Pub Date : 2023-12-12DOI: 10.1109/TSIPN.2023.3341410
Jun Hu;Ruonan Luo;Hongli Dong;Cai Chen;Hongjian Liu
This paper investigates the dynamic event-triggered fusion filtering problem for a class of uncertain multi-sensor rectangular descriptor systems with random state delay. The random state delay is depicted by a Bernoulli distributed random variable. In order to save the communication energy, a dynamic event-triggered mechanism (DETM) is employed to decide whether the measurements are transmitted to the local estimators. Firstly, by introducing the full-order transformation method, the rectangular descriptor systems are converted into the non-descriptor systems with full orders. Secondly, the local filter gains are designed to minimize the upper bounds of filtering error covariances (FECs), where the upper bounds of FECs and the filter gains depend on a group of free positive scalar parameters. To minimize the upper bounds of FECs, the scalar parameters are sought optimally by a numerical method, where the scalars obtained after optimization are called optimal parameters. Subsequently, the fusion filter of the original descriptor system is given by the inverse covariance intersection (ICI) fusion technique. Finally, the effectiveness and advantages of the proposed fusion filtering algorithm are illustrated by providing the experiments with circuit system application.
{"title":"Dynamic Event-Triggered Fusion Filtering for Multi-Sensor Rectangular Descriptor Systems With Random State Delay","authors":"Jun Hu;Ruonan Luo;Hongli Dong;Cai Chen;Hongjian Liu","doi":"10.1109/TSIPN.2023.3341410","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3341410","url":null,"abstract":"This paper investigates the dynamic event-triggered fusion filtering problem for a class of uncertain multi-sensor rectangular descriptor systems with random state delay. The random state delay is depicted by a Bernoulli distributed random variable. In order to save the communication energy, a dynamic event-triggered mechanism (DETM) is employed to decide whether the measurements are transmitted to the local estimators. Firstly, by introducing the full-order transformation method, the rectangular descriptor systems are converted into the non-descriptor systems with full orders. Secondly, the local filter gains are designed to minimize the upper bounds of filtering error covariances (FECs), where the upper bounds of FECs and the filter gains depend on a group of free positive scalar parameters. To minimize the upper bounds of FECs, the scalar parameters are sought optimally by a numerical method, where the scalars obtained after optimization are called optimal parameters. Subsequently, the fusion filter of the original descriptor system is given by the inverse covariance intersection (ICI) fusion technique. Finally, the effectiveness and advantages of the proposed fusion filtering algorithm are illustrated by providing the experiments with circuit system application.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"836-849"},"PeriodicalIF":3.2,"publicationDate":"2023-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139034118","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper we study the stability properties of aggregation graph neural networks (Agg-GNNs) considering perturbations of the underlying graph. An Agg-GNN is a hybrid architecture where information is defined on the nodes of a graph, but it is processed block-wise by Euclidean CNNs on the nodes after several diffusions on the graph shift operator. We derive stability bounds for the mapping operator associated to a generic Agg-GNN, and we specify conditions under which such operators can be stable to deformations. We prove that the stability bounds are defined by the properties of the filters in the first layer of the CNN that acts on each node. Additionally, we show that there is a close relationship between the number of aggregations, the filter's selectivity, and the size of the stability constants. We also conclude that in Agg-GNNs the selectivity of the mapping operators can be limited by the stability restrictions imposed on the first layer of the CNN stage, but this is compensated by the pointwise nonlinearities and filters in subsequent layers which are not subject to any restriction. This shows a substantial difference with respect to the stability properties of selection GNNs, where the selectivity of the filters in all layers is constrained by their stability. We provide numerical evidence corroborating the results derived, testing the behavior of Agg-GNNs in real life application scenarios considering perturbations of different magnitude.
{"title":"Stability of Aggregation Graph Neural Networks","authors":"Alejandro Parada-Mayorga;Zhiyang Wang;Fernando Gama;Alejandro Ribeiro","doi":"10.1109/TSIPN.2023.3341408","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3341408","url":null,"abstract":"In this paper we study the stability properties of aggregation graph neural networks (Agg-GNNs) considering perturbations of the underlying graph. An Agg-GNN is a hybrid architecture where information is defined on the nodes of a graph, but it is processed block-wise by Euclidean CNNs on the nodes after several diffusions on the graph shift operator. We derive stability bounds for the mapping operator associated to a generic Agg-GNN, and we specify conditions under which such operators can be stable to deformations. We prove that the stability bounds are defined by the properties of the filters in the first layer of the CNN that acts on each node. Additionally, we show that there is a close relationship between the number of aggregations, the filter's selectivity, and the size of the stability constants. We also conclude that in Agg-GNNs the selectivity of the mapping operators can be limited by the stability restrictions imposed on the first layer of the CNN stage, but this is compensated by the pointwise nonlinearities and filters in subsequent layers which are not subject to any restriction. This shows a substantial difference with respect to the stability properties of selection GNNs, where the selectivity of the filters in all layers is constrained by their stability. We provide numerical evidence corroborating the results derived, testing the behavior of Agg-GNNs in real life application scenarios considering perturbations of different magnitude.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"850-864"},"PeriodicalIF":3.2,"publicationDate":"2023-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139034299","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-01DOI: 10.1109/TSIPN.2023.3338467
Jiayi Lei;Yuan-Xin Li;Choon Ki Ahn
The topic of this study is adaptive distributed finite-time (FT) optimization of uncertain nonlinear high-order multi-agent systems (MASs) with disturbances. The proposed two-stage framework consists of an optimal FT estimator and an adaptive FT tracking controller. First, the estimator drives the optimization variables towards the optimal solution. In contrast to existing optimization control studies, high-order MASs subject to unknown dynamics are studied in this case. Second, by using the output of the estimator as a reference signal, the tracking controller allows all agents to approach the optimal point. The use of a command filter avoids the problem of discontinuous gradient functions, while it is possible to handle unknown nonlinear functions using fuzzy logic systems (FLSs). We prove, based on the FT stability criterion and convex optimization theory, that the proposed strategy minimizes the total objective function and results in a closed-loop system with bounded signals and FT convergence to the optimal solution. Finally, through a simulation example, the developed approach is verified.
本研究的主题是具有扰动的不确定非线性高阶多代理系统（MAS）的自适应分布式有限时间（FT）优化。所提出的两阶段框架包括最优有限时间估计器和自适应有限时间跟踪控制器。首先，估计器将优化变量推向最优解。与现有的优化控制研究不同，本案例研究的是受未知动态影响的高阶 MAS。其次，通过使用估计器的输出作为参考信号，跟踪控制器允许所有代理接近最优点。指令滤波器的使用避免了不连续梯度函数的问题，同时可以使用模糊逻辑系统（FLS）来处理未知的非线性函数。根据 FT 稳定性准则和凸优化理论，我们证明了所提出的策略能使总目标函数最小化，并能产生一个信号有界的闭环系统，且 FT 收敛到最优解。最后，通过一个仿真实例验证了所开发的方法。
{"title":"Optimizing Multi-Agent Systems With Uncertain Dynamics: A Finite-Time Adaptive Distributed Approach","authors":"Jiayi Lei;Yuan-Xin Li;Choon Ki Ahn","doi":"10.1109/TSIPN.2023.3338467","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3338467","url":null,"abstract":"The topic of this study is adaptive distributed finite-time (FT) optimization of uncertain nonlinear high-order multi-agent systems (MASs) with disturbances. The proposed two-stage framework consists of an optimal FT estimator and an adaptive FT tracking controller. First, the estimator drives the optimization variables towards the optimal solution. In contrast to existing optimization control studies, high-order MASs subject to unknown dynamics are studied in this case. Second, by using the output of the estimator as a reference signal, the tracking controller allows all agents to approach the optimal point. The use of a command filter avoids the problem of discontinuous gradient functions, while it is possible to handle unknown nonlinear functions using fuzzy logic systems (FLSs). We prove, based on the FT stability criterion and convex optimization theory, that the proposed strategy minimizes the total objective function and results in a closed-loop system with bounded signals and FT convergence to the optimal solution. Finally, through a simulation example, the developed approach is verified.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"865-874"},"PeriodicalIF":3.2,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139034117","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-12-01DOI: 10.1109/TSIPN.2023.3338452
Jian Liu;Yan Yang;Yongbao Wu;Seaar Al-Dabooni;Lei Xue;Donald C. Wunsch
In this article, the exponential synchronization (ES) of the reaction-diffusion systems on networks is studied under an asynchronous aperiodic intermittent control strategy. Different from the preceding studies, the control strategy of each node is different, which is more general and challenging. Meanwhile, to address the asynchrony problem of the asynchronous intermittent control, a new asynchronous average control rate (ACR) is constructed, which is different for each node. The ACR is greater than the lower bound of the control rate in the existing literature, which makes the results of this article less conservative. Then, by constructing the Lyapunov function and adopting the graph theory, some ES criteria are given for the reaction-diffusion systems on networks. Finally, the effectiveness of the algorithms is verified by the numerical simulations.
{"title":"Exponential Synchronization of Reaction-Diffusion Systems on Networks via Asynchronous Intermittent Control","authors":"Jian Liu;Yan Yang;Yongbao Wu;Seaar Al-Dabooni;Lei Xue;Donald C. Wunsch","doi":"10.1109/TSIPN.2023.3338452","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3338452","url":null,"abstract":"In this article, the exponential synchronization (ES) of the reaction-diffusion systems on networks is studied under an asynchronous aperiodic intermittent control strategy. Different from the preceding studies, the control strategy of each node is different, which is more general and challenging. Meanwhile, to address the asynchrony problem of the asynchronous intermittent control, a new asynchronous average control rate (ACR) is constructed, which is different for each node. The ACR is greater than the lower bound of the control rate in the existing literature, which makes the results of this article less conservative. Then, by constructing the Lyapunov function and adopting the graph theory, some ES criteria are given for the reaction-diffusion systems on networks. Finally, the effectiveness of the algorithms is verified by the numerical simulations.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"825-835"},"PeriodicalIF":3.2,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138678676","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-28DOI: 10.1109/TSIPN.2023.3335615
Mohammad-Hassan Ahmad Yarandi;Massoud Babaie-Zadeh
Using smoothness criteria to separate smooth graph signals from their summation is an approach that has recently been proposed (Mohammadi et al., 2023) and shown to have a unique solution up to the uncertainty of the average values of source signals. In this correspondence, closed-form solutions of both exact and approximate decompositions of that approach are presented. This closed-form solution in the exact decomposition also answers the open problem of the estimation error. Additionally, in the case of Gaussian source signals in the presence of additive Gaussian noise, it is shown that the optimization problem of that approach is equivalent to the Maximum A Posteriori (MAP) estimation of the sources.
{"title":"A Closed-Form Solution for Graph Signal Separation Based on Smoothness","authors":"Mohammad-Hassan Ahmad Yarandi;Massoud Babaie-Zadeh","doi":"10.1109/TSIPN.2023.3335615","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3335615","url":null,"abstract":"Using smoothness criteria to separate smooth graph signals from their summation is an approach that has recently been proposed (Mohammadi et al., 2023) and shown to have a unique solution up to the uncertainty of the average values of source signals. In this correspondence, closed-form solutions of both exact and approximate decompositions of that approach are presented. This closed-form solution in the exact decomposition also answers the open problem of the estimation error. Additionally, in the case of Gaussian source signals in the presence of additive Gaussian noise, it is shown that the optimization problem of that approach is equivalent to the Maximum A Posteriori (MAP) estimation of the sources.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"823-824"},"PeriodicalIF":3.2,"publicationDate":"2023-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138550231","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-23DOI: 10.1109/TSIPN.2023.3334496
Jun Hu;Shuting Fan;Cai Chen;Hongjian Liu;Xiaojian Yi
The paper investigates the distributed fusion filtering problem for time-varying multi-rate nonlinear systems (TVMRNSs) with sensor resolutions based on the encoding-decoding scheme (EDS) over sensor networks, where the iterative method is applied to the transformation of TVMRNSs. In order to enhance signal interference-resistant capability and improve transmission efficiency, the EDS based on dynamic quantization is introduced during the measurement transmission. On the basis of the decoded measurements, a local distributed filter is constructed, where an upper bound on the local filtering error (LFE) covariance is derived and the local filter gains are obtained by minimizing the trace of the upper bound. Subsequently, the fusion filtering algorithm is presented according to the covariance intersection fusion criterion. In addition, a sufficient condition is provided via reasonable assumptions to ensure the uniform boundedness of the upper bound on the LFE covariance. Finally, a moving target tracking practical example is taken to show the superiority of the proposed filtering algorithm and discuss the monotonicity of the mean-square error of the fusion filter with respect to the sensor resolutions and quantization intervals.
{"title":"Encoding-Decoding-Based Distributed Fusion Filtering for Multi-Rate Nonlinear Systems With Sensor Resolutions","authors":"Jun Hu;Shuting Fan;Cai Chen;Hongjian Liu;Xiaojian Yi","doi":"10.1109/TSIPN.2023.3334496","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3334496","url":null,"abstract":"The paper investigates the distributed fusion filtering problem for time-varying multi-rate nonlinear systems (TVMRNSs) with sensor resolutions based on the encoding-decoding scheme (EDS) over sensor networks, where the iterative method is applied to the transformation of TVMRNSs. In order to enhance signal interference-resistant capability and improve transmission efficiency, the EDS based on dynamic quantization is introduced during the measurement transmission. On the basis of the decoded measurements, a local distributed filter is constructed, where an upper bound on the local filtering error (LFE) covariance is derived and the local filter gains are obtained by minimizing the trace of the upper bound. Subsequently, the fusion filtering algorithm is presented according to the covariance intersection fusion criterion. In addition, a sufficient condition is provided via reasonable assumptions to ensure the uniform boundedness of the upper bound on the LFE covariance. Finally, a moving target tracking practical example is taken to show the superiority of the proposed filtering algorithm and discuss the monotonicity of the mean-square error of the fusion filter with respect to the sensor resolutions and quantization intervals.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"811-822"},"PeriodicalIF":3.2,"publicationDate":"2023-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138472866","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-11-06DOI: 10.1109/TSIPN.2023.3329301
Yun Chen;Yuhang Jin;Jianjun Bai;Mengze Zhu
This paper is concerned with the partial-sensor-measurements-based (PSMB) distributed estimation problem for a class of stochastic systems (SSs) with randomly occurring nonlinearities, persistent bounded noises and quantization effects. The observations of partial sensor nodes are available to be transmitted to the estimators. In order to enhance the utilization efficiency of limited resources, the Round-Robin protocol is deployed to schedule the data transmissions over communication networks. The sufficient condition is established to guarantee the mean-square exponential ultimate boundedness of the augmented estimation error system (AEES), and then the desired PSMB estimator gains are determined by minimizing the mean-square upper bound of the augmented estimation error vector subject to iterative matrix inequalities. Finally, an illustrative example demonstrates the effectiveness of proposed estimation scheme.
{"title":"Distributed Estimation by Partial Sensor Measurements Through Transmission Scheduling for Stochastic Systems","authors":"Yun Chen;Yuhang Jin;Jianjun Bai;Mengze Zhu","doi":"10.1109/TSIPN.2023.3329301","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3329301","url":null,"abstract":"This paper is concerned with the partial-sensor-measurements-based (PSMB) distributed estimation problem for a class of stochastic systems (SSs) with randomly occurring nonlinearities, persistent bounded noises and quantization effects. The observations of partial sensor nodes are available to be transmitted to the estimators. In order to enhance the utilization efficiency of limited resources, the Round-Robin protocol is deployed to schedule the data transmissions over communication networks. The sufficient condition is established to guarantee the mean-square exponential ultimate boundedness of the augmented estimation error system (AEES), and then the desired PSMB estimator gains are determined by minimizing the mean-square upper bound of the augmented estimation error vector subject to iterative matrix inequalities. Finally, an illustrative example demonstrates the effectiveness of proposed estimation scheme.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"800-810"},"PeriodicalIF":3.2,"publicationDate":"2023-11-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134794992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-10-30DOI: 10.1109/TSIPN.2023.3328275
Matin Macktoobian;Zhan Shu;Qing Zhao
Faults occurring in ad-hoc robot networks may fatally perturb their topologies leading to disconnection of subsets of those networks. Optimal topology synthesis is generally resource-intensive and time-consuming to be done in real time for large ad-hoc robot networks. One should only perform topology re-computations if the probability of topology recoverability after the occurrence of any fault surpasses that of its irrecoverability. We formulate this problem as a binary classification problem. Then, we develop a two-pathway data-driven model based on Bayesian Gaussian mixture models that predicts the solution to a typical problem by two different pre-fault and post-fault prediction pathways. The results, obtained by the integration of the predictions of those pathways, clearly indicate the success of our model in solving the topology (ir)recoverability prediction problem compared to the best of current strategies found in the literature.
{"title":"Topology Recoverability Prediction for Ad-Hoc Robot Networks: A Data-Driven Fault-Tolerant Approach","authors":"Matin Macktoobian;Zhan Shu;Qing Zhao","doi":"10.1109/TSIPN.2023.3328275","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3328275","url":null,"abstract":"Faults occurring in ad-hoc robot networks may fatally perturb their topologies leading to disconnection of subsets of those networks. Optimal topology synthesis is generally resource-intensive and time-consuming to be done in real time for large ad-hoc robot networks. One should only perform topology re-computations if the probability of topology recoverability after the occurrence of any fault surpasses that of its irrecoverability. We formulate this problem as a binary classification problem. Then, we develop a two-pathway data-driven model based on Bayesian Gaussian mixture models that predicts the solution to a typical problem by two different pre-fault and post-fault prediction pathways. The results, obtained by the integration of the predictions of those pathways, clearly indicate the success of our model in solving the topology (ir)recoverability prediction problem compared to the best of current strategies found in the literature.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"786-799"},"PeriodicalIF":3.2,"publicationDate":"2023-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"109157461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This article focuses on the practical fixed-time consensus tracking problem of second-order multi-agent systems (MASs) with mismatched disturbances and matched disturbances under directed graph. Actually, the leader can be a virtual signal or an actual agent. Considering these two situations, followers can track the leader in a finite time with uniform bound. By using the adding-a-power-integrator method, two fixed-time control protocols are proposed, thus the practical fixed-time tracking consensus is achieved in finite time, and settling time is independent of initial states. Finally, several simulations are given to further illustrate the effectiveness of the theory.
{"title":"Practical Fixed-Time Consensus Tracking for Second-Order Multi-Agent Systems With Mismatched Disturbances","authors":"Jiayi Gong;Fuyong Wang;Zhongxin Liu;Zengqiang Chen","doi":"10.1109/TSIPN.2023.3328276","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3328276","url":null,"abstract":"This article focuses on the practical fixed-time consensus tracking problem of second-order multi-agent systems (MASs) with mismatched disturbances and matched disturbances under directed graph. Actually, the leader can be a virtual signal or an actual agent. Considering these two situations, followers can track the leader in a finite time with uniform bound. By using the adding-a-power-integrator method, two fixed-time control protocols are proposed, thus the practical fixed-time tracking consensus is achieved in finite time, and settling time is independent of initial states. Finally, several simulations are given to further illustrate the effectiveness of the theory.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"761-769"},"PeriodicalIF":3.2,"publicationDate":"2023-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"109157457","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper presents a personalized graph federated learning (PGFL) framework in which distributedly connected servers and their respective edge devices collaboratively learn device or cluster-specific models while maintaining the privacy of every individual device. The proposed approach exploits similarities among different models to provide a more relevant experience for each device, even in situations with diverse data distributions and disproportionate datasets. Furthermore, to ensure a secure and efficient approach to collaborative personalized learning, we study a variant of the PGFL implementation that utilizes differential privacy, specifically zero-concentrated differential privacy, where a noise sequence perturbs model exchanges. Our mathematical analysis shows that the proposed privacy-preserving PGFL algorithm converges to the optimal cluster-specific solution for each cluster in linear time. It also reveals that exploiting similarities among clusters could lead to an alternative output whose distance to the original solution is bounded and that this bound can be adjusted by modifying the algorithm's hyperparameters. Further, our analysis shows that the algorithm ensures local differential privacy for all clients in terms of zero-concentrated differential privacy. Finally, the effectiveness of the proposed PGFL algorithm is showcased through numerical experiments conducted in the context of regression and classification tasks using some of the National Institute of Standards and Technology's (NIST's) datasets, namely, MNIST, and MedMNIST.
{"title":"Personalized Graph Federated Learning With Differential Privacy","authors":"Francois Gauthier;Vinay Chakravarthi Gogineni;Stefan Werner;Yih-Fang Huang;Anthony Kuh","doi":"10.1109/TSIPN.2023.3325963","DOIUrl":"https://doi.org/10.1109/TSIPN.2023.3325963","url":null,"abstract":"This paper presents a personalized graph federated learning (PGFL) framework in which distributedly connected servers and their respective edge devices collaboratively learn device or cluster-specific models while maintaining the privacy of every individual device. The proposed approach exploits similarities among different models to provide a more relevant experience for each device, even in situations with diverse data distributions and disproportionate datasets. Furthermore, to ensure a secure and efficient approach to collaborative personalized learning, we study a variant of the PGFL implementation that utilizes differential privacy, specifically zero-concentrated differential privacy, where a noise sequence perturbs model exchanges. Our mathematical analysis shows that the proposed privacy-preserving PGFL algorithm converges to the optimal cluster-specific solution for each cluster in linear time. It also reveals that exploiting similarities among clusters could lead to an alternative output whose distance to the original solution is bounded and that this bound can be adjusted by modifying the algorithm's hyperparameters. Further, our analysis shows that the algorithm ensures local differential privacy for all clients in terms of zero-concentrated differential privacy. Finally, the effectiveness of the proposed PGFL algorithm is showcased through numerical experiments conducted in the context of regression and classification tasks using some of the National Institute of Standards and Technology's (NIST's) datasets, namely, MNIST, and MedMNIST.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"9 ","pages":"736-749"},"PeriodicalIF":3.2,"publicationDate":"2023-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"109157459","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：