Information Sciences最新文献

{"title":"Energy-related controllability of corona product networks","authors":"Qiang Zhang ,&nbsp;Junjie Huang ,&nbsp;Bo Liu ,&nbsp;Housheng Su ,&nbsp;Alatancang Chen","doi":"10.1016/j.ins.2025.122654","DOIUrl":"10.1016/j.ins.2025.122654","url":null,"abstract":"<div><div>This article studies the energy-related controllability for a category of ‘large’ composite networks generated by ‘small’ simple factor networks with Laplacian dynamics under a leader-follower framework via corona product. Different from most existing literature on network controllability, this work characterizes the controllability of corona product networks (CPNs) from an energy point of view. This can quantify the difficulty of controlling CPNs based on controllability Gramian measures, involving average controllability and volumetric control energy, etc., where the energy is triggered by the leaders. The energy-related controllability of a CPN can be explored from the eigenvalues and eigenvectors of its factor networks. An algorithm for solving the maximum average controllability is provided, which can help one select the leaders to optimize network control and be applied in practice.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"723 ","pages":"Article 122654"},"PeriodicalIF":6.8,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145050624","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Bi-objective optimization for electric vehicle scheduling under vehicle-to-grid integration","authors":"Bing Yu,&nbsp;Yong Liu,&nbsp;Liang Ma","doi":"10.1016/j.ins.2025.122649","DOIUrl":"10.1016/j.ins.2025.122649","url":null,"abstract":"<div><div>Vehicle-to-grid (V2G) technology leverages the distributed energy-storage potential of electric vehicles (EVs), transforming the challenges of large-scale EV integration into opportunities to enhance grid flexibility and reliability. This study investigates the optimization of EV charging-discharging schedules by exploiting V2G capabilities. First, considering the spatiotemporal distribution of EVs, a Markov chain is constructed to describe probabilistic transitions between spatiotemporal states, which is then embedded in a traffic-network-based path decision model. Second, a dynamic battery energy consumption model is established, incorporating multiple factors that influence battery performance. Using Monte Carlo simulation results, a bi-objective optimization model is formulated to schedule charging and discharging, simultaneously minimizing (i) total cost — including user recharging time and battery degradation — and (ii) grid-load fluctuation. Given the NP-hard nature of the problem, an improved multi-objective bitterling fish optimization (IMOBFO) algorithm is developed to balance global exploration and local exploitation. Empirical studies in a region of Shanghai compare three strategies: disordered charging, ordered charging, and the proposed optimized charging–discharging strategy. Experimental results confirm the feasibility of the proposed model and the effectiveness of IMOBFO. Comparative analysis with seven other algorithms further validates the superior performance and stability of IMOBFO according to multiple multi-objective evaluation metrics.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"723 ","pages":"Article 122649"},"PeriodicalIF":6.8,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145027832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dynamic event-triggered finite-time actor-critic-identifier-based approximate optimal control for unknown nonlinear drifted systems","authors":"Shuangsi Xue ,&nbsp;Junkai Tan ,&nbsp;Zihang Guo ,&nbsp;Qingshu Guan ,&nbsp;Hui Cao ,&nbsp;Badong Chen","doi":"10.1016/j.ins.2025.122651","DOIUrl":"10.1016/j.ins.2025.122651","url":null,"abstract":"<div><div>This paper presents a finite-time dynamic event-triggered actor-critic-identifier (FT-DET-ACI) framework for the optimal control problem of nonlinear systems with uncertain drift dynamics. A theoretical foundation is established by reformulating the value function within a finite-time stable space, which facilitates system state stabilization within predetermined temporal constraints. The proposed approach derives finite-time optimal controllers through a transformed Hamilton-Jacobi-Bellman (HJB) equation. To address unknown system dynamics, an integrated actor-critic-identifier architecture is constructed to concurrently approximate the value function, synthesize the finite-time optimal controller, and identify system parameters. A dynamic event-triggering rule is designed to reduce computational and communication loads by selectively updating the control signal. Lyapunov stability analysis is provided to demonstrate the finite-time convergence properties of the closed-loop system. Numerical experiments are conducted to validate the efficacy of the proposed FT-DET-ACI methodology.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"723 ","pages":"Article 122651"},"PeriodicalIF":6.8,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145021105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantization-aware matrix factorization for low bit rate image compression","authors":"Pooya Ashtari ,&nbsp;Pourya Behmandpoor ,&nbsp;Fateme Nateghi Haredasht ,&nbsp;Jonathan H. Chen ,&nbsp;Panagiotis Patrinos ,&nbsp;Sabine Van Huffel","doi":"10.1016/j.ins.2025.122646","DOIUrl":"10.1016/j.ins.2025.122646","url":null,"abstract":"<div><div>Lossy image compression is essential for efficient transmission and storage. Traditional compression methods mainly rely on discrete cosine transform (DCT) or singular value decomposition (SVD), both of which represent image data in continuous domains and, therefore, necessitate carefully designed quantizers. Notably, these methods consider quantization as a separate step, which prevents quantization errors from being incorporated into the compression process and degrades the reconstruction quality, particularly in SVD-based methods. To address this issue, we introduce a quantization-aware matrix factorization (QMF) to develop a novel lossy image compression method. QMF provides a low-rank representation of the image data as a product of two smaller matrices, with elements constrained to bounded integer values, thereby effectively integrating quantization with low-rank approximation. We propose an efficient, provably convergent iterative algorithm for QMF using a block coordinate descent scheme, with subproblems having closed-form solutions. Our experiments demonstrate that our method consistently outperforms JPEG at low bit rates below 0.25 bits per pixel. We also demonstrated that our method has an improved capability to preserve visual semantics compared to JPEG at low bit rates by evaluating an ImageNet pre-trained classifier on compressed images. The project is available at <span><span>https://github.com/pashtari/qmf</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"722 ","pages":"Article 122646"},"PeriodicalIF":6.8,"publicationDate":"2025-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145010826","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Attention-based spatial-temporal interactive couple neural networks for multivariate time series forecasting","authors":"Bingsheng Wei,&nbsp;Yonghua Hei,&nbsp;Yuan Wan","doi":"10.1016/j.ins.2025.122647","DOIUrl":"10.1016/j.ins.2025.122647","url":null,"abstract":"<div><div>Multivariate time series forecasting (MTSF) has been a significant research focus across various domains. Recent studies have utilized deep neural networks to identify pattern relationships in MTSF. Despite these developments, accurately forecasting multivariate time series remains challenging due to the trend of time series and spatial-temporal heterogeneity. In this paper, we propose a unified multivariate time series forecasting framework for long-term, short-term, and spatial-temporal forecasting with attention-based spatial-temporal interactive coupled neural networks (ASTIC). Specifically, we proposed a spatial-temporal interactive couple block that contains both temporal and spatial branches to investigate the relationships between global and local patterns in temporal and spatial perspectives. In the temporal branch, we design a hybrid network module capable of enhancing representation learning using convolution and attention mechanisms, which dynamically capture the local trendiness and long-term time dependence implicit in time series. In the spatial branch, a novel dynamic graph learners are designed to learn global and local spatial patterns. Then a novel interactive coupling method is proposed to link the two branches together. ASTIC predicts time series effectively by using a multilevel structure to model the trendiness of the series and mining the spatial-temporal heterogeneity. Experimental results show that our method outperforms state-of-the-art baseline methods on nine real-world datasets.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"723 ","pages":"Article 122647"},"PeriodicalIF":6.8,"publicationDate":"2025-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145020830","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Set membership filter with nonlinear state inequality constraints","authors":"Xiaowei Li,&nbsp;Xuqi Zhang,&nbsp;Zhiguo Wang,&nbsp;Xiaojing Shen","doi":"10.1016/j.ins.2025.122650","DOIUrl":"10.1016/j.ins.2025.122650","url":null,"abstract":"<div><div>Set membership filter is a promising method to provide a bounding estimation containing the true state for dynamic systems with unknown but bounded noises. In this paper, we investigate the state bounding estimation problem of nonlinear dynamic systems with nonlinear state inequality constraints. Three types of ellipsoidal state bounding estimation methods are proposed by incorporating nonlinear state inequality constraints into nonlinear set membership filter. They are called model reduction method, system measurement method, and constraint dimension reduction method, respectively. We analyze the computation complexity of the three methods, which decrease in the order of model reduction method, system measurement method, and constraint dimension reduction method. Due to the nonlinearity of the dynamic systems, all the three methods are approximation algorithms and the state estimation accuracy cannot be analyzed explicitly. Consequently, a typical illustrative numerical experiment is conducted to compare the performance of the three methods. The results show that the accuracy increases in the order of the model reduction method, the constraint dimension reduction method, and the system measurement method.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"723 ","pages":"Article 122650"},"PeriodicalIF":6.8,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145021106","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Minimum initial state estimation of labeled time Petri nets in the presence of unobservable transitions","authors":"Liang Li ,&nbsp;Chen Wang ,&nbsp;Huimin Zhang ,&nbsp;Ding Liu","doi":"10.1016/j.ins.2025.122618","DOIUrl":"10.1016/j.ins.2025.122618","url":null,"abstract":"<div><div>This paper investigates the estimation of minimum initial states (MISs) for partially observable labeled time Petri net (LTPN) systems. Particularly, an MIS is defined by two components: a minimal initial marking possessing the minimum achievable total token count, and a set of timing constraints that explicitly define the static firing delays for every transition enabled under this initial marking. The number of logic transition sequences (LTSs) consistent with a given time-label sequence (TLS) can be infinite. To ensure tractability, we adopt the assumption that only a finite number of unobservable transition firings can occur prior to the firing of any observable transition. Within this framework, the initial step involves deriving the set of minimal initial markings. This derivation is based on LTSs that exhibit logical consistency with the provided TLS. Subsequently, we define a route-modified state class graph (R-MSCG) that is generated by firing the logic consistent LTSs attached to minimal initial markings. By utilizing the transitions-related timing constraints in R-MSCGs, a method is presented to detect whether these logic consistent LTSs are timing consistent with the given TLS. Moreover, we report an algorithm to implement the MISs estimation. Finally, a part processing unit is provided to illustrate how to apply the proposed algorithms to estimate the MIS that allows the specified task sequence to be completed with the least required resources while meeting the timing constraints on the task executions.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"722 ","pages":"Article 122618"},"PeriodicalIF":6.8,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144989441","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mean square event-triggered formation control for multi-agent systems under stochastic switching topologies","authors":"Guoying Miao ,&nbsp;Yiming Zhao ,&nbsp;Tao Li ,&nbsp;Jinde Cao ,&nbsp;Hanen Karamti","doi":"10.1016/j.ins.2025.122643","DOIUrl":"10.1016/j.ins.2025.122643","url":null,"abstract":"<div><div>This paper studies the challenging issue of privacy-preserving formation control for second-order multi-agent systems under stochastic switching topologies, mainly protecting privacy of true initial formation errors. Considering the circumstances of communication noises among agents, distributed intermittent event-triggered privacy-preserving formation algorithms including the fuzzy logic system are given. Meanwhile, a novel privacy-preserving mask function is proposed, where indistinguishable space is increased by 26.2% than ones in existing literatures. Based on stochastic stability analysis and graph theory, sufficient conditions of mean square formation with the bound error are derived. Moreover, optimal controllers for the leader are proposed within the intermittent event-triggered privacy-preserving framework. In order to deal with nonlinear and unknown function, we exploit the idea of adaptive control to parameterize partial derivative of corresponding cost function, which is another different technique rather than the adaptive dynamic programming one. Finally, two simulation examples are given to indicate feasibility of our proposed theoretical results.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"722 ","pages":"Article 122643"},"PeriodicalIF":6.8,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145010825","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Durrmeyer deep neural networks: Bridging deep learning and dynamic brain functional connectivity","authors":"Ugur Kadak","doi":"10.1016/j.ins.2025.122623","DOIUrl":"10.1016/j.ins.2025.122623","url":null,"abstract":"<div><div>This paper introduces a new family of Durrmeyer Deep Neural Networks (DDNN), characterized by its deep multi-layer architecture and the integration of Gaussian-Based Density (GBD) functions as sigmoidal activation mechanisms. Unlike shallow neural network (NN) operators, which often struggle with higher-order dependencies, this deep-layered approach significantly enhances approximation accuracy and adaptability. We rigorously analyze the convergence properties of DDNN in <span><math><msub><mrow><mi>L</mi></mrow><mrow><mi>p</mi></mrow></msub></math></span> spaces, demonstrating its superior stability over Bernstein-type operators. Importantly, extensive numerical experiments consistently show that DDNN achieves significantly smaller approximation errors compared to shallow-type NN operators, confirming its superior performance in function approximation. The deep structure of DDNN enables the modeling of complex, higher-order dependencies, while the GBD-based activations provide flexibility and adaptability in capturing transient neural interactions. This design enables the operator to effectively balance signal preservation and noise reduction through integral-based smoothing techniques. By bridging classical approximation theory and deep learning, this study establishes DDNN as a powerful tool for dynamic signal processing.</div><div>Moreover, we apply the DDNN framework to dynamic functional connectivity (DFC) analysis, where it effectively uncovers time-varying connectivity patterns in key brain regions by circumventing the limitations of conventional moving window techniques. The results highlight DDNN's ability to preserve low-frequency components while substantially reducing high-frequency noise, providing a more accurate representation of evolving brain states. Further, in its robust application to fMRI data for Independent Component Analysis (ICA), DDNN consistently and markedly enhanced spatial sparsity: raw component sparsity values (ranging from approximately 0.16 to 0.24) are significantly improved to a much higher range (0.77 to 0.85), representing an average increase of over 0.60 across components. This substantial improvement in spatial localization, coupled with DDNN's capacity for selective temporal modulation, provides a more accurate representation of evolving brain states, advancing the methodological landscape of functional neuroimaging and signal processing.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"722 ","pages":"Article 122623"},"PeriodicalIF":6.8,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144997699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Three-way partitioning graphs with reinforcement learning for adaptive knowledge tracing","authors":"Kai Zhang,&nbsp;Lihao Bi,&nbsp;Jinrong Xiong,&nbsp;Qi Wu,&nbsp;Xinyu Zhu","doi":"10.1016/j.ins.2025.122644","DOIUrl":"10.1016/j.ins.2025.122644","url":null,"abstract":"<div><div>Knowledge tracing (KT) aims to model a learner's mastery level and predict future performance based on historical learning interactions. While recent advances have leveraged neural and graph-based techniques, most existing models fail to account for the heterogeneous and dynamic influences among concepts. To address these limitations, we propose a novel framework called <strong>T</strong>hree-way <strong>P</strong>artitioning <strong>R</strong>einforced <strong>K</strong>nowledge <strong>T</strong>racing (TPR-KT), which integrates three functional modules to enhance adaptivity, interpretability, and predictive accuracy. First, a Region-Aware Partitioning module, grounded in three-way decision theory (TWD), partitions each concept's neighborhood into positive, boundary, and negative regions to capture varying levels of influence. Second, a Region-Specific Processing module employs specialized neural architectures to differentially process each region based on its learning effects. Third, a Region-Adaptive Updating module, guided by reinforcement learning, dynamically adjusts region thresholds to align with learners' evolving cognitive states. Extensive experiments on real-world and synthetic datasets demonstrate that our model significantly outperforms state-of-the-art baselines, validating the effectiveness of region-level modeling and adaptive partitioning in knowledge tracing.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"721 ","pages":"Article 122644"},"PeriodicalIF":6.8,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144988510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}