of more than 45%in various indicators such as root mean square error(RMSE).Through comprehensive and authoritative physical verification,we estab-lished that MSU’s prediction results closely align with pressure field data obtained in real turbulence fields.This confirmation underscores the considerable potential of MSU for practical applications in real engineering scenarios.The code is available at https://github.com/zhangzm0128/MSU.展开更多
Network traffic identification is critical for maintaining network security and further meeting various demands of network applications.However,network traffic data typically possesses high dimensionality and complexi...Network traffic identification is critical for maintaining network security and further meeting various demands of network applications.However,network traffic data typically possesses high dimensionality and complexity,leading to practical problems in traffic identification data analytics.Since the original Dung Beetle Optimizer(DBO)algorithm,Grey Wolf Optimization(GWO)algorithm,Whale Optimization Algorithm(WOA),and Particle Swarm Optimization(PSO)algorithm have the shortcomings of slow convergence and easily fall into the local optimal solution,an Improved Dung Beetle Optimizer(IDBO)algorithm is proposed for network traffic identification.Firstly,the Sobol sequence is utilized to initialize the dung beetle population,laying the foundation for finding the global optimal solution.Next,an integration of levy flight and golden sine strategy is suggested to give dung beetles a greater probability of exploring unvisited areas,escaping from the local optimal solution,and converging more effectively towards a global optimal solution.Finally,an adaptive weight factor is utilized to enhance the search capabilities of the original DBO algorithm and accelerate convergence.With the improvements above,the proposed IDBO algorithm is then applied to traffic identification data analytics and feature selection,as so to find the optimal subset for K-Nearest Neighbor(KNN)classification.The simulation experiments use the CICIDS2017 dataset to verify the effectiveness of the proposed IDBO algorithm and compare it with the original DBO,GWO,WOA,and PSO algorithms.The experimental results show that,compared with other algorithms,the accuracy and recall are improved by 1.53%and 0.88%in binary classification,and the Distributed Denial of Service(DDoS)class identification is the most effective in multi-classification,with an improvement of 5.80%and 0.33%for accuracy and recall,respectively.Therefore,the proposed IDBO algorithm is effective in increasing the efficiency of traffic identification and solving the problem of the original DBO algorithm that converges slowly and falls into the local optimal solution when dealing with high-dimensional data analytics and feature selection for network traffic identification.展开更多
模型深度的不断增加和处理序列长度的不一致对循环神经网络在不同处理器上的性能优化提出巨大挑战。针对自主研制的长向量处理器FT-M7032,实现了一个高效的循环神经网络加速引擎。该引擎采用行优先矩阵向量乘算法和数据感知的多核并行方...模型深度的不断增加和处理序列长度的不一致对循环神经网络在不同处理器上的性能优化提出巨大挑战。针对自主研制的长向量处理器FT-M7032,实现了一个高效的循环神经网络加速引擎。该引擎采用行优先矩阵向量乘算法和数据感知的多核并行方式,提高矩阵向量乘的计算效率;采用两级内核融合优化方法降低临时数据传输的开销;采用手写汇编优化多种算子,进一步挖掘长向量处理器的性能潜力。实验表明,长向量处理器循环神经网络推理引擎可获得较高性能,相较于多核ARM CPU以及Intel Golden CPU,类循环神经网络模型长短记忆网络可获得最高62.68倍和3.12倍的性能加速。展开更多
A network intrusion detection system is critical for cyber security against llegitimate attacks.In terms of feature perspectives,network traffic may include a variety of elements such as attack reference,attack type,a...A network intrusion detection system is critical for cyber security against llegitimate attacks.In terms of feature perspectives,network traffic may include a variety of elements such as attack reference,attack type,a subcategory of attack,host information,malicious scripts,etc.In terms of network perspectives,network traffic may contain an imbalanced number of harmful attacks when compared to normal traffic.It is challenging to identify a specific attack due to complex features and data imbalance issues.To address these issues,this paper proposes an Intrusion Detection System using transformer-based transfer learning for Imbalanced Network Traffic(IDS-INT).IDS-INT uses transformer-based transfer learning to learn feature interactions in both network feature representation and imbalanced data.First,detailed information about each type of attack is gathered from network interaction descriptions,which include network nodes,attack type,reference,host information,etc.Second,the transformer-based transfer learning approach is developed to learn detailed feature representation using their semantic anchors.Third,the Synthetic Minority Oversampling Technique(SMOTE)is implemented to balance abnormal traffic and detect minority attacks.Fourth,the Convolution Neural Network(CNN)model is designed to extract deep features from the balanced network traffic.Finally,the hybrid approach of the CNN-Long Short-Term Memory(CNN-LSTM)model is developed to detect different types of attacks from the deep features.Detailed experiments are conducted to test the proposed approach using three standard datasets,i.e.,UNsWNB15,CIC-IDS2017,and NSL-KDD.An explainable AI approach is implemented to interpret the proposed method and develop a trustable model.展开更多
The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedd...The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedded devices.In order to reduce the complexity and overhead of deploying neural networks on Integeronly hardware,most current quantization methods use a symmetric quantization mapping strategy to quantize a floating-point neural network into an integer network.However,although symmetric quantization has the advantage of easier implementation,it is sub-optimal for cases where the range could be skewed and not symmetric.This often comes at the cost of lower accuracy.This paper proposed an activation redistribution-based hybrid asymmetric quantizationmethod for neural networks.The proposedmethod takes data distribution into consideration and can resolve the contradiction between the quantization accuracy and the ease of implementation,balance the trade-off between clipping range and quantization resolution,and thus improve the accuracy of the quantized neural network.The experimental results indicate that the accuracy of the proposed method is 2.02%and 5.52%higher than the traditional symmetric quantization method for classification and detection tasks,respectively.The proposed method paves the way for computationally intensive neural network models to be deployed on devices with limited computing resources.Codes will be available on https://github.com/ycjcy/Hybrid-Asymmetric-Quantization.展开更多
Popular fermented golden pomfret(Trachinotus ovatus)is prepared via spontaneous fermentation;however,the mechanisms underlying the regulation of its flavor development remain unclear.This study shows the roles of the ...Popular fermented golden pomfret(Trachinotus ovatus)is prepared via spontaneous fermentation;however,the mechanisms underlying the regulation of its flavor development remain unclear.This study shows the roles of the complex microbiota and the dynamic changes in microbial community and flavor compounds during fish fermentation.Single-molecule real-time sequencing and molecular networking analysis revealed the correlations among different microbial genera and the relationships between microbial taxa and volatile compounds.Mechanisms underlying flavor development were also elucidated via KEGG based functional annotations.Clostridium,Shewanella,and Staphylococcus were the dominant microbial genera.Forty-nine volatile compounds were detected in the fermented fish samples,with thirteen identified as characteristic volatile compounds(ROAV>1).Volatile profiles resulted from the interactions among the microorganisms and derived enzymes,with the main metabolic pathways being amino acid biosynthesis/metabolism,carbon metabolism,and glycolysis/gluconeogenesis.This study demonstrated the approaches for distinguishing key microbiota associated with volatile compounds and monitoring the industrial production of high-quality fermented fish products.展开更多
BACKGROUND Currently,traditional Chinese medicine(TCM)formulas are commonly being used as adjunctive therapy for ulcerative colitis in China.Network meta-analysis,a quantitative and comprehensive analytical method,can...BACKGROUND Currently,traditional Chinese medicine(TCM)formulas are commonly being used as adjunctive therapy for ulcerative colitis in China.Network meta-analysis,a quantitative and comprehensive analytical method,can systematically compare the effects of different adjunctive treatment options for ulcerative colitis,providing scientific evidence for clinical decision-making.AIM To evaluate the clinical efficacy and safety of commonly used TCM for the treatment of ulcerative colitis(UC)in clinical practice through a network metaanalysis.METHODS Clinical randomized controlled trials of these TCM formulas used for the adjuvant treatment of UC were searched from the establishment of the databases to July 1,2022.Studies that met the inclusion criteria were screened and evaluated for literature quality and risk of bias according to the Cochrane 5.1 standard.The methodological quality of the studies was assessed using ReviewManager(RevMan)5.4,and a funnel plot was constructed to test for publication bias.ADDIS 1.16 statistical software was used to perform statistical analysis of the treatment measures and derive the network relationship and ranking diagrams of the various intervention measures.RESULTS A total of 64 randomized controlled trials involving 5456 patients with UC were included in this study.The adjuvant treatment of UC using five TCM formulations was able to improve the clinical outcome of the patients.Adjuvant treatment with Baitouweng decoction(BTWT)showed a significant effect[mean difference=36.22,95%confidence interval(CI):7.63 to 65.76].For the reduction of tumor necrosis factor in patients with UC,adjunctive therapy with BTWT(mean difference=−9.55,95%CI:−17.89 to−1.41),Shenlingbaizhu powder[SLBZS;odds ratio(OR)=0.19,95%CI:0.08 to 0.39],and Shaoyao decoction(OR=−23.02,95%CI:−33.64 to−13.14)was effective.Shaoyao decoction was more effective than BTWT(OR=0.12,95%CI:0.03 to 0.39),SLBZS(OR=0.19,95%CI:0.08 to 0.39),and Xi Lei powder(OR=0.34,95%CI:0.13 to 0.81)in reducing tumor necrosis factor and the recurrence rate of UC.CONCLUSION TCM combined with mesalazine is more effective than mesalazine alone in the treatment of UC.展开更多
A large number of network security breaches in IoT networks have demonstrated the unreliability of current Network Intrusion Detection Systems(NIDSs).Consequently,network interruptions and loss of sensitive data have ...A large number of network security breaches in IoT networks have demonstrated the unreliability of current Network Intrusion Detection Systems(NIDSs).Consequently,network interruptions and loss of sensitive data have occurred,which led to an active research area for improving NIDS technologies.In an analysis of related works,it was observed that most researchers aim to obtain better classification results by using a set of untried combinations of Feature Reduction(FR)and Machine Learning(ML)techniques on NIDS datasets.However,these datasets are different in feature sets,attack types,and network design.Therefore,this paper aims to discover whether these techniques can be generalised across various datasets.Six ML models are utilised:a Deep Feed Forward(DFF),Convolutional Neural Network(CNN),Recurrent Neural Network(RNN),Decision Tree(DT),Logistic Regression(LR),and Naive Bayes(NB).The accuracy of three Feature Extraction(FE)algorithms is detected;Principal Component Analysis(PCA),Auto-encoder(AE),and Linear Discriminant Analysis(LDA),are evaluated using three benchmark datasets:UNSW-NB15,ToN-IoT and CSE-CIC-IDS2018.Although PCA and AE algorithms have been widely used,the determination of their optimal number of extracted dimensions has been overlooked.The results indicate that no clear FE method or ML model can achieve the best scores for all datasets.The optimal number of extracted dimensions has been identified for each dataset,and LDA degrades the performance of the ML models on two datasets.The variance is used to analyse the extracted dimensions of LDA and PCA.Finally,this paper concludes that the choice of datasets significantly alters the performance of the applied techniques.We believe that a universal(benchmark)feature set is needed to facilitate further advancement and progress of research in this field.展开更多
文摘Tunnel boring machines(TBMs)have been widely utilised in tunnel construction due to their high efficiency and reliability.Accurately predicting TBM performance can improve project time management,cost control,and risk management.This study aims to use deep learning to develop real-time models for predicting the penetration rate(PR).The models are built using data from the Changsha metro project,and their performances are evaluated using unseen data from the Zhengzhou Metro project.In one-step forecast,the predicted penetration rate follows the trend of the measured penetration rate in both training and testing.The autoregressive integrated moving average(ARIMA)model is compared with the recurrent neural network(RNN)model.The results show that univariate models,which only consider historical penetration rate itself,perform better than multivariate models that take into account multiple geological and operational parameters(GEO and OP).Next,an RNN variant combining time series of penetration rate with the last-step geological and operational parameters is developed,and it performs better than other models.A sensitivity analysis shows that the penetration rate is the most important parameter,while other parameters have a smaller impact on time series forecasting.It is also found that smoothed data are easier to predict with high accuracy.Nevertheless,over-simplified data can lose real characteristics in time series.In conclusion,the RNN variant can accurately predict the next-step penetration rate,and data smoothing is crucial in time series forecasting.This study provides practical guidance for TBM performance forecasting in practical engineering.
基金The authors acknowledge the funding provided by the National Key R&D Program of China(2021YFA1401200)Beijing Outstanding Young Scientist Program(BJJWZYJH01201910007022)+2 种基金National Natural Science Foundation of China(No.U21A20140,No.92050117,No.62005017)programBeijing Municipal Science&Technology Commission,Administrative Commission of Zhongguancun Science Park(No.Z211100004821009)This work was supported by the Synergetic Extreme Condition User Facility(SECUF).
文摘Optical neural networks have significant advantages in terms of power consumption,parallelism,and high computing speed,which has intrigued extensive attention in both academic and engineering communities.It has been considered as one of the powerful tools in promoting the fields of imaging processing and object recognition.However,the existing optical system architecture cannot be reconstructed to the realization of multi-functional artificial intelligence systems simultaneously.To push the development of this issue,we propose the pluggable diffractive neural networks(P-DNN),a general paradigm resorting to the cascaded metasurfaces,which can be applied to recognize various tasks by switching internal plug-ins.As the proof-of-principle,the recognition functions of six types of handwritten digits and six types of fashions are numerical simulated and experimental demonstrated at near-infrared regimes.Encouragingly,the proposed paradigm not only improves the flexibility of the optical neural networks but paves the new route for achieving high-speed,low-power and versatile artificial intelligence systems.
基金supported by the Notional Natural Science Foundation of China,No.81960417 (to JX)Guangxi Key Research and Development Program,No.GuiKeA B20159027 (to JX)the Natural Science Foundation of Guangxi Zhuang Autonomous Region,No.2022GXNSFBA035545 (to YG)。
文摘Immune changes and inflammatory responses have been identified as central events in the pathological process of spinal co rd injury.They can greatly affect nerve regeneration and functional recovery.However,there is still limited understanding of the peripheral immune inflammato ry response in spinal cord inju ry.In this study.we obtained microRNA expression profiles from the peripheral blood of patients with spinal co rd injury using high-throughput sequencing.We also obtained the mRNA expression profile of spinal cord injury patients from the Gene Expression Omnibus(GEO)database(GSE151371).We identified 54 differentially expressed microRNAs and 1656 diffe rentially expressed genes using bioinformatics approaches.Functional enrichment analysis revealed that various common immune and inflammation-related signaling pathways,such as neutrophil extracellular trap formation pathway,T cell receptor signaling pathway,and nuclear factor-κB signal pathway,we re abnormally activated or inhibited in spinal cord inju ry patient samples.We applied an integrated strategy that combines weighted gene co-expression network analysis,LASSO logistic regression,and SVM-RFE algorithm and identified three biomarke rs associated with spinal cord injury:ANO10,BST1,and ZFP36L2.We verified the expression levels and diagnostic perfo rmance of these three genes in the original training dataset and clinical samples through the receiver operating characteristic curve.Quantitative polymerase chain reaction results showed that ANO20 and BST1 mRNA levels were increased and ZFP36L2 mRNA was decreased in the peripheral blood of spinal cord injury patients.We also constructed a small RNA-mRNA interaction network using Cytoscape.Additionally,we evaluated the proportion of 22 types of immune cells in the peripheral blood of spinal co rd injury patients using the CIBERSORT tool.The proportions of naive B cells,plasma cells,monocytes,and neutrophils were increased while the proportions of memory B cells,CD8^(+)T cells,resting natural killer cells,resting dendritic cells,and eosinophils were markedly decreased in spinal cord injury patients increased compared with healthy subjects,and ANO10,BST1 and ZFP26L2we re closely related to the proportion of certain immune cell types.The findings from this study provide new directions for the development of treatment strategies related to immune inflammation in spinal co rd inju ry and suggest that ANO10,BST2,and ZFP36L2 are potential biomarkers for spinal cord injury.The study was registe red in the Chinese Clinical Trial Registry(registration No.ChiCTR2200066985,December 12,2022).
文摘Background In early adolescence,youth are highly prone to suicidal behaviours.Identifying modifiable risk factors during this critical phase is a priority to inform effective suicide prevention strategies.Aims To explore the risk and protective factors of suicidal behaviours(ie,suicidal ideation,plans and attempts)in early adolescence in China using a social-ecological perspective.Methods Using data from the cross-sectional project‘Healthy and Risky Behaviours Among Middle School Students in Anhui Province,China',stratified random cluster sampling was used to select 5724 middle school students who had completed self-report questionnaires in November 2020.Network analysis was employed to examine the correlates of suicidal ideation,plans and attempts at four levels,namely individual(sex,academic performance,serious physical llness/disability,history of self-harm,depression,impulsivity,sleep problems,resilience),family(family economic status,relationship with mother,relationship with father,family violence,childhood abuse,parental mental illness),school(relationship with teachers,relationship with classmates,school-bullying victimisation and perpetration)and social(social support,satisfaction with society).Results In total,37.9%,19.0%and 5.5%of the students reported suicidal ideation,plans and attempts in the past 6 months,respectively.The estimated network revealed that suicidal ideation,plans and attempts were collectively associated with a history of self-harm,sleep problems,childhood abuse,school bullying and victimisation.Centrality analysis indicated that the most influential nodes in the network were history of self-harm and childhood abuse.Notably,the network also showed unique correlates of suicidal ideation(sex,weight=0.60;impulsivity,weight=0.24;family violence,weight=0.17;relationship with teachers,weight=-0.03;school-bullying perpetration,weight=0.22),suicidal plans(social support,weight=-0.15)and suicidal attempts(relationship with mother,weight=-0.10;parental mental llness,weight=0.61).Conclusions This study identified the correlates of suicidal ideation,plans and attempts,and provided practical implications for suicide prevention for young adolescents in China.Firstly,this study highlighted the importance of joint interventions across multiple departments.Secondly,the common risk factors of suicidal ideation,plans and attempts were elucidated.Thirdly,this study proposed target interventions to address the unique influencing factors of suicidal ideation,plans and attempts.
文摘A Wireless Sensor Network(WSN)is constructed with numerous sensors over geographical regions.The basic challenge experienced while designing WSN is in increasing the network lifetime and use of low energy.As sensor nodes are resource constrained in nature,novel techniques are essential to improve lifetime of nodes in WSN.Nodes energy is considered as an important resource for sensor node which are battery powered based.In WSN,energy is consumed mainly while data is being transferred among nodes in the network.Several research works are carried out focusing on preserving energy of nodes in the network and made network to live longer.Moreover,this network is threatened by attacks like vampire attack where the network is loaded by fake traffic.Here,Dual Encoding Recurrent Neural network(DERNNet)is proposed for classifying the vampire nodes s node in the network.Moreover,the Grey Wolf Optimization(GWO)algorithm helps for transferring the data by determining best solutions to optimally select the aggregation points;thereby maximizing battery/lifetime of the network nodes.The proposed method is evaluated with three standard approaches namely Knowledge and Intrusion Detection based Secure Atom Search Routing(KIDSASR),Risk-aware Reputation-based Trust(RaRTrust)model and Activation Function-based Trusted Neighbor Selection(AF-TNS)in terms of various parameters.These existing methods may lead to wastage of energy due to vampire attack,which further reduce the lifetime and increase average energy consumed in the network.Hence,the proposed DERNNet method achieves 31.4%of routing overhead,23%of end-to-end delay,78.6%of energy efficiency,94.8%of throughput,28.2%of average latency,92.4%of packet delivery ratio,85.2%of network lifetime,and 94.3%of classification accuracy.
基金Princess Nourah bint Abdulrahman University Researchers Supporting Project number(PNURSP2024R 343)PrincessNourah bint Abdulrahman University,Riyadh,Saudi ArabiaDeanship of Scientific Research at Northern Border University,Arar,Kingdom of Saudi Arabia,for funding this researchwork through the project number“NBU-FFR-2024-1092-02”.
文摘Phishing attacks present a persistent and evolving threat in the cybersecurity land-scape,necessitating the development of more sophisticated detection methods.Traditional machine learning approaches to phishing detection have relied heavily on feature engineering and have often fallen short in adapting to the dynamically changing patterns of phishingUniformResource Locator(URLs).Addressing these challenge,we introduce a framework that integrates the sequential data processing strengths of a Recurrent Neural Network(RNN)with the hyperparameter optimization prowess of theWhale Optimization Algorithm(WOA).Ourmodel capitalizes on an extensive Kaggle dataset,featuring over 11,000 URLs,each delineated by 30 attributes.The WOA’s hyperparameter optimization enhances the RNN’s performance,evidenced by a meticulous validation process.The results,encapsulated in precision,recall,and F1-score metrics,surpass baseline models,achieving an overall accuracy of 92%.This study not only demonstrates the RNN’s proficiency in learning complex patterns but also underscores the WOA’s effectiveness in refining machine learning models for the critical task of phishing detection.
基金the TCL Science and Technology Innovation Fundthe Youth Science and Technology Talent Promotion Project of Jiangsu Association for Science and Technology,Grant/Award Number:JSTJ‐2023‐017+4 种基金Shenzhen Municipal Science and Technology Innovation Council,Grant/Award Number:JSGG20220831105002004National Natural Science Foundation of China,Grant/Award Number:62201468Postdoctoral Research Foundation of China,Grant/Award Number:2022M722599the Fundamental Research Funds for the Central Universities,Grant/Award Number:D5000210966the Guangdong Basic and Applied Basic Research Foundation,Grant/Award Number:2021A1515110079。
文摘Convolutional neural networks depend on deep network architectures to extract accurate information for image super‐resolution.However,obtained information of these con-volutional neural networks cannot completely express predicted high‐quality images for complex scenes.A dynamic network for image super‐resolution(DSRNet)is presented,which contains a residual enhancement block,wide enhancement block,feature refine-ment block and construction block.The residual enhancement block is composed of a residual enhanced architecture to facilitate hierarchical features for image super‐resolution.To enhance robustness of obtained super‐resolution model for complex scenes,a wide enhancement block achieves a dynamic architecture to learn more robust information to enhance applicability of an obtained super‐resolution model for varying scenes.To prevent interference of components in a wide enhancement block,a refine-ment block utilises a stacked architecture to accurately learn obtained features.Also,a residual learning operation is embedded in the refinement block to prevent long‐term dependency problem.Finally,a construction block is responsible for reconstructing high‐quality images.Designed heterogeneous architecture can not only facilitate richer structural information,but also be lightweight,which is suitable for mobile digital devices.Experimental results show that our method is more competitive in terms of performance,recovering time of image super‐resolution and complexity.The code of DSRNet can be obtained at https://github.com/hellloxiaotian/DSRNet.
基金supported by the Japan Society for the Promotion of Science(JSPS)KAKENHI(JP22H03643)Japan Science and Technology Agency(JST)Support for Pioneering Research Initiated by the Next Generation(SPRING)(JPMJSP2145)+2 种基金JST Through the Establishment of University Fellowships Towards the Creation of Science Technology Innovation(JPMJFS2115)the National Natural Science Foundation of China(52078382)the State Key Laboratory of Disaster Reduction in Civil Engineering(CE19-A-01)。
文摘Accurately predicting fluid forces acting on the sur-face of a structure is crucial in engineering design.However,this task becomes particularly challenging in turbulent flow,due to the complex and irregular changes in the flow field.In this study,we propose a novel deep learning method,named mapping net-work-coordinated stacked gated recurrent units(MSU),for pre-dicting pressure on a circular cylinder from velocity data.Specifi-cally,our coordinated learning strategy is designed to extract the most critical velocity point for prediction,a process that has not been explored before.In our experiments,MSU extracts one point from a velocity field containing 121 points and utilizes this point to accurately predict 100 pressure points on the cylinder.This method significantly reduces the workload of data measure-ment in practical engineering applications.Our experimental results demonstrate that MSU predictions are highly similar to the real turbulent data in both spatio-temporal and individual aspects.Furthermore,the comparison results show that MSU predicts more precise results,even outperforming models that use all velocity field points.Compared with state-of-the-art methods,MSU has an average improvement of more than 45%in various indicators such as root mean square error(RMSE).Through comprehensive and authoritative physical verification,we estab-lished that MSU’s prediction results closely align with pressure field data obtained in real turbulence fields.This confirmation underscores the considerable potential of MSU for practical applications in real engineering scenarios.The code is available at https://github.com/zhangzm0128/MSU.
基金supported by the National Natural Science Foundation of China under Grant 61602162the Hubei Provincial Science and Technology Plan Project under Grant 2023BCB041.
文摘Network traffic identification is critical for maintaining network security and further meeting various demands of network applications.However,network traffic data typically possesses high dimensionality and complexity,leading to practical problems in traffic identification data analytics.Since the original Dung Beetle Optimizer(DBO)algorithm,Grey Wolf Optimization(GWO)algorithm,Whale Optimization Algorithm(WOA),and Particle Swarm Optimization(PSO)algorithm have the shortcomings of slow convergence and easily fall into the local optimal solution,an Improved Dung Beetle Optimizer(IDBO)algorithm is proposed for network traffic identification.Firstly,the Sobol sequence is utilized to initialize the dung beetle population,laying the foundation for finding the global optimal solution.Next,an integration of levy flight and golden sine strategy is suggested to give dung beetles a greater probability of exploring unvisited areas,escaping from the local optimal solution,and converging more effectively towards a global optimal solution.Finally,an adaptive weight factor is utilized to enhance the search capabilities of the original DBO algorithm and accelerate convergence.With the improvements above,the proposed IDBO algorithm is then applied to traffic identification data analytics and feature selection,as so to find the optimal subset for K-Nearest Neighbor(KNN)classification.The simulation experiments use the CICIDS2017 dataset to verify the effectiveness of the proposed IDBO algorithm and compare it with the original DBO,GWO,WOA,and PSO algorithms.The experimental results show that,compared with other algorithms,the accuracy and recall are improved by 1.53%and 0.88%in binary classification,and the Distributed Denial of Service(DDoS)class identification is the most effective in multi-classification,with an improvement of 5.80%and 0.33%for accuracy and recall,respectively.Therefore,the proposed IDBO algorithm is effective in increasing the efficiency of traffic identification and solving the problem of the original DBO algorithm that converges slowly and falls into the local optimal solution when dealing with high-dimensional data analytics and feature selection for network traffic identification.
文摘模型深度的不断增加和处理序列长度的不一致对循环神经网络在不同处理器上的性能优化提出巨大挑战。针对自主研制的长向量处理器FT-M7032,实现了一个高效的循环神经网络加速引擎。该引擎采用行优先矩阵向量乘算法和数据感知的多核并行方式,提高矩阵向量乘的计算效率;采用两级内核融合优化方法降低临时数据传输的开销;采用手写汇编优化多种算子,进一步挖掘长向量处理器的性能潜力。实验表明,长向量处理器循环神经网络推理引擎可获得较高性能,相较于多核ARM CPU以及Intel Golden CPU,类循环神经网络模型长短记忆网络可获得最高62.68倍和3.12倍的性能加速。
文摘A network intrusion detection system is critical for cyber security against llegitimate attacks.In terms of feature perspectives,network traffic may include a variety of elements such as attack reference,attack type,a subcategory of attack,host information,malicious scripts,etc.In terms of network perspectives,network traffic may contain an imbalanced number of harmful attacks when compared to normal traffic.It is challenging to identify a specific attack due to complex features and data imbalance issues.To address these issues,this paper proposes an Intrusion Detection System using transformer-based transfer learning for Imbalanced Network Traffic(IDS-INT).IDS-INT uses transformer-based transfer learning to learn feature interactions in both network feature representation and imbalanced data.First,detailed information about each type of attack is gathered from network interaction descriptions,which include network nodes,attack type,reference,host information,etc.Second,the transformer-based transfer learning approach is developed to learn detailed feature representation using their semantic anchors.Third,the Synthetic Minority Oversampling Technique(SMOTE)is implemented to balance abnormal traffic and detect minority attacks.Fourth,the Convolution Neural Network(CNN)model is designed to extract deep features from the balanced network traffic.Finally,the hybrid approach of the CNN-Long Short-Term Memory(CNN-LSTM)model is developed to detect different types of attacks from the deep features.Detailed experiments are conducted to test the proposed approach using three standard datasets,i.e.,UNsWNB15,CIC-IDS2017,and NSL-KDD.An explainable AI approach is implemented to interpret the proposed method and develop a trustable model.
基金The Qian Xuesen Youth Innovation Foundation from China Aerospace Science and Technology Corporation(Grant Number 2022JY51).
文摘The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedded devices.In order to reduce the complexity and overhead of deploying neural networks on Integeronly hardware,most current quantization methods use a symmetric quantization mapping strategy to quantize a floating-point neural network into an integer network.However,although symmetric quantization has the advantage of easier implementation,it is sub-optimal for cases where the range could be skewed and not symmetric.This often comes at the cost of lower accuracy.This paper proposed an activation redistribution-based hybrid asymmetric quantizationmethod for neural networks.The proposedmethod takes data distribution into consideration and can resolve the contradiction between the quantization accuracy and the ease of implementation,balance the trade-off between clipping range and quantization resolution,and thus improve the accuracy of the quantized neural network.The experimental results indicate that the accuracy of the proposed method is 2.02%and 5.52%higher than the traditional symmetric quantization method for classification and detection tasks,respectively.The proposed method paves the way for computationally intensive neural network models to be deployed on devices with limited computing resources.Codes will be available on https://github.com/ycjcy/Hybrid-Asymmetric-Quantization.
基金supported by the National Natural Science Foundation of China(32001733)the Earmarked fund for CARS(CARS-47)+3 种基金Guangxi Natural Science Foundation Program(2021GXNSFAA196023)Guangdong Basic and Applied Basic Research Foundation(2021A1515010833)Young Talent Support Project of Guangzhou Association for Science and Technology(QT20220101142)the Special Scientific Research Funds for Central Non-profit Institutes,Chinese Academy of Fishery Sciences(2020TD69)。
文摘Popular fermented golden pomfret(Trachinotus ovatus)is prepared via spontaneous fermentation;however,the mechanisms underlying the regulation of its flavor development remain unclear.This study shows the roles of the complex microbiota and the dynamic changes in microbial community and flavor compounds during fish fermentation.Single-molecule real-time sequencing and molecular networking analysis revealed the correlations among different microbial genera and the relationships between microbial taxa and volatile compounds.Mechanisms underlying flavor development were also elucidated via KEGG based functional annotations.Clostridium,Shewanella,and Staphylococcus were the dominant microbial genera.Forty-nine volatile compounds were detected in the fermented fish samples,with thirteen identified as characteristic volatile compounds(ROAV>1).Volatile profiles resulted from the interactions among the microorganisms and derived enzymes,with the main metabolic pathways being amino acid biosynthesis/metabolism,carbon metabolism,and glycolysis/gluconeogenesis.This study demonstrated the approaches for distinguishing key microbiota associated with volatile compounds and monitoring the industrial production of high-quality fermented fish products.
文摘BACKGROUND Currently,traditional Chinese medicine(TCM)formulas are commonly being used as adjunctive therapy for ulcerative colitis in China.Network meta-analysis,a quantitative and comprehensive analytical method,can systematically compare the effects of different adjunctive treatment options for ulcerative colitis,providing scientific evidence for clinical decision-making.AIM To evaluate the clinical efficacy and safety of commonly used TCM for the treatment of ulcerative colitis(UC)in clinical practice through a network metaanalysis.METHODS Clinical randomized controlled trials of these TCM formulas used for the adjuvant treatment of UC were searched from the establishment of the databases to July 1,2022.Studies that met the inclusion criteria were screened and evaluated for literature quality and risk of bias according to the Cochrane 5.1 standard.The methodological quality of the studies was assessed using ReviewManager(RevMan)5.4,and a funnel plot was constructed to test for publication bias.ADDIS 1.16 statistical software was used to perform statistical analysis of the treatment measures and derive the network relationship and ranking diagrams of the various intervention measures.RESULTS A total of 64 randomized controlled trials involving 5456 patients with UC were included in this study.The adjuvant treatment of UC using five TCM formulations was able to improve the clinical outcome of the patients.Adjuvant treatment with Baitouweng decoction(BTWT)showed a significant effect[mean difference=36.22,95%confidence interval(CI):7.63 to 65.76].For the reduction of tumor necrosis factor in patients with UC,adjunctive therapy with BTWT(mean difference=−9.55,95%CI:−17.89 to−1.41),Shenlingbaizhu powder[SLBZS;odds ratio(OR)=0.19,95%CI:0.08 to 0.39],and Shaoyao decoction(OR=−23.02,95%CI:−33.64 to−13.14)was effective.Shaoyao decoction was more effective than BTWT(OR=0.12,95%CI:0.03 to 0.39),SLBZS(OR=0.19,95%CI:0.08 to 0.39),and Xi Lei powder(OR=0.34,95%CI:0.13 to 0.81)in reducing tumor necrosis factor and the recurrence rate of UC.CONCLUSION TCM combined with mesalazine is more effective than mesalazine alone in the treatment of UC.
文摘A large number of network security breaches in IoT networks have demonstrated the unreliability of current Network Intrusion Detection Systems(NIDSs).Consequently,network interruptions and loss of sensitive data have occurred,which led to an active research area for improving NIDS technologies.In an analysis of related works,it was observed that most researchers aim to obtain better classification results by using a set of untried combinations of Feature Reduction(FR)and Machine Learning(ML)techniques on NIDS datasets.However,these datasets are different in feature sets,attack types,and network design.Therefore,this paper aims to discover whether these techniques can be generalised across various datasets.Six ML models are utilised:a Deep Feed Forward(DFF),Convolutional Neural Network(CNN),Recurrent Neural Network(RNN),Decision Tree(DT),Logistic Regression(LR),and Naive Bayes(NB).The accuracy of three Feature Extraction(FE)algorithms is detected;Principal Component Analysis(PCA),Auto-encoder(AE),and Linear Discriminant Analysis(LDA),are evaluated using three benchmark datasets:UNSW-NB15,ToN-IoT and CSE-CIC-IDS2018.Although PCA and AE algorithms have been widely used,the determination of their optimal number of extracted dimensions has been overlooked.The results indicate that no clear FE method or ML model can achieve the best scores for all datasets.The optimal number of extracted dimensions has been identified for each dataset,and LDA degrades the performance of the ML models on two datasets.The variance is used to analyse the extracted dimensions of LDA and PCA.Finally,this paper concludes that the choice of datasets significantly alters the performance of the applied techniques.We believe that a universal(benchmark)feature set is needed to facilitate further advancement and progress of research in this field.
