Multi-Modality Spatio-Temporal Forecasting (MoSTF) extends traditional spatio-temporal forecasting by incorporating diverse traffic modalities. Despite significant recent strides in spatio-temporal modeling, existing approaches often fail to explicitly model the coupling relationships between different modality variables. Accurate MoSTF is challenging, as it requires modeling (1) temporal dynamic heterogeneity under exogenous influences and (2) heterogeneous spatial dependencies alongside complex cross-variable couplings. To address these challenges, we propose the Dual-Domain Spectral Filtering Network (DSFNet). Our framework employs dual-domain spectral filtering to capture heterogeneous spatial patterns and explicitly model the relationships between variables. Unlike graph-based message passing or dense attention over node-modality pairs, DSFNet factorizes space-modality interactions into feature-domain and spatial-domain spectral operators, enabling scalable modeling of nonlocal dependencies and cross-modality couplings. Furthermore, we introduce an external gating mechanism to adaptively regulate temporal dynamics under external influences. We validate our method through extensive experiments on five representative real-world traffic datasets. Compared with the second-best baselines, DSFNet reduces MAE by 3.21

This paper addresses the observer-based control problem for two-time-scale Markov networked control systems operating under a coding-decoding scheme. To overcome communication bandwidth limitations and safeguard data privacy, the discrepancy between the observer state data and the designed duplicator decoded data is encoded into a specific sequence of codewords before transmission. The decoder converts these codewords into decoded forms upon receipt. A theoretical framework for the coding-decoding scheme is established using a uniform quantization method to derive criteria ensuring system detectability. Based on this framework, a decoded state-dependent controller is designed, and sufficient conditions for the stability of the estimated error system are provided. By constructing appropriate Lyapunov functions and utilizing the linear matrix inequality method, a sufficient condition for the exponential mean-square boundedness of the closed-loop system is derived. Numerical simulation demonstrates the effectiveness of the proposed control scheme.

In previous studies on the bifurcation analysis of neural networks, the precise relationship between dynamic bifurcation and system parameters has often remained unclear. This paper investigates the dynamic bifurcation and control of delayed fractional-order bidirectional associative memory inertial neural networks (FOBAMINNs). First, the stability of the FOBAMINNs without delay is analyzed by examining the corresponding characteristic equations. Then, its dynamics with delay, which is taken as the bifurcation parameter, is studied. Sufficient conditions for delay-induced Hopf bifurcation are established. Consequently, an explicit analytical formula is derived to elucidate the relationship between the bifurcation point and system parameters. To effectively control the dynamic bifurcation, a feedback controller is designed. Theoretical analysis demonstrates that the emergence of bifurcation points can be effectively delayed by appropriately selecting the feedback control gain. Simulation results confirm the validity of the main results.

Multimodal image registration aims to align images within a common geometric space and is essential for tasks like image fusion and segmentation. However, existing methods often focus only on rigid or non-rigid transformations, which limits their effectiveness when both are present. We propose a coarse-to-fine image registration framework, termed SA-CFNet. It performs rigid image registration in the coarse stage and non-rigid image registration in the fine stage. It consists of two key components: the Rigid Transformation Network (RNet) and Semantic-Guided Deformation Network (SGDNet). RNet estimates a 6D transformation matrix that applies rotation and translation to the moving image, yielding a coarsely registered result. To establish reliable correspondences, we use a semantic similarity-based strategy that leverages semantic information from the images. SGDNet refines the registration by estimating a deformation field, allowing the coarse_registration_image to be further adjusted for fine registration. The deformation field is also calculated using semantic similarity. We employ a coarse-to-fine estimation strategy to enhance the accuracy of the deformation field. We conducted experiments on image rigid transformation and non-rigid transformation using datasets such as RoadScene, TNO, and M3FD. To validate the effectiveness for downstream tasks, we designed an experiment on multimodal image fusion. The results show that our method remains robust in handling different rotations and translations for rigid image registration and achieves state-of-the-art performance in image registration tasks.

In this paper, we propose a class of asymmetric networks with two different coupling styles to highlight the impact of network parameters and leader position on network coherence. Firstly, the exact analytical expressions for the leaderless coherence of two networks are derived. Secondly, we select several distinct schemes for assigning leaders to the networks and derive the corresponding expressions between leader-follower coherence and network parameters. Thirdly, the coherence and robustness properties of the networks are further analyzed based on the previously derived theoretical results. Finally, the Kirchhoff index, global mean first-passage time, and Laplacian energy of the networks are analyzed. Based on the analysis results of this paper, we can conclude that leader position and network parameters have a profound impact on coherence. Specifically, the coupled networks with no chain connections in the middle exhibit better coherence performance. The analytical results obtained above provide a new perspective and theoretical foundation for the optimal design of coordination mechanisms in distributed systems.

Stochastic disturbance intensifies the influence of uncertainty on closed-loop systems, and conventional control methods for stochastic nonlinear systems usually fail to ensure estimation accuracy of uncertain terms, degrading control performance. This paper studies composite adaptive fuzzy self-triggered (ST) control for uncertain stochastic nonlinear strict feedback systems with partially immeasurable states. To estimate unknown states, a fuzzy state observer is implemented. A series-parallel estimation model is constructed to get a prediction error used to design a composite fuzzy adaptive law to improve fuzzy approximation accuracy. The issue of ``explosion of complexity'' in the backstepping controller design is addressed by employing a command filter, and an appropriate error compensation system is proposed to reduce the effect of filter errors. A novel ST strategy is designed to minimize consumption of communication resources. By utilizing stochastic theory, a composite adaptive fuzzy ST controller is designed, which ensures tracking error converges to a small region of the origin and all signals in the closed-loop system are uniformly ultimately bounded in mean square. Moreover, the proposed approach exhibits satisfactory tracking performance and efficient communication capability. Finally, the effectiveness and applicability of the proposed method are validated via two numerical simulations.

This paper mainly studies global mu-consensus in the mean square problem of the nonlinear multi-agent systems (MASs) with unbounded time-varying delays and stochastic delayed impulses. Different from previous works, stochastic delayed impulsive effects are considered in general nonlinear delayed MASs to simulate the sudden changes that the systems encounter during operation. In addition, a novel event-triggered mechanism is designed, where it is only necessary to monitor the state of the MASs at impulsive instants. In this way, Zeno phenomenon can be effectively avoided. Moreover, the frequency of information exchange among agents and controller updates can be reduced, leading to a reduction in control costs. Then, based on the designed event-triggered control strategy and some techniques of stochastic analysis, the global consensus in the mean square criteria for the nonlinear delayed MASs under stochastic delayed impulses are proposed. Finally, examples are presented to illustrate the validity of the results obtained.

The accurate association mining between micro ribonucleic acids (micro RNA) and diseases could help understand the complex mechanism of diseases. However, the computational algorithms fail to sufficiently extract discriminative patterns from complex biological networks. In this work, we devised a cross-view graph representation learning model based on graph convolution and hierarchical attention with a variational information bottleneck strategy for the potential association prediction of micro RNA-disease pairs. The proposed model not only captures different view-specific similarities, but also discards the noise attribute and superfluous information of nodes. Specifically, our model first built a heterogeneous biological graph by the micro RNA and disease similarity and known associations. Then, the special graph convolutional module is devised to extract the multi-view local interaction features, and the hierarchical attention layer combines the multi-view representation of nodes to capture higher-order discriminative representations. Moreover, the variational information bottleneck retains relevant information while minimizing the redundant features of micro RNA and diseases to obtain discriminative representations. Finally, the adaptive loss function is designed to optimize the performance. Extensive experimental results demonstrate the superior performance of our model in comparison to state-of-the-art models, and indicate that it is a reliable representation learning and prediction model for the micro RNA-disease associations.

This paper explores the cooperative output regulation control of fractional-order multi-agent systems (FOMASs) under external Denial-of-Service (DoS) attacks and internal sensor faults. A distributed resilient observer is constructed to estimate the states of the exosystem. A switching adaptive law is designed to estimate the unmeasurable states of the FOMAS caused by sensor faults, in combination with the output of the sensors. Furthermore, filters and compensation signals are designed to avoid complexity explosion and the need for high-order derivatives of the observer. Based on the aforementioned designs and a backstepping framework, a distributed fault-tolerant control framework is developed, and the stability of the system is proven based on the Lyapunov-stability criterion. The developed control framework can uniformly handle both external DoS attacks and internal sensor faults. Finally, two simulation examples are given to verify the effectiveness of the proposed scheme.

Traditional consensus control for multiagent systems (MASs) often suffers from excessive communication burden due to continuous control input updates. This article presents a sampled-data event-triggered (SDET) adaptive fuzzy bipartite consensus control scheme for fractional-order MASs. A fuzzy state observer is constructed to estimate unmeasurable states and unknown nonlinearities. An SDET condition integrates event errors, fuzzy weight estimates, and bipartite consensus errors. Distinct from existing input-variation-based triggering mechanisms, the proposed SDET mechanism updates and transmits the control and adaptive laws only at aperiodic sampled-data instants, significantly reducing computational and communication load. To address the nonsmooth and piecewise constant virtual control signals induced by the SDET mechanism, a fractional-order command filter is incorporated into the backstepping design, effectively eliminating the explosion of complexity and facilitating rigorous stability analysis. Within a hybrid-system framework, the boundedness of all closed-loop signals is rigorously proven, and a strictly positive lower bound on interevent intervals is derived to exclude Zeno behavior. Simulation results on a fractional-order single-machine infinite-bus power system demonstrate the effectiveness and superiority of the proposed scheme.

This paper investigates the mean-square leader-following consensus of stochastic delayed multi-agent systems (MASs) subject to hybrid cyber attacks. To begin with, for stochastic non-delayed and delayed MASs, novel non-delayed and delayed piecewise differential inequalities are designed, effectively integrating an event-triggered mechanism, intermittent, and impulsive control. Secondly, a novel closed-loop aperiodic intermittent impulsive control is designed using a Lyapunov-based event-triggered mechanism with exponential terms. Furthermore, under the impulsive control subject to Denial-of-Service attack and deception attack, sufficient criteria for secure consensus of stochastic non-delayed and delayed MASs are established through the proposed non-delayed and delayed piecewise differential inequality and closed-loop aperiodic intermittent control. Finally, two numerical simulations are presented to further validate the effectiveness of the theoretical analysis.

In this paper, the design of a fuzzy iterative learning tracking control scheme is investigated for a class of uncertain strict-feedback nonlinear systems. A key contribution of this paper is to handle composite uncertainties including varying trial lengths, unknown nonlinearities, and an unknown time-varying parameter. Firstly, a time-axis extension mechanism is introduced to deal with data discontinuities caused by variable trial lengths. Additionally, fuzzy logic systems are employed to approximate unknown nonlinearities, while sliding mode control is utilized to compensate for approximation errors. Furthermore, both original and projection-based learning laws are developed to estimate the unknown time-varying parameter. Notably, the dynamic surface control technique is adopted to construct error propagation channels for ensuring accurate trajectory tracking. Finally, on a representative electromechanical system model, simulation results are presented to demonstrate the effectiveness of the proposed control strategy. Note to Practitioners-Industrial control systems performing repetitive tasks, such as robotic assembly or manufacturing cycles, often struggle to maintain precise trajectory tracking due to composite uncertainties. These include unpredictable variations in task duration (causing incomplete data), unknown system dynamics, and changing operational parameters. This paper tackles these practical challenges by introducing a fuzzy iterative learning controller designed for a class of uncertain strict-feedback nonlinear systems. Our approach enables robust and precise trajectory tracking without requiring full prior knowledge of the system. Key innovations include a mechanism to seamlessly handle interrupted or variable-length operational cycles, intelligent components that automatically learn and compensate for unknown system behaviors, and methods to accurately estimate changing parameters during operation. The controller continuously refines its performance across repeated operational cycles. Validated on an electromechanical testbed, the method demonstrates reliable tracking accuracy where traditional controllers might fail. Future work will focus on reducing computational demands and extending the framework to networked multi-system operations.

Circular ribonucleic acids(circRNA) are key regulators in complex diseases, offering insights into disease mechanisms and therapeutic opportunities. Data sparsity and the challenge of extracting both local and global interactions across diverse biomolecules limit the performance of existing methods. In this work, we combine the merits of biomolecule graph modeling and representation learning from meta-relations, and devise an adaptive dual-channel graph transformer to extract complex interaction patterns among diverse biomolecules. We construct a meta-relation-guided heterogeneous biological graph integrating circRNAs, diseases, and other biomolecules. We then devise an adaptive dual-channel graph transformer with a dual gate control strategy to extract relation-specific interdependencies via node- and edge-level attention, while a feature boost connection enhances topological and semantic information. Extensive experiments on benchmark datasets show that the proposed model outperforms state-of-the-art methods, achieving improvements of 2.41% in AUROC and 6.68% in AUPR. The model effectively handles complex biological graph data, improving both association inference and interpretability for circRNA function prediction, and opens new avenues for exploring circRNA as therapeutic targets and vaccine carriers in complex diseases.

This study investigates a class of quasi-consensus problems with input delay. For the nonlinear heterogeneous multi-agent systems (MASs) with or without input delay, an event-triggered control protocol and an impulsive controller are designed to decrease the negative influence of input delay. Based on Lyapunov stability theory, sufficient conditions for achieving leader-following quasi-consensus (LFQC) are derived. The relationships among the event-triggering threshold, input delay, and parameters of the system dynamics are elucidated. It is proven that the constructed event-triggering function can exclude Zeno behavior. Finally, a simulation example is provided to verify the validity and feasibility of the theoretical conclusions.

This article addresses the problem of encryption -decryption-based bipartite synchronization control for a class of discrete-time coupled neural networks (CNNs), in which the nodes exhibit both cooperative and antagonistic interactions. Initially, a Markov chain with concealed operating modes is used to describe Markov jump CNNs (MJCNNs) with switching topologies (STs). In this framework, a hidden Markov model (HMM) is incorporated, whose emission values express the system mode. Next, the decentralized adaptive event-triggered strategy is proposed to alleviate the communication burden caused by interactions between nodes. Moreover, an encryption-decryption algorithm (EDA) that takes into account identity authentication is programmed to encrypt the data at the triggering moment of each node, thereby securing the data interaction privacy. Then, the observation-mode-based bipartite synchronization control law is formulated to fulfill the control demands of the plant. Furthermore, some sufficient conditions for the networks to be mean square synchronized and satisfy the H-infinity performance are obtained based on the Lyapunov stability theory. Finally, two simulation examples involving chaotic neural networks (NNs) are presented to verify the effectiveness of the proposed method.

A novel predefined-time optimal formation control scheme is developed for fixed-wing multi-unmanned aerial vehicle (UAV) systems subject to model uncertainties and external disturbances. Unlike current state-of-the-art methods, the proposed optimal controller integrates a predefined-time estimation mechanism within an actor-critic framework. This integration guarantees both optimal performance and system stability within a user-specified settling time. Within this framework, the proposed actor weight estimation protocol is jointly updated with the critic in an online manner, bypassing the limitations of traditional reinforcement learning (RL) algorithms reliant on independent updates via projection operators. A rigorous proof of bounded estimation errors in RL updates is provided, obviating the need for the conventional premise of a priori error bounds. Furthermore, an improved distributed formation tracking control framework is designed. This framework not only inherits the capability to maintain relative position relationships but also ensures heading synchronization with the leader. Accordingly, a devised disturbance observer suppresses the propagation effect of complex disturbances within the relative motion model. Experimental validation is conducted using a dedicated hardware-in-the-loop (HIL) testbed, integrating commercial flight controllers with a real-time simulator. Both comprehensive numerical simulations and experimental results conclusively demonstrate the effectiveness and superior performance of the proposed control scheme.

This paper addresses the critical challenge of distributed fault-tolerant projective group consensus control for multi-agent systems operating under both the input constraints and persistent external disturbances. Unlike conventional approaches, this study proposes a novel framework that provides comprehensive solution for multi-fault scenarios by modeling diverse fault dynamics through time-varying polynomial formulations and polyhedral structures. This advanced representation accommodates a wide spectrum of fault modes without requiring strict matching conditions, thereby significantly enhancing both flexibility and practical applicability. To achieve accurate estimation of system faults and disturbances, we develop a distributed composite observer. This observer utilizes relative output error signals between neighboring agents while effectively decoupling the interactions between faults and disturbances. Furthermore, we introduce an innovative distributed anti-saturation fault-tolerant control strategy that synergistically integrates the observer's estimates with global output information. This approach guarantees robust performance despite simultaneous input constraints and multiple fault conditions. Through simulation studies, the proposed control framework demonstrates its effectiveness, confirming its potential for real-world implementation in complex multi-agent systems.

This study handles the robust sampled-data H infinity fuzzy control analysis for a category of nonlinear partial differential systems (NPDSs) holding disturbances. As for now, the Takagi-Sugeno (T-S) fuzzy model serves superior by describing a broad category of nonlinear systems, and therefore, originally, a T-S fuzzy model is employed to illustrate the nonlinear parabolic partial differential systems. Here, the primary focus of this research is on designing a resilient sampled-data H-infinity fuzzy estimator-based controller which is competent in stabilizing the T-S fuzzy closed-loop partial differential systems (PDSs) and to tolerate the disruption under a specified level. By the virtue of Lyapunov stability theory, Green's formula and several inequality techniques, the robust stabilization design problem based on a sampled-data fuzzy H-infinity estimator is effectively addressed using a set of linear matrix inequalities (LMIs). Moreover, the impacts of the diffusion phenomenon and the designed controller are clearly reflected in the derived criteria. Further, the acquired criteria can be checked for their practicability by the virtue of MATLAB LMI control toolbox. Finally, simulation results are presented to demonstrate the effectiveness of the proposed criteria.

In this article, we study the stability and bifurcations of a fractional-order neutral neural network with three types of delays. A four-dimensional fractional-order neutral-type neural network (FONTNN) is firstly established. Secondly, the bifurcation results of the proposed FONTNN are extracted by the analytical method of characteristic equations and Cramer’s rule. It demonstrates that FONTNN can neatly improve the performance stability of the system in comparison with integer-order neural networks. Moreover, the influence of fractional orders is nicely explored. It discovers that the bifurcation points extremely depend on fractional orders. The stability performance can be adjusted by selecting some appropriate fractional orders. The authenticity of the developed theoretical outcomes is ultimately corroborated through numerical simulations.

This paper constructs a coupled population-economy reaction-diffusion model with asymmetric density-dependent diffusion, aiming to theoretically reveal how the mechanism of “congestion inhibiting diffusion and agglomeration enhancing diffusion” drives the self-organization of urban and regional spatial patterns. The critical conditions for Turing instability and the characteristic wavelength are rigorously derived, clarifying the regulatory role of density-dependent parameters on the instability threshold. Furthermore, the amplitude equations are derived using weakly nonlinear analysis, revealing the selection and stability mechanisms of spatial patterns near the bifurcation point. The interaction between Hopf bifurcation and Turing instability is also explored, and by establishing coupled amplitude equations, the resulting rich spatiotemporal dynamics are characterized. Numerical simulations are in excellent agreement with theoretical predictions and further reveal that, compared to linear diffusion, the density-dependent diffusion significantly accelerates the formation of spatial heterogeneous structures by introducing dynamic feedback mechanisms.