With the rapid development of the Internet,network security and data privacy are increasingly valued.Although classical Network Intrusion Detection System(NIDS)based on Deep Learning(DL)models can provide good detecti...With the rapid development of the Internet,network security and data privacy are increasingly valued.Although classical Network Intrusion Detection System(NIDS)based on Deep Learning(DL)models can provide good detection accuracy,but collecting samples for centralized training brings the huge risk of data privacy leakage.Furthermore,the training of supervised deep learning models requires a large number of labeled samples,which is usually cumbersome.The“black-box”problem also makes the DL models of NIDS untrustworthy.In this paper,we propose a trusted Federated Learning(FL)Traffic IDS method called FL-TIDS to address the above-mentioned problems.In FL-TIDS,we design an unsupervised intrusion detection model based on autoencoders that alleviates the reliance on marked samples.At the same time,we use FL for model training to protect data privacy.In addition,we design an improved SHAP interpretable method based on chi-square test to perform interpretable analysis of the trained model.We conducted several experiments to evaluate the proposed FL-TIDS.We first determine experimentally the structure and the number of neurons of the unsupervised AE model.Secondly,we evaluated the proposed method using the UNSW-NB15 and CICIDS2017 datasets.The exper-imental results show that the unsupervised AE model has better performance than the other 7 intrusion detection models in terms of precision,recall and f1-score.Then,federated learning is used to train the intrusion detection model.The experimental results indicate that the model is more accurate than the local learning model.Finally,we use an improved SHAP explainability method based on Chi-square test to analyze the explainability.The analysis results show that the identification characteristics of the model are consistent with the attack characteristics,and the model is reliable.展开更多
In quantum noise stream cipher(QNSC)systems,it is difficult to compensate fiber nonlinearity by digital signal processing(DSP)due to interactions between chromatic dispersion(CD),amplified spontaneous emission(ASE)noi...In quantum noise stream cipher(QNSC)systems,it is difficult to compensate fiber nonlinearity by digital signal processing(DSP)due to interactions between chromatic dispersion(CD),amplified spontaneous emission(ASE)noise from erbiumdoped fiber amplifier(EDFA)and Kerr nonlinearity.Nonlinearity equalizer(NLE)based on machine learning(ML)algorithms have been extensively studied.However,most NLE based on supervised ML algorithms have high training overhead and computation complexity.In addition,the performance of these algorithms have a lot of randomness.This paper proposes two clustering algorithms based on Fuzzylogic C-Means Clustering(FLC)to compensate the fiber nonlinearity in quadrature amplitude modulation(QAM)-based QNSC system,including FLC based on subtractive clustering(SC)and annealing evolution(AE)algorithm.The performance of FLC-SC and FLC-AE are evaluated through simulation and experiment.The proposed algorithms can promptly obtain suitable initial centroids and choose optimal initial centroids of the clusters to achieve the global optimal initial centroids especially for high order modulation scheme.In the simulation,different parameter configurations are considered,including fiber length,optical signal-to-noise ratio(OSNR),clipping ratio and resolution of digital to analog converter(DAC).Further-more,we measure the Q-factor of transmission signal with different launched powers,DAC resolution and laser linewidth in the optical back-to-back(BTB)experiment with 80-km single mode fiber.Both simulation and experimental results show that the proposed techniques can greatly mitigate the signal impairments.展开更多
The task of modeling and analyzing intercepted multifunction radars(MFRs)pulse trains is vital for cognitive electronic reconnaissance.Existing methodologies predominantly rely on prior information or heavily constrai...The task of modeling and analyzing intercepted multifunction radars(MFRs)pulse trains is vital for cognitive electronic reconnaissance.Existing methodologies predominantly rely on prior information or heavily constrained models,posing challenges for non-cooperative applications.This paper introduces a novel approach to model MFRs using a Bayesian network,where the conditional probability density function is approximated by an autoregressive kernel mixture network(ARKMN).Utilizing the estimated probability density function,a dynamic programming algorithm is proposed for denoising and detecting change points in the intercepted MFRs pulse trains.Simulation results affirm the proposed method's efficacy in modeling MFRs,outperforming the state-of-the-art in pulse train denoising and change point detection.展开更多
Time-series anomaly detection plays a crucial role in the operation of offshore wind turbines.Various wind turbine monitoring systems rely on time-series data to monitor and identify anomalies in real-time,as well as ...Time-series anomaly detection plays a crucial role in the operation of offshore wind turbines.Various wind turbine monitoring systems rely on time-series data to monitor and identify anomalies in real-time,as well as to initiate early warning processes.However,for offshore wind turbines with a high data density,conventional methods have high computational overhead in detecting anomalies while failing to accurately detect anomalies due to variations in data scales.To address this challenge,we propose an efficient anomaly detection method with contrastive learning,called Hawkeye.Hawkeye is based on residual clustering,an unsupervised anomaly detection method for multivariate time-series data.To ensure accurate anomaly detection,a trend-capturing prediction module is also combined with an automatic labeling module.As a result,the most common information can be learned from multivariate time-series data to reconstruct data trends.By evaluating Hawkeye on public datasets and real-world offshore wind turbine operational datasets,the results show that Hawkeye’s F_(1)-score improves by an average of 14% compared with Isolation Forest,and its size shrinks by up to 11.5 times on the largest dataset compared with other methods.The proposed Hawkeye is potential to real-time monitoring and early warning systems for wind turbines,accelerating the development of intelligent operation and maintenance.展开更多
Locality preserving projection （LPP） is a typical and popular dimensionality reduction （DR） method,and it can potentially find discriminative projection directions by preserving the local geometric structure in da...Locality preserving projection （LPP） is a typical and popular dimensionality reduction （DR） method,and it can potentially find discriminative projection directions by preserving the local geometric structure in data. However,LPP is based on the neighborhood graph artificially constructed from the original data,and the performance of LPP relies on how well the nearest neighbor criterion work in the original space. To address this issue,a novel DR algorithm,called the self-dependent LPP （sdLPP） is proposed. And it is based on the fact that the nearest neighbor criterion usually achieves better performance in LPP transformed space than that in the original space. Firstly,LPP is performed based on the typical neighborhood graph; then,a new neighborhood graph is constructed in LPP transformed space and repeats LPP. Furthermore,a new criterion,called the improved Laplacian score,is developed as an empirical reference for the discriminative power and the iterative termination. Finally,the feasibility and the effectiveness of the method are verified by several publicly available UCI and face data sets with promising results.展开更多
A methodology is presented whereby a neural network is used to learn the inverse kinematic relationships of the position and orientation of a six joint manipulator. The arm solution for the orientation of a manipulato...A methodology is presented whereby a neural network is used to learn the inverse kinematic relationships of the position and orientation of a six joint manipulator. The arm solution for the orientation of a manipulator using a self organizing neural net is studied in this paper. A new training model of the self organizing neural network is proposed by thoroughly studying Martinetz, Ritter and Schulten′s self organizing neural network based on Kohonen′s self organizing mapping algorithm using a Widrow Hoff type error correction rule and closely combining the characters of the inverse kinematic relationship for a robot arm. The computer simulation results for a PUMA 560 robot show that the proposed method has a significant improvement over other methods documented in the references in self organizing capability and precision by training process.展开更多
Froth image features of coal flotation have been extracted and studied by neighboring grey level dependence matrix, spatial grey level dependence matrix and grey level histogram. In this paper, a basic algorithm of un...Froth image features of coal flotation have been extracted and studied by neighboring grey level dependence matrix, spatial grey level dependence matrix and grey level histogram. In this paper, a basic algorithm of unsupervised learning pattern classification is presented, and coal flotation froth images are classified by means of self organizing map (SOM). By extracting features from 51 flotation froth images with laboratory column, four types of froth images are classified. The correct rate of SOM cluster is satisfactory. And a good relationship of froth type with average ash content is also observed.展开更多
Purpose： Our study proposes a bootstrapping-based method to automatically extract data- usage statements from academic texts. Design/methodology/approach： The method for data-usage statements extraction starts with ...Purpose： Our study proposes a bootstrapping-based method to automatically extract data- usage statements from academic texts. Design/methodology/approach： The method for data-usage statements extraction starts with seed entities and iteratively learns patterns and data-usage statements from unlabeled text. In each iteration, new patterns are constructed and added to the pattern list based on their calculated score. Three seed-selection strategies are also proposed in this paper. Findings： The performance of the method is verified by means of experiments on real data collected from computer science journals. The results show that the method can achieve satisfactory performance regarding precision of extraction and extensibility of obtained patterns. Research limitations： While the triple representation of sentences is effective and efficient for extracting data-usage statements, it is unable to handle complex sentences. Additional features that can address complex sentences should thus be explored in the future. Practical implications： Data-usage statements extraction is beneficial for data-repository construction and facilitates research on data-usage tracking, dataset-based scholar search, and dataset evaluation. Originality/value： To the best of our knowledge, this paper is among the first to address the important task of automatically extracting data-usage statements from real data.展开更多
Mobile edge computing has emerged as a new paradigm to enhance computing capabilities by offloading complicated tasks to nearby cloud server.To conserve energy as well as maintain quality of service,low time complexit...Mobile edge computing has emerged as a new paradigm to enhance computing capabilities by offloading complicated tasks to nearby cloud server.To conserve energy as well as maintain quality of service,low time complexity algorithm is proposed to complete task offloading and server allocation.In this paper,a multi-user with multiple tasks and single server scenario is considered for small network,taking full account of factors including data size,bandwidth,channel state information.Furthermore,we consider a multi-server scenario for bigger network,where the influence of task priority is taken into consideration.To jointly minimize delay and energy cost,we propose a distributed unsupervised learning-based offloading framework for task offloading and server allocation.We exploit a memory pool to store input data and corresponding decisions as key-value pairs for model to learn to solve optimization problems.To further reduce time cost and achieve near-optimal performance,we use convolutional neural networks to process mass data based on fully connected networks.Numerical results show that the proposed algorithm performs better than other offloading schemes,which can generate near-optimal offloading decision timely.展开更多
Recently,the evolution of Generative Adversarial Networks(GANs)has embarked on a journey of revolutionizing the field of artificial and computational intelligence.To improve the generating ability of GANs,various loss...Recently,the evolution of Generative Adversarial Networks(GANs)has embarked on a journey of revolutionizing the field of artificial and computational intelligence.To improve the generating ability of GANs,various loss functions are introduced to measure the degree of similarity between the samples generated by the generator and the real data samples,and the effectiveness of the loss functions in improving the generating ability of GANs.In this paper,we present a detailed survey for the loss functions used in GANs,and provide a critical analysis on the pros and cons of these loss functions.First,the basic theory of GANs along with the training mechanism are introduced.Then,the most commonly used loss functions in GANs are introduced and analyzed.Third,the experimental analyses and comparison of these loss functions are presented in different GAN architectures.Finally,several suggestions on choosing suitable loss functions for image synthesis tasks are given.展开更多
Supervised learning methods（eg.PLS-DA,SVM,etc.） have been widely used with laser-induced breakdown spectroscopy（LIBS） to classify materials;however,it may induce a low correct classification rate if a test sample ...Supervised learning methods（eg.PLS-DA,SVM,etc.） have been widely used with laser-induced breakdown spectroscopy（LIBS） to classify materials;however,it may induce a low correct classification rate if a test sample type is not included in the training dataset.Unsupervised cluster analysis methods（hierarchical clustering analysis,K-means clustering analysis,and iterative self-organizing data analysis technique） are investigated in plastics classification based on the line intensities of LIBS emission in this paper.The results of hierarchical clustering analysis using four different similarity measuring methods（single linkage,complete linkage,unweighted pair-group average,and weighted pair-group average） are compared.In K-means clustering analysis,four kinds of choosing initial centers methods are applied in our case and their results are compared.The classification results of hierarchical clustering analysis,K-means clustering analysis,and ISODATA are analyzed.The experiment results demonstrated cluster analysis methods can be applied to plastics discrimination with LIBS.展开更多
基金supported by National Natural Science Fundation of China under Grant 61972208National Natural Science Fundation(General Program)of China under Grant 61972211+2 种基金National Key Research and Development Project of China under Grant 2020YFB1804700Future Network Innovation Research and Application Projects under Grant No.2021FNA020062021 Jiangsu Postgraduate Research Innovation Plan under Grant No.KYCX210794.
文摘With the rapid development of the Internet,network security and data privacy are increasingly valued.Although classical Network Intrusion Detection System(NIDS)based on Deep Learning(DL)models can provide good detection accuracy,but collecting samples for centralized training brings the huge risk of data privacy leakage.Furthermore,the training of supervised deep learning models requires a large number of labeled samples,which is usually cumbersome.The“black-box”problem also makes the DL models of NIDS untrustworthy.In this paper,we propose a trusted Federated Learning(FL)Traffic IDS method called FL-TIDS to address the above-mentioned problems.In FL-TIDS,we design an unsupervised intrusion detection model based on autoencoders that alleviates the reliance on marked samples.At the same time,we use FL for model training to protect data privacy.In addition,we design an improved SHAP interpretable method based on chi-square test to perform interpretable analysis of the trained model.We conducted several experiments to evaluate the proposed FL-TIDS.We first determine experimentally the structure and the number of neurons of the unsupervised AE model.Secondly,we evaluated the proposed method using the UNSW-NB15 and CICIDS2017 datasets.The exper-imental results show that the unsupervised AE model has better performance than the other 7 intrusion detection models in terms of precision,recall and f1-score.Then,federated learning is used to train the intrusion detection model.The experimental results indicate that the model is more accurate than the local learning model.Finally,we use an improved SHAP explainability method based on Chi-square test to analyze the explainability.The analysis results show that the identification characteristics of the model are consistent with the attack characteristics,and the model is reliable.
基金supported by NSFC Projects(Grant No.:61901053,61831003,62021005)the Fundamental Research Funds for the Central Universities(2021RC12).
文摘In quantum noise stream cipher(QNSC)systems,it is difficult to compensate fiber nonlinearity by digital signal processing(DSP)due to interactions between chromatic dispersion(CD),amplified spontaneous emission(ASE)noise from erbiumdoped fiber amplifier(EDFA)and Kerr nonlinearity.Nonlinearity equalizer(NLE)based on machine learning(ML)algorithms have been extensively studied.However,most NLE based on supervised ML algorithms have high training overhead and computation complexity.In addition,the performance of these algorithms have a lot of randomness.This paper proposes two clustering algorithms based on Fuzzylogic C-Means Clustering(FLC)to compensate the fiber nonlinearity in quadrature amplitude modulation(QAM)-based QNSC system,including FLC based on subtractive clustering(SC)and annealing evolution(AE)algorithm.The performance of FLC-SC and FLC-AE are evaluated through simulation and experiment.The proposed algorithms can promptly obtain suitable initial centroids and choose optimal initial centroids of the clusters to achieve the global optimal initial centroids especially for high order modulation scheme.In the simulation,different parameter configurations are considered,including fiber length,optical signal-to-noise ratio(OSNR),clipping ratio and resolution of digital to analog converter(DAC).Further-more,we measure the Q-factor of transmission signal with different launched powers,DAC resolution and laser linewidth in the optical back-to-back(BTB)experiment with 80-km single mode fiber.Both simulation and experimental results show that the proposed techniques can greatly mitigate the signal impairments.
基金supported by the National Natural Science Foundation of China under Grant 62301119。
文摘The task of modeling and analyzing intercepted multifunction radars(MFRs)pulse trains is vital for cognitive electronic reconnaissance.Existing methodologies predominantly rely on prior information or heavily constrained models,posing challenges for non-cooperative applications.This paper introduces a novel approach to model MFRs using a Bayesian network,where the conditional probability density function is approximated by an autoregressive kernel mixture network(ARKMN).Utilizing the estimated probability density function,a dynamic programming algorithm is proposed for denoising and detecting change points in the intercepted MFRs pulse trains.Simulation results affirm the proposed method's efficacy in modeling MFRs,outperforming the state-of-the-art in pulse train denoising and change point detection.
基金supported by Shanghai Electric Power Energy Technology Co.,Ltd.2023 Science and Technology Project under Grant No.33019006220801.
文摘Time-series anomaly detection plays a crucial role in the operation of offshore wind turbines.Various wind turbine monitoring systems rely on time-series data to monitor and identify anomalies in real-time,as well as to initiate early warning processes.However,for offshore wind turbines with a high data density,conventional methods have high computational overhead in detecting anomalies while failing to accurately detect anomalies due to variations in data scales.To address this challenge,we propose an efficient anomaly detection method with contrastive learning,called Hawkeye.Hawkeye is based on residual clustering,an unsupervised anomaly detection method for multivariate time-series data.To ensure accurate anomaly detection,a trend-capturing prediction module is also combined with an automatic labeling module.As a result,the most common information can be learned from multivariate time-series data to reconstruct data trends.By evaluating Hawkeye on public datasets and real-world offshore wind turbine operational datasets,the results show that Hawkeye’s F_(1)-score improves by an average of 14% compared with Isolation Forest,and its size shrinks by up to 11.5 times on the largest dataset compared with other methods.The proposed Hawkeye is potential to real-time monitoring and early warning systems for wind turbines,accelerating the development of intelligent operation and maintenance.
基金Supported by the National Natural Science Foundation of China (60973097)the Scientific Research Foundation of Liaocheng University(X0810029)~~
文摘Locality preserving projection （LPP） is a typical and popular dimensionality reduction （DR） method,and it can potentially find discriminative projection directions by preserving the local geometric structure in data. However,LPP is based on the neighborhood graph artificially constructed from the original data,and the performance of LPP relies on how well the nearest neighbor criterion work in the original space. To address this issue,a novel DR algorithm,called the self-dependent LPP （sdLPP） is proposed. And it is based on the fact that the nearest neighbor criterion usually achieves better performance in LPP transformed space than that in the original space. Firstly,LPP is performed based on the typical neighborhood graph; then,a new neighborhood graph is constructed in LPP transformed space and repeats LPP. Furthermore,a new criterion,called the improved Laplacian score,is developed as an empirical reference for the discriminative power and the iterative termination. Finally,the feasibility and the effectiveness of the method are verified by several publicly available UCI and face data sets with promising results.
文摘A methodology is presented whereby a neural network is used to learn the inverse kinematic relationships of the position and orientation of a six joint manipulator. The arm solution for the orientation of a manipulator using a self organizing neural net is studied in this paper. A new training model of the self organizing neural network is proposed by thoroughly studying Martinetz, Ritter and Schulten′s self organizing neural network based on Kohonen′s self organizing mapping algorithm using a Widrow Hoff type error correction rule and closely combining the characters of the inverse kinematic relationship for a robot arm. The computer simulation results for a PUMA 560 robot show that the proposed method has a significant improvement over other methods documented in the references in self organizing capability and precision by training process.
基金National Natural Science Foundation of China( 5 99740 32 )
文摘Froth image features of coal flotation have been extracted and studied by neighboring grey level dependence matrix, spatial grey level dependence matrix and grey level histogram. In this paper, a basic algorithm of unsupervised learning pattern classification is presented, and coal flotation froth images are classified by means of self organizing map (SOM). By extracting features from 51 flotation froth images with laboratory column, four types of froth images are classified. The correct rate of SOM cluster is satisfactory. And a good relationship of froth type with average ash content is also observed.
基金supported by the National Natural Science Foundation of China (Grant No.:71473183)
文摘Purpose： Our study proposes a bootstrapping-based method to automatically extract data- usage statements from academic texts. Design/methodology/approach： The method for data-usage statements extraction starts with seed entities and iteratively learns patterns and data-usage statements from unlabeled text. In each iteration, new patterns are constructed and added to the pattern list based on their calculated score. Three seed-selection strategies are also proposed in this paper. Findings： The performance of the method is verified by means of experiments on real data collected from computer science journals. The results show that the method can achieve satisfactory performance regarding precision of extraction and extensibility of obtained patterns. Research limitations： While the triple representation of sentences is effective and efficient for extracting data-usage statements, it is unable to handle complex sentences. Additional features that can address complex sentences should thus be explored in the future. Practical implications： Data-usage statements extraction is beneficial for data-repository construction and facilitates research on data-usage tracking, dataset-based scholar search, and dataset evaluation. Originality/value： To the best of our knowledge, this paper is among the first to address the important task of automatically extracting data-usage statements from real data.
基金presented in part at the EAI CHINACOM 2020supported in part by Natural Science Foundation of Jiangxi Province (Grant No.20202BAB212003)+1 种基金Projects of Humanities and Social Sciences of universities in Jiangxi (JC18224)Science and technology project of Jiangxi Provincial Department of Education(GJJ210817, GJJ210854)
文摘Mobile edge computing has emerged as a new paradigm to enhance computing capabilities by offloading complicated tasks to nearby cloud server.To conserve energy as well as maintain quality of service,low time complexity algorithm is proposed to complete task offloading and server allocation.In this paper,a multi-user with multiple tasks and single server scenario is considered for small network,taking full account of factors including data size,bandwidth,channel state information.Furthermore,we consider a multi-server scenario for bigger network,where the influence of task priority is taken into consideration.To jointly minimize delay and energy cost,we propose a distributed unsupervised learning-based offloading framework for task offloading and server allocation.We exploit a memory pool to store input data and corresponding decisions as key-value pairs for model to learn to solve optimization problems.To further reduce time cost and achieve near-optimal performance,we use convolutional neural networks to process mass data based on fully connected networks.Numerical results show that the proposed algorithm performs better than other offloading schemes,which can generate near-optimal offloading decision timely.
文摘Recently,the evolution of Generative Adversarial Networks(GANs)has embarked on a journey of revolutionizing the field of artificial and computational intelligence.To improve the generating ability of GANs,various loss functions are introduced to measure the degree of similarity between the samples generated by the generator and the real data samples,and the effectiveness of the loss functions in improving the generating ability of GANs.In this paper,we present a detailed survey for the loss functions used in GANs,and provide a critical analysis on the pros and cons of these loss functions.First,the basic theory of GANs along with the training mechanism are introduced.Then,the most commonly used loss functions in GANs are introduced and analyzed.Third,the experimental analyses and comparison of these loss functions are presented in different GAN architectures.Finally,several suggestions on choosing suitable loss functions for image synthesis tasks are given.
基金supported by Beijing Natural Science Foundation of China(No.4132063)
文摘Supervised learning methods（eg.PLS-DA,SVM,etc.） have been widely used with laser-induced breakdown spectroscopy（LIBS） to classify materials;however,it may induce a low correct classification rate if a test sample type is not included in the training dataset.Unsupervised cluster analysis methods（hierarchical clustering analysis,K-means clustering analysis,and iterative self-organizing data analysis technique） are investigated in plastics classification based on the line intensities of LIBS emission in this paper.The results of hierarchical clustering analysis using four different similarity measuring methods（single linkage,complete linkage,unweighted pair-group average,and weighted pair-group average） are compared.In K-means clustering analysis,four kinds of choosing initial centers methods are applied in our case and their results are compared.The classification results of hierarchical clustering analysis,K-means clustering analysis,and ISODATA are analyzed.The experiment results demonstrated cluster analysis methods can be applied to plastics discrimination with LIBS.
