Finding important edges in a graph is a crucial problem for various research fields, such as network epidemics, signal processing, machine learning, and sensor networks. In this paper, we tackle the problem based on sampling theory on graphs. We convert the original graph to a line graph where its nodes and edges, respectively, represent the original edges and the connections between the edges. We then perform node sampling of the line graph based on the edge smoothness assumption: this process selects the most critical edges in the original graph. We present a general framework of edge sampling based on graph sampling theory and reveal a theoretical relationship between the degree of the original graph and the line graph. We also propose an acceleration method for edge sampling in the proposed framework by using the relationship between the two types of Laplacian of the node and edge domains. Experimental results in synthetic and real-world graphs validate the effectiveness of our approach against some alternative edge selection methods.
{"title":"Edge Sampling of Graphs: Graph Signal Processing Approach With Edge Smoothness","authors":"Kenta Yanagiya;Koki Yamada;Yasuo Katsuhara;Tomoya Takatani;Yuichi Tanaka","doi":"10.1109/TSIPN.2025.3639969","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3639969","url":null,"abstract":"Finding important edges in a graph is a crucial problem for various research fields, such as network epidemics, signal processing, machine learning, and sensor networks. In this paper, we tackle the problem based on sampling theory on graphs. We convert the original graph to a <italic>line graph</i> where its nodes and edges, respectively, represent the original edges and the connections between the edges. We then perform node sampling of the line graph based on the <italic>edge smoothness</i> assumption: this process selects the most critical edges in the original graph. We present a general framework of edge sampling based on graph sampling theory and reveal a theoretical relationship between the degree of the original graph and the line graph. We also propose an acceleration method for edge sampling in the proposed framework by using the relationship between the two types of Laplacian of the node and edge domains. Experimental results in synthetic and real-world graphs validate the effectiveness of our approach against some alternative edge selection methods.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1592-1604"},"PeriodicalIF":3.0,"publicationDate":"2025-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11288079","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145778202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-08DOI: 10.1109/TSIPN.2025.3640933
Ali Bashirgonbadi;Mohamad Reza Salehi;Hamid Soltanian-Zadeh
This study presents a novel vertex-frequency framework to quantify structure–function coupling across brain networks and explore its relationship with cognitive ability. Using vertex-frequency energy distribution applied to Blood Oxygenation Level Dependent (BOLD) signal mapped onto the structural brain graph, we assessed coupling across 360 regions defined by the Glasser atlas. The analysis revealed region-specific coupling patterns, particularly within the Default Mode Network (DMN) and Dorsal Attention Network (DAN) that significantly correlate with general cognitive ability (g-factor). Validation on two independent resting-state functional Magnetic Resonance Imaging (rs-fMRI) sessions from the Human Connectome Project demonstrated high reproducibility (r ≈ 0.98), demonstrating the consistency of the proposed method. In contrast to conventional Graph Fourier Transform (GFT)-based approaches, which showed sensitivity to frequency cutoff parameters, our method yielded consistent coupling estimates without requiring parameter tuning. These findings suggest that vertex-frequency analysis is useful for capturing localized structure–function interactions and their cognitive relevance.
{"title":"Vertex-Frequency Analysis of Brain Networks: Unveiling the Connection Between Structure-Function Coupling and Cognitive Ability","authors":"Ali Bashirgonbadi;Mohamad Reza Salehi;Hamid Soltanian-Zadeh","doi":"10.1109/TSIPN.2025.3640933","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3640933","url":null,"abstract":"This study presents a novel vertex-frequency framework to quantify structure–function coupling across brain networks and explore its relationship with cognitive ability. Using vertex-frequency energy distribution applied to Blood Oxygenation Level Dependent (BOLD) signal mapped onto the structural brain graph, we assessed coupling across 360 regions defined by the Glasser atlas. The analysis revealed region-specific coupling patterns, particularly within the Default Mode Network (DMN) and Dorsal Attention Network (DAN) that significantly correlate with general cognitive ability (g-factor). Validation on two independent resting-state functional Magnetic Resonance Imaging (rs-fMRI) sessions from the Human Connectome Project demonstrated high reproducibility (r ≈ 0.98), demonstrating the consistency of the proposed method. In contrast to conventional Graph Fourier Transform (GFT)-based approaches, which showed sensitivity to frequency cutoff parameters, our method yielded consistent coupling estimates without requiring parameter tuning. These findings suggest that vertex-frequency analysis is useful for capturing localized structure–function interactions and their cognitive relevance.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1582-1591"},"PeriodicalIF":3.0,"publicationDate":"2025-12-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145778256","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 : 2025-12-04DOI: 10.1109/TSIPN.2025.3639916
Ce Chuai;Fuyong Wang;Mingwei Sun;Mikulas Huba;Pavol Bistak
This article explores the resilient fault-tolerant containment control problem for nonlinear multi-agent systems with actuator faults and denial-of-service (DoS) attacks. To circument unknown agent dynamics, the nonlinear data mapping of agents with time-varying actuator fault information is established by the locally dynamic linearization technique. In the cyber layer, the stochastic DoS attack is supposed to follow the Bernoulli distribution with duration and frequency constraints, and a backward attack compensation strategy is built. In the physical layer, an adaptive varying actuator fault compensation mechanism derived from the improved projection algorithm is developed. Within this design, a data-driven distributed model-free adaptive fault-tolerant control (DMFA-FTC) method is formulated to ensure the dual security guarantees. By the nature of irreducible sub-stochastic matrices, the convergence condition of the method is provided. Finally, experiments affirm the DMFA-FTC method.
{"title":"Data-Driven Containment Control for Nonlinear Multi-Agent Systems Under DoS Attacks and Actuator Faults","authors":"Ce Chuai;Fuyong Wang;Mingwei Sun;Mikulas Huba;Pavol Bistak","doi":"10.1109/TSIPN.2025.3639916","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3639916","url":null,"abstract":"This article explores the resilient fault-tolerant containment control problem for nonlinear multi-agent systems with actuator faults and denial-of-service (DoS) attacks. To circument unknown agent dynamics, the nonlinear data mapping of agents with time-varying actuator fault information is established by the locally dynamic linearization technique. In the cyber layer, the stochastic DoS attack is supposed to follow the Bernoulli distribution with duration and frequency constraints, and a backward attack compensation strategy is built. In the physical layer, an adaptive varying actuator fault compensation mechanism derived from the improved projection algorithm is developed. Within this design, a data-driven distributed model-free adaptive fault-tolerant control (DMFA-FTC) method is formulated to ensure the dual security guarantees. By the nature of irreducible sub-stochastic matrices, the convergence condition of the method is provided. Finally, experiments affirm the DMFA-FTC method.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1569-1581"},"PeriodicalIF":3.0,"publicationDate":"2025-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145729304","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 : 2025-11-25DOI: 10.1109/TSIPN.2025.3636753
Qinghuan Yang;Fuyong Wang;Tuo Zhou;Zhongxin Liu
This paper investigates the secure consensus of leader-following multi-agent systems under event-triggered control against hybrid cyber attacks. The system faces dual challenges: hybrid attacks threaten consensus security, and communication burden requires reduction— balancing these is the core issue. To resolve this, firstly, a hybrid cyber attack model with constraints on attack frequency and duration is constructed, more general than existing single-attack models. Then, an attack detection scheme based on watermarking and Kullback-Leibler divergence detector is designed, where transmitted data is encrypted/decrypted via watermarks to aid attack identification. Additionally, to reduce communication burden, distributed static and dynamic event-triggered control mechanisms are proposed and analyzed, respectively. Sufficient conditions for asymptotic leader-following consensus are established by exploring correlations between attack frequency, duration, and event-triggering parameters. The non-existence of Zeno behavior is theoretically demonstrated. Finally, simulation examples with wheeled mobile robots are performed to validate the results.
{"title":"Secure Event-Triggered Consensus Control of Multi-Agent Systems Under Hybrid Cyber Attacks","authors":"Qinghuan Yang;Fuyong Wang;Tuo Zhou;Zhongxin Liu","doi":"10.1109/TSIPN.2025.3636753","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3636753","url":null,"abstract":"This paper investigates the secure consensus of leader-following multi-agent systems under event-triggered control against hybrid cyber attacks. The system faces dual challenges: hybrid attacks threaten consensus security, and communication burden requires reduction— balancing these is the core issue. To resolve this, firstly, a hybrid cyber attack model with constraints on attack frequency and duration is constructed, more general than existing single-attack models. Then, an attack detection scheme based on watermarking and Kullback-Leibler divergence detector is designed, where transmitted data is encrypted/decrypted via watermarks to aid attack identification. Additionally, to reduce communication burden, distributed static and dynamic event-triggered control mechanisms are proposed and analyzed, respectively. Sufficient conditions for asymptotic leader-following consensus are established by exploring correlations between attack frequency, duration, and event-triggering parameters. The non-existence of Zeno behavior is theoretically demonstrated. Finally, simulation examples with wheeled mobile robots are performed to validate the results.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1500-1514"},"PeriodicalIF":3.0,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674714","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 : 2025-11-25DOI: 10.1109/TSIPN.2025.3636755
Renyi Wang;Songsong Cheng;Yuan Fan;Jianbin Qiu
Distributed weakly convex optimization is a significant class of problems in signal and information processing, with wide-ranging applications such as sparse dictionary learning, low-rank matrix completion, and robust phase retrieval. Most existing distributed algorithms for solving this type of problem are designed based on exact gradient information. However, it is challenging to obtain this information as closed-form analytical expressions are often unavailable in certain circumstances. In this article, we propose a gradient estimation scheme for distributed weakly convex optimization problems, estimating the gradient information using finite differences in orthogonal random directions. This approach is more general and has better estimation effectiveness than existing methods based on stochastic vectors. Furthermore, we design a projected zeroth-order gradient tracking algorithm, which effectively solves the considered problem over an unbalanced communication topology. We also demonstrate that the proposed algorithm converges to a stationary point with a rate of ${mathcal {O}}(ln K/{sqrt{K}})$ from the perspective of the Moreau envelope. Finally, we provide two examples to verify the effectiveness of our algorithm.
{"title":"Distributed Zeroth-Order Gradient Tracking for Weakly Convex Optimization Over Unbalanced Graphs","authors":"Renyi Wang;Songsong Cheng;Yuan Fan;Jianbin Qiu","doi":"10.1109/TSIPN.2025.3636755","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3636755","url":null,"abstract":"Distributed weakly convex optimization is a significant class of problems in signal and information processing, with wide-ranging applications such as sparse dictionary learning, low-rank matrix completion, and robust phase retrieval. Most existing distributed algorithms for solving this type of problem are designed based on exact gradient information. However, it is challenging to obtain this information as closed-form analytical expressions are often unavailable in certain circumstances. In this article, we propose a gradient estimation scheme for distributed weakly convex optimization problems, estimating the gradient information using finite differences in orthogonal random directions. This approach is more general and has better estimation effectiveness than existing methods based on stochastic vectors. Furthermore, we design a projected zeroth-order gradient tracking algorithm, which effectively solves the considered problem over an unbalanced communication topology. We also demonstrate that the proposed algorithm converges to a stationary point with a rate of <inline-formula><tex-math>${mathcal {O}}(ln K/{sqrt{K}})$</tex-math></inline-formula> from the perspective of the Moreau envelope. Finally, we provide two examples to verify the effectiveness of our algorithm.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1515-1526"},"PeriodicalIF":3.0,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674799","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 : 2025-11-25DOI: 10.1109/TSIPN.2025.3636760
Jiong Yu;Lei Xue;Jian Liu;Yongbao Wu;Changyin Sun
This paper proposes an aperiodically intermittent dynamic event-triggered control (AIDE-TC) to investigate the practical predefined-time synchronization (PTS) for complex networks (CNs) with time-varying coupling. Notably, the proposed AIDE-TC strategy is based on the average control rate, which makes the theoretical conditions more applicable to practical scenarios. In contrast to fixed-time synchronization, we design a time-varying function to ensure that all states of CNs converge to a common adjustable region within a user-defined time. Moreover, unlike the existing literature that considers bounded time-varying coupling weights, this paper employs average time-varying coupling weights to relax the requirements for the coupling weights. More importantly, a new auxiliary function is designed to overcome the difficulties caused by intermittent control in practical PTS. By constructing a Lyapunov function with the auxiliary function, we derive the practical PTS criterion for CNs with time-varying coupling. Meanwhile, the Zeno phenomenon is excluded when the coupling weights have an upper bound. Finally, we present a numerical simulation applied to an islanded microgrid system to verify the effectiveness of the obtained results.
{"title":"Practical Predefined-Time Synchronization for Complex Networks With Time-Varying Coupling via Intermittent Dynamic Event-Triggered Control","authors":"Jiong Yu;Lei Xue;Jian Liu;Yongbao Wu;Changyin Sun","doi":"10.1109/TSIPN.2025.3636760","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3636760","url":null,"abstract":"This paper proposes an aperiodically intermittent dynamic event-triggered control (AIDE-TC) to investigate the practical predefined-time synchronization (PTS) for complex networks (CNs) with time-varying coupling. Notably, the proposed AIDE-TC strategy is based on the average control rate, which makes the theoretical conditions more applicable to practical scenarios. In contrast to fixed-time synchronization, we design a time-varying function to ensure that all states of CNs converge to a common adjustable region within a user-defined time. Moreover, unlike the existing literature that considers bounded time-varying coupling weights, this paper employs average time-varying coupling weights to relax the requirements for the coupling weights. More importantly, a new auxiliary function is designed to overcome the difficulties caused by intermittent control in practical PTS. By constructing a Lyapunov function with the auxiliary function, we derive the practical PTS criterion for CNs with time-varying coupling. Meanwhile, the Zeno phenomenon is excluded when the coupling weights have an upper bound. Finally, we present a numerical simulation applied to an islanded microgrid system to verify the effectiveness of the obtained results.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1543-1556"},"PeriodicalIF":3.0,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674802","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}
The environment undergoes perpetual changes that are influenced by a combination of endogenous and exogenous factors. Consequently, it exerts a substantial influence on an individual's physical and psychological state, directly or indirectly affecting the evolutionary dynamics of a population described by a network, which in turn can also alter the environment. Furthermore, the evolution of strategies, shaped by reputation, can diverge due to variations in multiple factors. To explore the potential consequences of the mentioned situations, this paper studies how game and reputation dynamics alter the evolution of cooperation. Concretely, game transitions are determined by individuals' behaviors and external uncontrollable factors. The cooperation level of its neighbors reflects individuals' reputation, and further, a general fitness function regarding payoff and reputation is provided. Within the context of the donation game, we investigate the relevant outcomes associated with the aforementioned evolutionary process, considering various topologies for distinct interactions. Additionally, a biased mutation is introduced to gain a deeper insight into the strategy evolution. We detect a substantial increase in the cooperation level through intensive simulations, and some important phenomena are observed, e.g., the unilateral increase of the value of prosocial behavior limits promotion in cooperative behavior in square-lattice networks.
{"title":"Evolutionary Dynamics Based on Reputation in Networked Populations With Game Transitions","authors":"Yuji Zhang;Minyu Feng;Jürgen Kurths;Attila Szolnoki","doi":"10.1109/TSIPN.2025.3636748","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3636748","url":null,"abstract":"The environment undergoes perpetual changes that are influenced by a combination of endogenous and exogenous factors. Consequently, it exerts a substantial influence on an individual's physical and psychological state, directly or indirectly affecting the evolutionary dynamics of a population described by a network, which in turn can also alter the environment. Furthermore, the evolution of strategies, shaped by reputation, can diverge due to variations in multiple factors. To explore the potential consequences of the mentioned situations, this paper studies how game and reputation dynamics alter the evolution of cooperation. Concretely, game transitions are determined by individuals' behaviors and external uncontrollable factors. The cooperation level of its neighbors reflects individuals' reputation, and further, a general fitness function regarding payoff and reputation is provided. Within the context of the donation game, we investigate the relevant outcomes associated with the aforementioned evolutionary process, considering various topologies for distinct interactions. Additionally, a biased mutation is introduced to gain a deeper insight into the strategy evolution. We detect a substantial increase in the cooperation level through intensive simulations, and some important phenomena are observed, e.g., the unilateral increase of the value of prosocial behavior limits promotion in cooperative behavior in square-lattice networks.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1557-1568"},"PeriodicalIF":3.0,"publicationDate":"2025-11-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145729322","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 : 2025-11-24DOI: 10.1109/TSIPN.2025.3636747
Jun Hu;Bingxin Lei;Raquel Caballero-Águila;Hui Yu
This paper presents a distributed state estimation (SE) approach utilizing an adaptive event-triggered mechanism (AETM), designed for time-varying nonlinear complex networks subject to uncertain coupling strengths (UCSs) and data integrity attacks. The AETM is introduced to regulate the frequency of data transmission by adaptively adjusting the triggering thresholds. In addition, data integrity attacks with two forms of attack are considered, namely a multiplicative attack and a linear attack. In particular, these two types of attacks are launched with a certain order of priority. The objective of this paper is to design a distributed SE approach under the AETM, considering UCSs and data integrity attacks, where a locally optimized upper bound (UB) on the estimation error covariance (EEC) is obtained using the method of mathematical induction and minimized by properly selecting the estimator gain. Furthermore, a sufficient condition in relation to the uniform boundedness of the UB on the EEC is deduced. Besides, the monotonicity of the trace of the minimized UB of EEC with respect to the attack probability and strength is also analyzed and shown. Finally, simulation experiments with comparative results are presented to illustrate the validity of the proposed distributed SE method in localizing multiple mobile robots.
{"title":"Adaptive Event-Triggered State Estimation for Nonlinear Complex Networks With Uncertain Coupling Strengths and Data Integrity Attacks","authors":"Jun Hu;Bingxin Lei;Raquel Caballero-Águila;Hui Yu","doi":"10.1109/TSIPN.2025.3636747","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3636747","url":null,"abstract":"This paper presents a distributed state estimation (SE) approach utilizing an adaptive event-triggered mechanism (AETM), designed for time-varying nonlinear complex networks subject to uncertain coupling strengths (UCSs) and data integrity attacks. The AETM is introduced to regulate the frequency of data transmission by adaptively adjusting the triggering thresholds. In addition, data integrity attacks with two forms of attack are considered, namely a multiplicative attack and a linear attack. In particular, these two types of attacks are launched with a certain order of priority. The objective of this paper is to design a distributed SE approach under the AETM, considering UCSs and data integrity attacks, where a locally optimized upper bound (UB) on the estimation error covariance (EEC) is obtained using the method of mathematical induction and minimized by properly selecting the estimator gain. Furthermore, a sufficient condition in relation to the uniform boundedness of the UB on the EEC is deduced. Besides, the monotonicity of the trace of the minimized UB of EEC with respect to the attack probability and strength is also analyzed and shown. Finally, simulation experiments with comparative results are presented to illustrate the validity of the proposed distributed SE method in localizing multiple mobile robots.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1527-1542"},"PeriodicalIF":3.0,"publicationDate":"2025-11-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145674766","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}
Kernel-based reconstruction of graph signals was extensively studied in graph signal processing domain, which has been which has been verified to be efficient for real-world applications. However, a fast sampling approach under the kernel-based assumption is still missing in this field, that can further improve the reconstruction performance. In this paper, we will propose an eigen-decomposition-free (ED-free) sampling method for general kernel-based graph signals. We first define graph signals as vectors in reproducing kernel Hilbert space (RKHS), and then introduce an explicit kernel-based recovery method. Then, we focus on the A-optimal sampling function on submatrix of a constantly shifted kernel matrix to stabilize the recovery performance. For fast optimization, we define node-dependent vectors called localized graph filter impulse responses (LogFIRs)—the impulse responses of square root of kernel matrix, which can be estimated fast without ED. Using these vectors, we reformulate our sampling objective and develop a low-complexity algorithm to evaluate nodes via only simple scalar multiplications. Further, we find that the proposed algorithm tends to sample nodes whose supported LogFIRs are as orthogonal as possible. Besides, we prove that our proposed sampling method can improve the recovery performance of graph signals under bandlimited and smooth models via exquisitely choosing specific kernels. Finally, extensive simulations on synthetic and real-world graph signals have validated that our proposed method achieves state-of-the-art execution time and has superior reconstruction performance compared to other deterministic sampling alternatives.
{"title":"Fast Kernel-Based Sampling of Graph Signals Using Localized Graph Filter Impulse Responses","authors":"Fen Wang;Cheng Yang;Wuyue Zhang;Xue Zhang;Baoyi Xu","doi":"10.1109/TSIPN.2025.3626207","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3626207","url":null,"abstract":"Kernel-based reconstruction of graph signals was extensively studied in graph signal processing domain, which has been which has been verified to be efficient for real-world applications. However, a fast sampling approach under the kernel-based assumption is still missing in this field, that can further improve the reconstruction performance. In this paper, we will propose an eigen-decomposition-free (ED-free) sampling method for general kernel-based graph signals. We first define graph signals as vectors in reproducing kernel Hilbert space (RKHS), and then introduce an explicit kernel-based recovery method. Then, we focus on the A-optimal sampling function on submatrix of a constantly shifted kernel matrix to stabilize the recovery performance. For fast optimization, we define node-dependent vectors called <italic>localized graph filter impulse responses</i> (LogFIRs)—the impulse responses of square root of kernel matrix, which can be estimated fast without ED. Using these vectors, we reformulate our sampling objective and develop a low-complexity algorithm to evaluate nodes via only simple scalar multiplications. Further, we find that the proposed algorithm tends to sample nodes whose supported LogFIRs are as orthogonal as possible. Besides, we prove that our proposed sampling method can improve the recovery performance of graph signals under bandlimited and smooth models via exquisitely choosing specific kernels. Finally, extensive simulations on synthetic and real-world graph signals have validated that our proposed method achieves state-of-the-art execution time and has superior reconstruction performance compared to other deterministic sampling alternatives.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1473-1489"},"PeriodicalIF":3.0,"publicationDate":"2025-11-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145612066","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 : 2025-11-06DOI: 10.1109/TSIPN.2025.3629606
Haodong Chen;Jianquan Lu;Lulu Li;Jungang Lou
State estimation plays a crucial role in Boolean networks as it determines the capability of predicting and regulating network dynamic behaviors. Particularly, state-dependent packet dropouts (SDPDs) are of critical importance, as it enables more accurate modeling of communication constraints in practical systems, thereby significantly affecting the reliability of state estimation. This paper addresses the state estimation problem in stochastic Boolean networks (SBNs) with SDPDs through a novel interactive coupled estimation framework. Using the semi-tensor product technique, we first derive the algebraic form of SDPD-affected SBN, where SDPDs are characterized by a transition probability matrix linking region labels to states. Our framework decomposes the estimation process into two interactively coordinated components: region label estimation and state estimation. We develop a matrix-based recursive method to compute the forward state-label probability distribution, enabling optimal region label sequence estimation. Based on the estimated region labels and the observation sequence, we then derive a state estimator to reconstruct the internal states of SDPD-affected SBNs. The effectiveness of this dual estimation structure is validated through a biological case study.
{"title":"State Estimation of Stochastic Boolean Networks Facing State-Dependent Packet Dropouts: An Interactive Coupled Estimation Method","authors":"Haodong Chen;Jianquan Lu;Lulu Li;Jungang Lou","doi":"10.1109/TSIPN.2025.3629606","DOIUrl":"https://doi.org/10.1109/TSIPN.2025.3629606","url":null,"abstract":"State estimation plays a crucial role in Boolean networks as it determines the capability of predicting and regulating network dynamic behaviors. Particularly, state-dependent packet dropouts (SDPDs) are of critical importance, as it enables more accurate modeling of communication constraints in practical systems, thereby significantly affecting the reliability of state estimation. This paper addresses the state estimation problem in stochastic Boolean networks (SBNs) with SDPDs through a novel interactive coupled estimation framework. Using the semi-tensor product technique, we first derive the algebraic form of SDPD-affected SBN, where SDPDs are characterized by a transition probability matrix linking region labels to states. Our framework decomposes the estimation process into two interactively coordinated components: region label estimation and state estimation. We develop a matrix-based recursive method to compute the forward state-label probability distribution, enabling optimal region label sequence estimation. Based on the estimated region labels and the observation sequence, we then derive a state estimator to reconstruct the internal states of SDPD-affected SBNs. The effectiveness of this dual estimation structure is validated through a biological case study.","PeriodicalId":56268,"journal":{"name":"IEEE Transactions on Signal and Information Processing over Networks","volume":"11 ","pages":"1490-1499"},"PeriodicalIF":3.0,"publicationDate":"2025-11-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145612044","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：