Adapting Large Multimodal Models (LMMs) to real-world scenarios poses the dual challenges of learning from sequential data streams while handling frequent modality incompleteness, a task known as Continual Missing Modality Learning (CMML). However, existing works on CMML have predominantly relied on prompt tuning, a technique that struggles with this task due to cross-task interference between its learnable prompts in their shared embedding space. A naive application of Low-Rank Adaptation (LoRA) with modality-shared module will also suffer modality interference from competing gradients. To this end, we propose DeLo, the first framework to leverage a novel dual-decomposed low-rank expert architecture for CMML. Specifically, this architecture resolves modality interference through decomposed LoRA expert, dynamically composing LoRA update matrix with rank-one factors from disentangled modality-specific factor pools. Embedded within a task-partitioned framework that structurally prevents catastrophic forgetting, this expert system is supported by two key mechanisms: a Cross-Modal Guided Routing strategy to handle incomplete data and a Task-Key Memory for efficient, task-agnostic inference. Extensive experiments on established CMML benchmarks demonstrate that our method significantly outperforms state-of-the-art approaches. This highlights the value of a principled, architecturally-aware LoRA design for real-world multimodal challenges.

This article addresses general decay synchronization matter for multiplex and directed networks via delayed feedback control. Decay synchronization is regarded as a class of ψ-type synchronization, which derives from the generalizations of ψ-type function and ψ-type stability. By exploiting appropriate nonlinear control positioned on a portion of multiplex networks, synchronization for the network system is solved with decay rate. In comparison with previous multiplex networks, the present model contains asymmetric, with non-cooperative factors and not connected outer matrices, combined with negative elements of inner matrices in the article, which improves existing results well. We propose a synchronization method for multiplex network under this constraint from angle of inner matrices. It is proved that if weighted group of union new matrices for each dimension is strongly connected, then decay synchronization and anti-synchronization can be realized under delayed feedback controller. Moreover, some specific modes for synchronization are illustrated more precisely. In addition, reaction-diffusion systems are also further conducted as an application. Simulations are given for verifying the validity of gained results.

In this paper, we tackle Generalized Category Discovery (GCD) by drawing inspiration from the platypus—a creature that uniquely blends features from different species. Our method bridges the gap between known and unknown categories through a novel cross-instance feature learning paradigm. Unlike traditional GCD methods, we dynamically mix patches from multiple images, integrating features across instances. This process is enhanced by a progressive mixing strategy that evolves from focusing on labeled data to incorporating both labeled and unlabeled data. Complemented by a hierarchical contrastive learning framework that enforces constraints at global, mixed-origin, and inter-mixed levels, our approach effectively generalizes across different feature spaces. Extensive experiments on six benchmark datasets demonstrate our method’s superior performance in recognizing known classes and discovering new ones, setting a new standard in GCD tasks.

Remote sensing object detection plays a critical role in analyzing aerial imagery, enabling the accurate localization of arbitrarily oriented objects. Traditional detection frameworks often suffer from computational inefficiencies and feature misalignment due to the reliance on pre-defined rotated anchors and fixed convolutional sampling patterns. To address these limitations, we propose ASOD-Net, an Arbitrary Sampling and Offset-based Detection Network, which integrates Arbitrary Sampling Convolution (ASConv) and a differential offset-based bounding box representation. ASConv dynamically adjusts sampling positions to improve feature adaptability, while the offset-based bounding box representation enhances regression stability and reduces dependency on anchor angles. Extensive experiments on DOTA, HRSC2016, and UCAS-AOD demonstrate that our approach significantly improves detection accuracy, robustness, and computational efficiency.

Causal mediation analysis is crucial for deconstructing complex mechanisms of action. However, in current mediation analysis, complex structures derived from causal discovery lack direct interpretation of mediation pathways, while traditional mediation analysis and effect estimation are limited by the reliance on pre-specified pathways, leading to a disconnection between structure discovery and causal mechanism understanding. Therefore, a unified framework integrating structure discovery, pathway identification, and effect estimation systematically quantifies mediation pathways under structural uncertainty, enabling automated identification and inference of mediation pathways. To this end, we propose Structure-Informed Guided Mediation Analysis (SIGMA), which guides automated mediation pathway identification through probabilistic causal structure discovery and uncertainty quantification, enabling end-to-end propagation of structural uncertainty from structure learning to effect estimation. Specifically, SIGMA employs differentiable Flow-Structural Equation Models to learn structural posteriors, generating diverse Directed Acyclic Graphs (DAGs) to quantify structural uncertainty. Based on these DAGs, we introduce the Path Stability Score to evaluate the marginal probability of pathways, identifying high-confidence mediation paths. For identified mediation pathways, we integrate Efficient Influence Functions with Bayesian model averaging to fuse within-structure estimation uncertainty and between-structure effect variation, propagating uncertainty to the final effect estimates. In synthetic data experiments, SIGMA achieves state-of-the-art performance in pathway identification accuracy and effect quantification precision under structures uncertainty, concurrent multiple pathways, and nonlinear scenarios. In real-world applications using Human Phenotype Project data, SIGMA identifies mediation effects of sleep quality on cardiovascular health through inflammatory and metabolic pathways, uncovering previously unspecified multiple mediation paths.

This paper investigates the practical prescribed-time (PT) synchronization and control for the multiweighted complex networks (MCNs) with or without external disturbances. First, due to the possible out of memory and practical error tolerance, we define a class of general time-varying regulatory functions for practical PT stability, which can contain many previous special forms. The classic PT regulatory function is replaced with a positive constant after the practical settling time according to the tolerant precision. Dynamical systems are proved to achieve practical PT stability at the practical settling time, which can be calculated under a given condition, then exponentially converge after this time. Both the settling time and the tolerance precision can be prescribed arbitrarily. Next, taking the external disturbances into account, both PT and practical PT stability are strictly obtained. Furthermore, practical PT synchronization of MCNs is also investigated, i.e., its synchronization error can enter to a prescribed-precision region after prescribed time surely and exponentially converge to zero. At last, numerical simulations are provided to evaluate the effectiveness of theoretical results.

Remote sensing data are of considerable significance for monitoring global climate, detecting harmful algae bloom, and so on. However, due to sensor failures, cloud cover, and thick aerosols, the collected remote sensing data for various ocean factors, such as chlorophyll-a (Chl-a) concentration and sea surface temperature (SST), often have a high missing rate, which seriously hinders their applications. Existing data imputation models mostly ignore highly sparse spatial locations or perform badly when the data missing rate is high due to the lack of available information. Large language models (LLMs) possess powerful representation learning capabilities and can effectively capture sequential correlations even with extremely limited information, thereby presenting the promising potential for highly sparse remote sensing data imputation. Therefore, we proposed a novel LLM-based spatiotemporal model for highly sparse remote sensing data imputation, i.e., LLM4HRS. First, LLM4HRS develops an LLM-based bidirectional temporal representation learning module to learn forward and backward temporal dependencies in data sequences and fuses them together to obtain comprehensive temporal representations. Next, LLM4HRS constructs a LLM-based spatial representation learning module to learn spatial correlations with the learned temporal representation. Finally, a spatiotemporal representation fusion and data imputation module is developed to achieve data imputation. Experiments on the SST and Chl-a remote sensing datasets demonstrate that LLM4HRS significantly outperforms existing data imputation models, with its advantages becoming more pronounced as the masking rate increases. Furthermore, when extended to the remote sensing PAR data in a large region, LLM4HRS still achieves the best performance, further validating its broad applicability for remote sensing data imputation. The code of LLM4HRS is publicly available athttps://github.com/ssyuwang/LLM4HRS-master.

This article concerns output synchronization (OSyn) and proportional-integral-derivative (PID) control problems for networks with directed and multiple output communications. Outer coupling matrices can be direct, competitive, and even not connected, and inner coupling matrices are also utilized to discuss the OSyn. At first, a new assumption for the uncoupled node is proposed, which plays an important role in the following analysis. Then, a sufficient condition is established for networks with a single outer coupling matrix. Moreover, OSyn is also addressed by exploiting simpler and more practical PID controller. Furthermore, we solve OSyn for directed networks with multiple communications in virtue of two developed techniques, where PID control criteria can be established. Examples are provided to verify the effectiveness of those obtained results.

In this paper, we study the synchronization problem of directed multi-level complex networks (MLevelN). Firstly, directed multi-layer complex networks with multiple weights are successfully transformed to directed multi-layer complex networks with a single weight by using the newly developed re-arranging variables’ order technique. Moreover, as a typical representative of the MLevelN model, we propose some novel Lyapunov functions by the Kronecker product of the left eigenvectors for multi-layer networks, and we also illustrate the processing procedure can be simply extended to more complicated MLevelN. Furthermore, the pinning control strategies for various synchronization types are also designed with constant and adaptive strengths respectively. The obtained criteria can greatly improve the conservation of previous works, whose effectiveness is demonstrated by numerical simulations.

Monitoring Chlorophyll-a (Chl-a) concentration in ocean is of considerable significance for early warning of algae disasters, marine ecological environment protection, etc. However, due to various uncontrollable factors such as cloud cover and thick aerosols, the missing rate of observed Chl-a data is quite high, which seriously hinders its applications. The existing data imputation methods are mostly applicable to Chl-a data with a low missing rate, and perform much worse when the missing rate reaches 0.7 or above. To address this issue, we proposed a self-supervised spatiotemporal imputation (S2-STI) model for highly sparse Chl-a data imputation by fusing multisource satellite data. First, to obtain comprehensive information about the missed Chl-a data, we designed a multisource sparse data fusion (MSDF) module to fuse multisource satellite data, including Chl-a data, sea surface temperature (SST) data, and photosynthetically available radiation data. MSDF constructs a spatialtemporal graph network to learn high-quality spatiotemporal representations of SST and photosynthetically active radiation (PAR) data, and fuses the learned representations with Chl-a data to enrich the information for imputation. Then, we developed a generation-based data imputation (GDI) module to model the distribution of Chl-a data based on the outputs of MSDF. Considering the high data sparsity, we designed a self-supervised training strategy to train S2-STI in the absence of ground truth. Finally, we leverage the data generated by the GDI module in the trained S2-STI model to fill in missing values in the sparse Chl-a data. Experiments on real datasets show that S2-STI achieves much better performance than the existing data imputation methods. Specifically, for the Chl-a data with a missing rate of 0.9, S2-STI improves by at least 12% in terms of masked autoencoder (MAE) error when compared with the strong baseline methods.

We conduct the study of passivity and control for multiweighted and directed fractional-order network systems (MDFONSs) in this article. A new concept of fractional-order passivity (FOP) is defined, which also contains the integer-order passivity. In the literature of multiweighted networks, many papers usually study its passivity only from the viewpoint of outer coupling matrices (OMs), which are also assumed to be connected and undirected. In this article, we add the viewpoint of inner coupling matrices (IMs), and the OMs can be directed and not connected, which can greatly improve the existing results. By means of decomposing IMs into their main diagonal matrices and residual matrices, we obtain that if the weighted combination of multiple OMs for each dimension is strongly connected, then FOP can be realized. Of course, the above results also hold for diagonal IMs, which is commonly addressed in previous works. Moreover, synchronization, adaptive coupling strengths and pinning control are also discussed. Besides, FOP and control rules for multiweighted and directed fractional-order reaction-diffusion network systems (MDFORDNSs) are derived by applying this strategy. Numerical examples are ultimately employed to examine the effectiveness of these gained results.

In this article, the passivity and control issues for directed uncertain coupled dynamical networks are solved. The presented model is directly coupled with multiple coupling matrices and parametric uncertainty, while previous literatures of multiweighted networks usually suppose that outer coupling matrices (OMs) are connected, undirected, and certain. The viewpoint of inner coupling matrices (IMs) in this article is added and OMs can be directed and not connected, which is a great improvement on the existing results. First, for all diagonal IMs, considering each dimension separately, we can derive if the weighted combination of multiple OMs for each dimension is strongly connected, then passivity and pinning control rules can be established. In addition, we also discuss the situation that IMs are positive definite but not diagonal. By means of the weighted combination of normalized left eigenvectors (NLEVec) corresponding to zero eigenvalue for multiple coupling matrices, we prove if the Chebyshev distance (Cheb-Dist) among these NLEVec is less than a tolerant deviation interval, then passivity, synchronization, and pinning control criteria are acquired. Moreover, a matter of adaptive coupling strengths is also settled. Examples are provided to verify the validity of established results.

The study of complex networks with multiweights (CNMWs) has been a hot topic recently. For a network with a single weight, previous studies have shown that they can promote synchronization, but for CNMWs, there are no rigorous analyses about the role of coupling matrices. In this brief, the complex network is allowed to be directed, which is the main difference with previous studies and may make the synchronization analysis difficult for multiple couplings. At first, we prove that if the inner coupling matrices are all diagonal, then synchronization can be realized only if the weighted sum (or union) of multiple coupling matrices is strongly connected, which bridges the gap between single-weighted and multiweighted networks. Moreover, we also consider the case that inner coupling matrices are positive definite but not diagonal. We design two techniques for this hard problem. One technique is to decompose inner coupling matrices into diagonal matrices and residual matrices. The other one is to measure the similarity between outer coupling matrices. In virtue of the normalized left eigenvectors (NLEVecs) corresponding to the zero eigenvalue of coupling matrices, we prove that if the Chebyshev distance between NLEVec is less than some value, defined as the allowable deviation bound, then the synchronization and control will be realized with sufficiently large coupling strengths. Furthermore, adaptive rules are also designed for coupling strength.

This article is dedicated to addressing the generalized decay synchronization problem of directly coupled stochastic dynamic networks with multiple time delays. First, by means of the presentations of the$\psi $-type function and$\psi $-type stability, the definitions of decay synchronization$(\psi $-type synchronization) are introduced. After that, by constructing appropriate feedback controllers and the Lyapunov functional method, we study the decay synchronization for multiple coupled dynamic networks. Moreover, for different coupling matrices, their zero eigenvalues are correspondingly denoted by different normalized left eigenvectors (NLEVec), so it is difficult to establish the functional based on these NLEVec. With the help of the weighted combination of NLEVec for multiple coupling matrices, under the assumption that the Chebyshev gap among NLEVec is less than an allowable deviation interval, some generalized decay (anti-) synchronization conditions are established. Furthermore, we discuss exponential synchronization and polynomial synchronization as special cases, and more relevant criteria are acquired. Examples are provided to verify the effectiveness of those obtained conditions.

We investigate a general way to achieve finite-time synchronization (FnTSYN) for networks via aperiodically intermittent control (AInC). At first, we propose a general sufficient criteria under AInC for finite-time stability (FnTSta). Then, a general model for complex networks under this sufficient condition is given. Furthermore, we apply the proposed condition to the spatially coupled reaction-diffusion neural networks (RDNNs) to realize its FnTSYN. Finally, simulations are given to verify the results.

This article uses a multiple compliant degree-of-freedoms (DOFs) mechanism (i.e., a spring) to facilitate compliant insertion in precision assembly and proposes an efficient insertion strategy accordingly. The addition of a spring increases the insertion compliance while resulting in the object being not directly controllable. The proposed strategy handles both vertical insertion and inclined insertion with an unknown posture according to force feedback. We investigate horizontal compliance when the spring is compressed or stretched and introduce the withdrawal process for exceptional situations by taking advantage of the insertion compliance. The inclined insertion is a process of inserting while estimating the hole posture and a radial compensation strategy is presented while not affecting the axial length of the spring. Efficient insertion planning is discussed in the presence of uncertainty caused by the spring for both insertion types. Experiments are carried out to demonstrate the validation of the proposed method.

We are making good progresses in our impact and reputation over the last year. According to the latest data released by Scopus on February 11, 2019, our CiteScore hits its historical high to 3.94, and TCSS ranks 8th out of the 226 journals (top 3.54%) in the field of social sciences. This is a solid improvement compared with the corresponding data in 2017 (CiteScore: 2.36, Rank: 17/226, top 8%). Thanks and congratulations to our authors, reviewers, and members of our editorial board. The current issue includes 17 regular papers and a brief discussion on social education.

New intelligent era calls for new learners and thus urgently needs a series of artificial intelligence. As a good educational platform for teaching artificial intelligence, smart cars have aroused concern and practices of all parties. However, at present, most courses and training pay more attention to basic knowledge and technology of smart cars, seldom to training based on artificial intelligence curriculum system and comprehensive competency integrating science, technology, art and management. Therefore, based on concept of iSTREAM (intelligence for Science, Technology, Robotics, Engineering, Art, and Management) and Raspberry intelligent vehicle teaching platform, this paper introduced a smart car-themed artificial intelligence courses including basic courses, specialized courses, specialized technical courses and elective courses. This course can guide learners to develop smart cars based on visual cognition, in-depth learning, VR and 3D printing integrated artistic creativity. It combines disciplines such as science, technology, art, games and management to upgrade a single knowledge and technology course into a comprehensive competency course that integrates knowledge, skills, emotion and management. Practice in Beijing NO.13 and NO.101 High School shows that this course allows students to experience scientific research process, learn artificial intelligence related knowledge and skills, understand scientific way of thinking and scientific research methods, stimulate learners' responsibility and scientific passion, and cultivate leadership skills through self-learning and partly project management.

In this paper, the finite-time and fixed-time cluster synchronization problem for complex networks with or without pinning control are discussed. Finite-time (or fixed-time) synchronization has been a hot topic in recent years, which means that the network can achieve synchronization in finite-time, and the settling time depends on the initial values for finite-time synchronization (or the settling time is bounded by a constant for any initial values for fixed-time synchronization). To realize the finite-time and fixed-time cluster synchronization, some simple distributed protocols with or without pinning control are designed and the effectiveness is rigorously proved. Several sufficient criteria are also obtained to clarify the effects of coupling terms for finite-time and fixed-time cluster synchronization. Especially, when the cluster number is one, the cluster synchronization becomes the complete synchronization problem; when the network has only one node, the coupling term between nodes will disappear, and the synchronization problem becomes the simplest master-slave case, which also includes the stability problem for nonlinear systems like neural networks. All these cases are also discussed. Finally, numerical simulations are presented to demonstrate the correctness of obtained theoretical results.

Reducing operation and the maintenance costs of wind turbines has become a primary issue of wind farm owners and operators. Since the supervisory control and data acquisition (SCADA) system has been widely used in wind farms, it is costeffective to use SCADA data to realize condition monitoring. To this end, this paper proposes a method to calculate health index of wind turbines based on SCADA data. This method uses kernel density estimatison (KDE), then calculate the relative health index (RHI) of all wind turbines in a certain wind farm. To show the effectiveness of the method, we apply our method to real SCADA data of several wind farms. The result shows that RHI can reflect the health status of each wind turbine. Therefore, it is convenient to maintain the turbines with low RHI, thus to save maintenance costs.

In the paper, considering the limitation of effective method in E-learning area, a recommendation framework for E-Learning based on deep learning is proposed. Our model is based on deep learning, which has strong capability to learn from large-scale data. It has some improvements than previous methods. First, it is based on the conventional K-Nearnest Neighbor(KNN) method to train a model, thus its accuracy is guaranteed. Second, it can recommend the new item whose similarity can not be calculated. Third, it greatly reduces the heavy burden for a running system, which is useful in real practice of recommendation systems. In conclusion, the proposed framework can offer a new recommendation method for more personlized learning in the future.

The stability of the zero solution plays an important role in the investigation of positive systems. In this brief, we discuss the μ -stability of positive nonlinear systems with unbounded time-varying delays. The system is modeled by the continuous-time ordinary differential equation. Under some assumptions on the nonlinear functions, such as homogeneous, cooperative, and nondecreasing, we propose a novel transform, by which the nonlinear system reduces to a new system. Thus, we analyze its dynamics, which can simplify the nonlinear homogenous functions with respect to the arbitrary dilation map to those with respect to the standard dilation map. We finally get some new criteria for the global μ -stability taking the degree into consideration. A numerical example is given to demonstrate the validity of obtained results.

In this paper, we investigate the global exponential stability for complex-valued recurrent neural networks with asynchronous time delays by decomposing complex-valued networks to real and imaginary parts and construct an equivalent real-valued system. The network model is described by a continuous-time equation. There are two main differences of this paper with previous works: 1) time delays can be asynchronous, i.e., delays between different nodes are different, which make our model more general and 2) we prove the exponential convergence directly, while the existence and uniqueness of the equilibrium point is just a direct consequence of the exponential convergence. Using three generalized norms, we present some sufficient conditions for the uniqueness and global exponential stability of the equilibrium point for delayed complex-valued neural networks. These conditions in our results are less restrictive because of our consideration of the excitatory and inhibitory effects between neurons; so previous works of other researchers can be extended. Finally, some numerical simulations are given to demonstrate the correctness of our obtained results.

In this paper, pinning synchronization problem for nonlinear coupled networks is investigated, which can be recurrently connected neural networks, cellular neural networks, Hodgkin-Huxley models, Lorenz chaotic oscillators, and so on. Nodes in the network are assumed to be identical and nodes' dynamical behaviors are described by continuous-time equations. The network topology is undirected and static. At first, the scope of accepted nonlinear coupling functions is defined, and the effect of nonlinear coupling functions on synchronization is carefully discussed. Then, the pinning control technique is used for synchronization, especially the control type is aperiodically intermittent. Some sufficient conditions to guarantee global synchronization are presented. Furthermore, the adaptive approach is also applied on the pinning control, and a centralized adaptive algorithm is designed and its validity is also proved. Finally, several numerical simulations are given to verify the obtained theoretical results.

In this paper, we investigate the exponential synchronization problem for linearly coupled networks with delay by pinning a simple aperiodically intermittent controller. The network topology can be directed. Different from previous works, the intermittent control can be aperiodic. Two types of delay are considered. The first case is that the delay is time-varying and large, and in this case, there is no restriction imposed on the delay and the control (and/or rest) width. The other one is that the delay is small enough so that it is less than the minimum of control width. Different approaches are provided to investigate these two cases, and some criteria are given to realize exponential synchronization. Furthermore, by applying the adaptive approach to the second model, we establish a general adaptive theory for intermittent control, which can be applied not only to networks without time delay, but also to delayed networks, regardless of whether the intermittent control is periodic or aperiodic. Finally, the numerical simulations are given to verify the validness of the theoretical results.

In this paper, we investigate the cluster synchronization problem for linearly coupled networks, which can be recurrently connected neural networks, cellular neural networks, Hodgkin-Huxley models, Lorenz chaotic oscillators, etc., by adding some simple intermittent pinning controls. We assume the nodes in the network to be identical and the coupling matrix to be asymmetric. Some sufficient conditions to guarantee global cluster synchronization are presented. Furthermore, a centralized adaptive intermittent control is introduced and theoretical analysis is provided. Then, by applying the adaptive approach on the diagonal submatrices of the asymmetric coupling matrix, we also get the corresponding cluster synchronization result. Finally, numerical simulations are given to verify the theoretical results.