Out-of-distribution (OOD) detection plays a critical role in ensuring the robustness of machine learning models in open-world settings. While extensive efforts have been made in vision, language, and graph domains, the challenge of OOD detection in hypergraph-structured data remains unexplored. In this work, we formalize the problem of hypergraph out-of-distribution (HOOD) detection, which aims to identify nodes or hyperedges whose high-order relational contexts differ significantly from those seen during training. We propose HyperGOOD, a unified energy-based detection framework that integrates multi-scale spectral decomposition with structure-aware uncertainty propagation. By preserving both low- and high-frequency signals and diffusing uncertainty across the hypergraph, HyperGOOD effectively captures subtle and relationally entangled anomalies. Experimental results on nine hypergraph datasets demonstrate the effectiveness of our approach, establishing a new foundation for robust hypergraph learning under distributional shifts.

Multimodal recommendation provides richer and more accurate personalized recommendations by jointly modeling user’s historical behaviors and different modality of items, such as text, image, audio, and video in online platforms. Most existing work of multimodal recommendation focuses on leveraging modal features and modal correlation graph structures to learn user preferences. Due to insufficient exploration of user collaborative preferences and the noise during high-order multimodal data connections, valuable information may be lost, leading to deviations in understanding user preferences. Therefore, an I ndependent T opological Preference-Aware and Co operative H ypergraph D iffusion-based M ultimodal Rec ommender Model (ITCoHD-MRec) is necessary for online platforms. This article aims to develop an ITCoHD-MRec that incorporates topological perception as well as generative diffusion models in multimodal hypergraph recommendation to make the model more adaptive and robust in complex environments. Firstly, leveraging a Graph Convolutional Network (GCN), the model independently captures user preference representations for both collaborative relevance and modal relevance from the user-item interaction graph, which contains ID embeddings and modal features. This enables the extraction of deeper associations between users and items. Secondly, leveraging topological pruning techniques, the model learns differentiated features in different modal blocks to prevent node representations from becoming homogenized. This helps further identify user preferred connectivity patterns and removes redundant noisy connections. Finally, by employing the diffusion model, information regarding the higher-order interaction patterns between attributes and items within the hypergraph structure is propagated. This effectively captures the potential global dependencies between attributes and items, thereby providing deeper associations enriched with more substantial semantic information for subsequent recommendation tasks. The model autonomously learns different features and higher-order connectivity of nodes, which enables the model to obtain a wider and more accurate perception of user preferences in complex interaction environments. Experimental comparisons with 15 models on four real datasets—Baby, Sports, Clothing, and Electronics show that the model improves the recall by 0.85%–3.57% and the normalized discounted cumulative gain by 2.31%–3.43%, which validates the effectiveness of ITCoHD-MRec.

The out-of-distribution (OOD) detection on graph-structured data is crucial for deploying graph neural networks securely in open-world scenarios. However, existing methods have overlooked the prevalent scenario of multi-label classification in real-world applications. In this work, we investigate the unexplored issue of OOD detection within multi-label node classification tasks. We propose ML-GOOD, a simple yet sufficient approach that utilizes an energy function to gauge the OOD score for each label. We further develop a strategy for amalgamating multiple label energies, allowing for the comprehensive utilization of label information to tackle the primary challenges encountered in multi-label scenarios. Extensive experimentation conducted on seven diverse sets of real-world multi-label graph datasets, encompassing cross-domain scenarios. The results show that the AUROC of ML-GOOD is improved by 5.26% in intra-domain and 6.54% in cross-domain compared to the previous methods. These empirical validations not only affirm the robustness of our methodology but also illuminate new avenues for further exploration within this burgeoning field of research.

Time series forecasting has various applications, such as meteorological rainfall prediction, traffic flow analysis, financial forecasting, and operational load monitoring for various systems. Due to the sparsity of time series data, relying solely on time-domain or frequency-domain modeling limits the model's ability to fully leverage multi-domain information. Moreover, when applied to time series forecasting tasks, traditional attention mechanisms tend to over-focus on irrelevant historical information, which may introduce noise into the prediction process, leading to biased results. We proposed WDformer, a wavelet-based differential Transformer model. This study employs the wavelet transform to conduct a multi-resolution analysis of time series data. By leveraging the advantages of joint representation in the time-frequency domain, it accurately extracts the key information components that reflect the essential characteristics of the data. Furthermore, we apply attention mechanisms on inverted dimensions, allowing the attention mechanism to capture relationships between multiple variables. When performing attention calculations, we introduced the differential attention mechanism, which computes the attention score by taking the difference between two separate softmax attention matrices. This approach enables the model to focus more on important information and reduce noise. WDformer has achieved state-of-the-art (SOTA) results on multiple challenging real-world datasets, demonstrating its accuracy and effectiveness. Code is available at https://github.com/xiaowangbc/WDformer.

With the growing adoption of Hypergraph Neural Networks (HNNs) to model higher-order relationships in complex data, concerns about their security and robustness have become increasingly important. However, current security research often overlooks the unique structural characteristics of hypergraph models when developing adversarial attack and defense strategies. To address this gap, we demonstrate that hypergraphs are particularly vulnerable to node injection attacks, which align closely with real-world applications. Through empirical analysis, we develop a relatively unnoticeable attack approach by monitoring changes in homophily and leveraging this self-regulating property to enhance stealth. Building on these insights, we introduce HyperNear, i.e., $\underline{N}$ode inj$\underline{E}$ction $\underline{A}$ttacks on hype$\underline{R}$graph neural networks, the first node injection attack framework specifically tailored for HNNs. HyperNear integrates homophily-preserving strategies to optimize both stealth and attack effectiveness. Extensive experiments show that HyperNear achieves excellent performance and generalization, marking the first comprehensive study of injection attacks on hypergraphs. Our code is available at https://github.com/ca1man-2022/HyperNear.

Facial expressions in videos inherently mirror the dynamic nature of real-world facial events. Consequently, facial expression recognition (FER) should employ a dynamic graph-based representation to effectively capture the relational structure of facial expressions rather than relying on conventional grid or sequence methods. However, existing graph-based approaches have their limitations. Frame-level graph methods provide a coarse representation of the facial graph across time and space, while landmark-based graph methods need to introduce additional facial landmarks, resulting in a static graph structure. To address these challenges, we propose spatial-temporal relation-aware dynamic graph convolutional networks (ST-RDGCN). This fine-grained relation modeling approach enables the dynamic modeling of evolving facial expressions in videos through dynamic space-time graphs, eliminating the need for facial landmarks. ST-RDGCN encompasses three graph construction paradigms: dynamic independent space graph, dynamic joint space-time graph, and dynamic cross space-time graph. Furthermore, we propose a relation-aware space-time graph convolution (RSTG-Conv) operator to learn informative spatiotemporal correlations in dynamic space-time graphs. In extensive experimental evaluations, our ST-RDGCN demonstrates state-of-the-art performance on the five popular video-based FER datasets, achieving overall accuracy scores of 99.69%, 91.67%, 56.51%, 69.37%, and 49.03% on the CK+, Oulu-CASIA, AFEW, DFEW, and FERV39k datasets, respectively. In particular, our ST-RDGCN outperforms the current best method by 3.6% in UAR on the most challenging FERV39k dataset. Furthermore, our analysis reveals that the dynamic cross space-time graph scheme is the most effective among the three dynamic graph construction schemes.

This paper presents a novel approach that leverages two models to integrate features from numerous unlabeled images, addressing the challenge of semi-supervised salient object detection (SSOD). Unlike conventional methods that rely on selecting high-quality pseudo labels, our method identifies the model that produces consistent predictions for original images and their color transformation versions from two models to infer reliable pseudo labels for all unlabeled images, improving the diversity of the training set. Specifically, we propose adaptive selection indicators to quantify prediction differences and guide the updates of the two models using the unlabeled set alternatively. Initially, two models used in our framework are trained on the labeled set. Once the adaptive selection indicator conditions are satisfied, one model is designated as the proxy, generating pseudo labels, while the other serves as the saliency model, which is further trained using these pseudo labels. Subsequently, the updated saliency model optimizes the proxy model’s parameters according to another adaptive selection indicator. Experimental results and ablation studies on six benchmark salient object detection datasets confirm the effectiveness and robustness of our method. Our approach achieves performance comparable to recent fully supervised methods while using only one eighth of the labeled data, demonstrating its potential for efficient and scalable SSOD. This paper is publicly available athttps://github.com/Liyuan0905/CATNet.

Recently, researchers have proposed many deep generative models, including generative adversarial networks (GANs) and denoising diffusion models. Although significant breakthroughs have been made and empirical success has been achieved with the GAN, its mathematical underpinnings remain relatively unknown. This paper focuses on a rigorous mathematical theoretical framework: the composite-functional-gradient GAN (CFG). Specifically, we reveal the theoretical connection between the CFG model and score-based models. We find that the CFG discriminator’s training objective is equivalent to finding an optimal$D(\mathrm {x})$. The optimal$D(\mathrm {x})$’s gradient differentiates the integral of the differences between the score functions of real and synthesized samples. Conversely, training the CFG generator involves finding an optimal$G(\mathrm {x})$that minimizes this difference. In this paper, we aim to derive an annealed weight preceding the CFG discriminator’s weight. This new explicit theoretical explanation model is called the annealed CFG method. To overcome the annealed CFG method’s limitation, as the method is not readily applicable to the state-of-the-art (SOTA) GAN model, we propose a nested annealed training scheme (NATS). This scheme keeps the annealed weight from the CFG method and can be seamlessly adapted to various GAN models, no matter their structural, loss, or regularization differences. We conduct thorough experimental evaluations on various benchmark datasets for image generation. The results show that our annealed CFG and NATS methods significantly improve the synthesized samples’ quality and diversity. This improvement is clear when comparing the CFG method and the SOTA GAN models.

Graph anomaly detection (GAD) has attracted increasing interest due to its critical role in diverse real-world applications. Graph neural networks (GNNs) offer a promising avenue for GAD, leveraging their exceptional capacity to model complex graph structures and relationships. However, existing GNN-based models encounter challenges in addressing the GAD's fundamental issue-anomaly camouflage, where anomalies mimic normal instances, leading to indistinguishable features. In this article, we propose a novel approach, termed correlation information enhanced GAD (CIE-GAD). Specifically, drawing on the observation that the distribution of homophilic and heterophilic edges differs between abnormal and normal samples, we construct a hypergraph to learn the co-occurrence relationships among adjacent edges. By enhancing the extraction of sample correlation information, we effectively tackle feature similarity caused by anomaly camouflage, thereby enhancing the performance of GAD. Furthermore, we develop a spectral convolution mechanism based on node-level attention fusion, enabling the capture of multifrequency signals. This module performs adaptive fusion tailored to the unique frequency information requirements of each node, mitigating the local heterophily problem. Extensive experiments on various real-world GAD datasets demonstrate that the proposed CIE-GAD outperforms state-of-the-art methods. Notably, our approach achieves AUC-PR improvements of up to 3.47%, with an average gain of 1.5%, demonstrating its effectiveness in detecting anomalies in graph data.

The Kriging model has been widely used in regression-based surrogate-assisted evolutionary algorithms (SAEAs) for expensive multiobjective optimization by using one model to approximate one objective, and the fusion of all the models forms the fitness surrogate. However, when tackling expensive many-objective optimization problems, too many models are required to construct such a fitness surrogate, which incurs cumulative prediction uncertainty and higher computational cost. Considering that the fitness surrogate works to predict different objective values to help select promising solutions with good convergence and diversity, this article proposes a novel model reduction idea to change the many-models-based fitness surrogate to a two-models-based indicator surrogate (TIS) that directly approximates convergence and diversity indicators. Based on TIS, a hierarchical and ensemble SAEA (HES-EA) is proposed with three stages. First, the HES-EA transforms the many objectives of the real-evaluated solutions into two indicators (i.e., the convergence and diversity indicators) and divides these solutions into different clusters. Second, an HES consisting of a cluster surrogate and different TISs is trained through these clustered solutions and their indicators. Third, during the optimization process, the HES can predict the candidate solutions’ cluster information via the cluster surrogate and indicator information via the TISs. Promising solutions can thus be selected based on the predicted information via a clustering-based sequential selection strategy without real fitness evaluation consumption. Compared with state-of-the-art SAEAs on three widely used benchmark suites up to 184 instances and one real-world application, HES-EA shows its superiority in both optimization performance and computational cost.

Constrained multiobjective optimization problems often have complex feasible regions and constrained Pareto fronts. These factors bring great challenges to current constrained multiobjective optimization evolutionary algorithms (CMOEAs). To solve this problem and further balance the objective optimization and constraint satisfaction, we propose an indexes-based and partial restart-based constrained multiobjective optimization (IRCMO) algorithm. In IRCMO, a two-stage (i.e., development and enhancement) and tri-population framework is designed. IRCMO adopts the aggregative indexes-based evaluation and adaptive collaborative partial restart strategy to assist the evolution of the first and second populations. The third population is obtained by directed sampling, which is mostly located at the boundary of the feasible region and enhances the exploration ability of extreme solutions. At the end of each generation, a progressive dual-archive strategy is designed to screen the solutions distributed uniformly from three populations. Experimental results demonstrate that IRCMO is superior to the other six state-of-the-art CMOEAs on several constraint benchmark suites and real-world problems.

3D keypoint detection is of great interest to researchers in computer vision and graphics because it is an integral part of realizing many tasks, such as object tracking, 3D reconstruction, and shape registration. However, it is challenging to detect 3D keypoints quickly and stably due to the ambiguity of the keypoints and the presence of noise, density changes, and geometric distortions in the 3D point cloud. This paper proposes a novel 3D keypoint detection method based on point cloud structural saliency (PCSS) to realize stable and efficient 3D keypoint detection. First, we propose an effective point cloud feature descriptor called local spatial geometric feature, which can effectively combine spatial and geometric information to improve feature distinguishability. Second, we define a point cloud structural saliency representation that effectively characterizes the structured information in the point cloud. Finally, we generate 3D keypoints based on point cloud structural saliency using a non-maximum suppression method. We evaluate our method on five 3D keypoint benchmark datasets, and the experimental results demonstrate that it achieves state-of-the-art performance in 3D keypoint detection. Comparing it with previous keypoint detection methods further demonstrates the effectiveness and superiority of our method.

Intelligent Transportation System (ITS) is a critical component of smart cities, however, certain issues significantly limit the construction of ITS. On the one hand, as the core resource of ITS, traffic data often suffers from missing values due to sensor failure, communication interruption, and so on. On the other hand, traffic flow change is a complex dynamic process that resulted from periodic changes caused by social activities, which increases the difficulty of data completion. To address these issues, the Message Passing Period-Aware Imputation Network (MPPAIN) is proposed. Firstly, the spatial-temporal information is transmitted sequentially in time order by the Message Passing Block based on gated recurrent unit, and the missing values are preliminarily estimated. Then, the output is fed to the Period-Aware Block to find the main frequency components that represent the changes of the traffic flow in the frequency domain through the Fourier transform. Subsequently, the traffic flow is divided according to the major periods, and the second time estimation is completed by extracting the intra-periodic and inter-periodic features simultaneously through the convolutional neural network. Finally, a bi-directional structure is designed by reversing the input traffic flow in time order to further extract the spatial-temporal and periodic features from the future to the past. Experiments demonstrate that the proposed model has excellent data imputation capabilities in various simulated missing rates and missing scenarios on several real traffic datasets.

Open-vocabulary multi-object tracking (OVMOT) is a cutting research direction within the multi-object tracking field. It employs large multi-modal models to effectively address the challenge of tracking unseen objects within dynamic visual scenes. While models require robust domain generalization and temporal adaptability, OVTrack, the only existing open-vocabulary multi-object tracker, relies solely on static appearance information and lacks these crucial adaptive capabilities. In this paper, we propose OVSORT, a new framework designed to improve domain generalization and temporal information processing. Specifically, we first propose the Adaptive Contextual Normalization (ACN) technique in OVSORT, which dynamically adjusts the feature maps based on the dataset's statistical properties, thereby fine-tuning our model's to improve domain generalization. Then, we introduce motion cues for the first time. Using our Joint Motion and Appearance Tracking (JMAT) strategy, we obtain a joint similarity measure and subsequently apply the Hungarian algorithm for data association. Finally, our Hierarchical Adaptive Feature Update (HAFU) strategy adaptively adjusts feature updates according to the current state of each trajectory, which greatly improves the utilization of temporal information. Extensive experiments on the TAO validation set and test set confirm the superiority of OVSORT, which significantly improves the handling of novel and base classes. It surpasses existing methods in terms of accuracy and generalization, setting a new state-of-the-art for OVMOT.

Continuous zeroing neural network (ZNN) and its discrete ZNN (DZNN) are comprehensively developed in many optimization systems. In this article, a Milne-Hamming method with DZNN classified as an implicit method is proposed and discussed upon the previous researches. Specifically, the Milne-Hamming discrete ZNN (MHDZNN) model is aimed for time-varying nonlinear optimization (TV-NO) problem with functional limitations. This Milne-Hamming (MH) method is a four-step discretized formula with fixed parameters and is introduced to discretize the ZNN model. Theoretical analyses of the MHDZNN model derive that MHDZNN possesses a larger stepsize domain $\mu \in (0,1/2)$ of absolute stability. Its convergent error is of order $O(\tau ^{5})$ and the corresponding truncation error constant is $1/40$ , which shows intimate relation to the accuracy. Compared with the existing DZNN models such as four-step explicit methods with the same $O(\tau ^{5})$ pattern, the convergent error constant of MHDZNN is smaller by a factor and maximal stability domain is greater. Finally, numerical simulations and application to redundant manipulators are provided and studied to verify the effectiveness of the proposed MHDZNN model.

In this article, we derive the predictor-corrector (PC) methods with three-order convergent precision, together with a class of specific general linear three-step (GLTS) rules provided. Afterward, a time-varying optimization (TVO) problem, which is deemed as a discrete TVO has been formulated and studied. The classical discrete zeroing neural network via Zhang et al. discretization (ZD-DZNN) is often utilized to obtain the solution. Actually, the stepsize domain of the DZNN model is a great factor for the dynamical stability. To enlarge the stepsize domain of the DZNN model, specific GLTS-type PC-DZNN models are applied to solve the TVO problem. Theoretical analyses show that better stability of the DZNN can be achieved by PC methods. Numerical simulative comparisons between the proposed PC-DZNN models and the ZD-DZNN in terms of stability are provided for further illustrations. In addition, motion planning of a PA10 manipulator and physical kinematics on UR5 formed as a TVO problem has been solved efficiently by applying the specific GLTS-type PC-DZNN models.

The canonical approach to video action recognition dictates a neural network model to do a classic and standard 1-of-N majority vote task. They are trained to predict a fixed set of predefined categories, limiting their transferability on new datasets with unseen concepts. In this article, we provide a new perspective on action recognition by attaching importance to the semantic information of label texts rather than simply mapping them into numbers. Specifically, we model this task as a video-text matching problem within a multimodal learning framework, which strengthens the video representation with more semantic language supervision and enables our model to do zero-shot action recognition without any further labeled data or parameters' requirements. Moreover, to handle the deficiency of label texts and make use of tremendous web data, we propose a new paradigm based on this multimodal learning framework for action recognition, which we dub "pre-train, adapt and fine-tune." This paradigm first learns powerful representations from pre-training on a large amount of web image-text or video-text data. Then, it makes the action recognition task to act more like pre-training problems via adaptation engineering. Finally, it is fine-tuned end-to-end on target datasets to obtain strong performance. We give an instantiation of the new paradigm, ActionCLIP, which not only has superior and flexible zero-shot/few-shot transfer ability but also reaches a top performance on general action recognition task, achieving 83.8% top-1 accuracy on Kinetics-400 with a ViT-B/16 as the backbone. Code is available at https://github.com/sallymmx/ActionCLIP.git.

In online learning, personalized course recommendations that align with learners’ preferences and future needs are essential. Thus, the development of efficient recommender systems is crucial to guide learners to appropriate courses. Graph learning in recommender systems has been extensively studied, yet many models focus on low-frequency information, underscoring similar learner preferences and overlooking high-frequency data that indicates varied learning trajectories. Furthermore, course co-occurrence and sequential relationships are often insufficiently investigated. In this paper, we introduceEduGraph, a novel framework developed specifically for MOOC course recommendation systems.EduGraphis characterized by its incorporation of a learning path-based hypergraph, a unique perspective wherein learners are represented as hyperedges, and courses are delineated as vertices. The framework incorporates a framelet-based hypergraph convolution, integrating low-pass filters to highlight similarities and high-pass filters to underscore distinct learning paths among learners. Furthermore,EduGraphfeatures a dual hypergraph learning model, with channels designated for vertex and hyperedge encoding, fostering a collaborative information exchange that refines the learners’ preference embeddings. The empirical assessment ofEduGraphis conducted through a comprehensive comparison with many existing baselines, utilizing two distinct MOOC datasets. Our experimental studies not only emphasize the enhanced recommendation performance ofEduGraphbut also elucidate the significant contributions of its individual components, such as the integration of low-pass and high-pass filters and the framelet-wise collaborative strategy that effectively bridges hyperedge-level and vertex-level representations, augmenting the overall efficacy of the course recommendation system.

Modern Siamese trackers mainly rely on classifying and regressing pre-defined anchor boxes or per-pixel points, which are assigned as positive and negative samples based on box intersection-over-union (IoU) or point distance with corresponding ground-truth for training. However, this rigid configuration potentially involves some noisy and ambiguous positive samples, leading to an inconsistency problem between classification and regression, which limits the tracking performance. In this paper, we propose a novel probabilistic assignment approach that dynamically determines positive/negative samples for each instance. To be specific, we first customize the confidence scores of positive candidates by comprehensively exploring the outputs from both classification and regression heads, and fit these scores as a probability distribution. Therefore, it is intuitive to conduct adaptive label assignment according to their probabilities. Then, we also consider dynamic re-weighting factor for each positive sample, jointly optimizing the classification and regression losses in a synchronized manner. Moreover, we introduce a decoupled IoU prediction branch to bridge the gap between the training and inference objectives for accurate tracking. Thanks to well-aligned procedures, our method significantly improves the performance of both CNN-based and Transformer-based trackers. Extensive experiments conducted on several tracking benchmarks including LaSOT and GOT-10k, demonstrate the effectiveness and efficiency of the proposed probabilistic assignment tracker.

Message passing (MP) is crucial for effective graph neural networks (GNNs). Most local message-passing schemes have been shown to underperform on heterophily graphs due to the perturbation of updated representations caused by local redundant heterophily information. However, our experiment findings indicate that the distribution of heterophily information during MP can be disrupted by disentangling local neighborhoods. This finding can be applied to other GNNs, improving their performance on heterophily graphs in a more flexible manner compared to most heterophily GNNs with complex designs. This article proposes a new type of simple message-passing neural network called Flow2GNN. It uses a two-way flow message-passing scheme to enhance the ability of GNNs by disentangling and redistributing heterophily information in the topology space and the attribute space. Our proposed message-passing scheme consists of two steps in topology space and attribute space. First, we introduce a new disentangled operator with binary elements that disentangle topology information in-flow and out-flow between connected nodes. Second, we use an adaptive aggregation model that adjusts the flow amount between homophily and heterophily attribute information. Furthermore, we rigorously prove that disentangling in message-passing can reduce the generalization gap, offering a deeper understanding of how our model enhances other GNNs. The extensive experiment results show that the proposed model, Flow2GNN, not only outperforms state-of-the-art GNNs, but also helps improve the performance of other commonly used GNNs on heterophily graphs, including GCN, GAT, GCNII, and H2GCN, specifically for GCN, with up to a 25.88% improvement on the Wisconsin dataset.

To amplify the achievements on Zeroing neural network (ZNN) and widen the application of fuzzy logic system (FLS), a complex fuzzy parameter zeroing neural network (CFPZNN) model is established to address the time-variant complex Sylvester equation problem. Varying from the fixed parameters in conventional ZNN (CZNN) or time-varying parameters in ZNN (TVP-ZNN), the fuzzy parameter generated by the FLS fluctuates according with convergent error and adjusts the convergent rate adaptively. Three different activated functions (AFs) equipped with the CFP-ZNN model are analyzed and discussed. Finite convergence characteristic of the CFP-ZNN model with Signbi-power (SBP) is testified. Furthermore, various membership functions (MFs) and various fuzzy control output values are studied and compared to exhibit the performance of the CFPZNN model. Theoretical analyses and comparable simulation results among different ZNN-based neural network models in dealing with time-variant complex Sylvester equations are welly coincided.

To assist redundant manipulator to complete complex repetitive trajectory in a presented time, this article provides a solution to the kinematic assignment and presents a predefined time fuzzy zeroing neural network with event-triggered mechanism (ETM-PTFZNN). The repetitive kinematics of the redundant manipulator is originally formulated as a time-varying quadratic programming (TVQP) problem, and the ETM-PTFZNN is engaged to solve the corresponding TVQP, where the fuzzy predefined time (PT) convergence is obtained by the fuzzy system and PT activation function simultaneously. Furthermore, event-triggered mechanism is introduced to update the fuzzy parameters of the ETM-PTFZNN orderly, which greatly alleviates the calculation burden of the ETM-PTFZNN. Theoretical analyses and simulative profiles reveal that the ETM-PTFZNN model can realize PT characteristic, robustness, adaptive stability, and repetitive trajectory for the TVQP problem of kinematic planning for manipulators.

Multiple autonomous underwater vehicles (AUVs) are popular for executing submarine missions, which involve multiple targets distributed in a large and complex underwater environment. The path planning of multiple AUVs is a significant and challenging problem, which determines the location of surface points for AUV launch and plans the paths of AUVs for target traveling. Most existing works model the problem as a single-objective static optimization problem. However, the target missions may change over time, and multiple optimization objectives are usually expected for decision making. Thus, this article models the problem as a dynamic multiobjective optimization problem and proposes a cooperative evolutionary computation algorithm to provide diverse and high-quality solutions for decision makers. In the proposed method, solutions are represented using a bilayer encode scheme, in which the first layer indicates the surface location points and the second layer represents the traveling sequences of target missions. Multiple populations for multiple objectives framework is adopted to efficiently solve the dynamic multiobjective AUV optimization problem. In addition, a recombination-based sampling strategy is developed to improve convergence by fusing the information of multiple populations. Once a change occurs, an incremental response strategy is adopted to generate high-quality solutions for population evolution. Based on the dataset of New Zealand bathymetry, six complex underwater scenarios are constructed with a size of 50 km × 50 km× 10 km and 400 target missions for tests. Experimental results show that the proposed method outperforms the state-of-the-art algorithms in terms of solution diversity and optimality.

The perception layer of Internet of Things (IoT) not only needs to perceive service requests rapidly, but also considers reducing energy consumption intelligently. This issue becomes crucial in the scenario of large-scale IoT services. The existing research usually focus on one single aspect only, either energy consumption or perception rate. Inspired from the human visual direction-sensitive system and convolutional neural network, we propose an energy-efficient convolution-based partitioned collaborative perception algorithm (CPCPA) for large-scale IoT services. The perception range of each node is divided into multiple regions. First, by introducing the direction sensitive mechanism, the preferred search orientation can be determined quickly and become directional. Then, a selection operator of the partitioned region is designed to keep the search region updating and prevent CPCPA from getting stuck in local optimums. Meanwhile, a convolutional method is used to filter out unhelpful nodes to adapt the self-adaptive wake-up probability, which precisely controls the state switch of the nodes to reduce energy consumption. Finally, simulation results verify that CPCPA enables IoT to discover large-scale random service requests. The results also indicate that the proposed algorithm achieves better energy maintenance and maintains a higher perception rate than the state-of-the-art algorithms including directional sensitivity-based perception algorithm, sensor node activation method using bat algorithm, coverage aware scheduling for optimal placement, intelligent self-organizing scheme. An overall average perception rate improvement of 2.46% is achieved by CPCPA than the compared algorithms.

MOTIVATION: Protein-protein interaction (PPI) networks and transcriptional regulatory networks are critical in regulating cells and their signaling. A thorough understanding of PPIs can provide more insights into cellular physiology at normal and disease states. Although numerous methods have been proposed to predict PPIs, it is still challenging for interaction prediction between unknown proteins. In this study, a novel neural network named AFTGAN was constructed to predict multi-type PPIs. Regarding feature input, ESM-1b embedding containing much biological information for proteins was added as a protein sequence feature besides amino acid co-occurrence similarity and one-hot coding. An ensemble network was also constructed based on a transformer encoder containing an AFT module (performing the weight operation on vital protein sequence feature information) and graph attention network (extracting the relational features of protein pairs) for the part of the network framework. RESULTS: The experimental results showed that the Micro-F1 of the AFTGAN based on three partitioning schemes (BFS, DFS and the random mode) on the SHS27K and SHS148K datasets was 0.685, 0.711 and 0.867, as well as 0.745, 0.819 and 0.920, respectively, all higher than that of other popular methods. In addition, the experimental comparisons confirmed the performance superiority of the proposed model for predicting PPIs of unknown proteins on the STRING dataset. AVAILABILITY AND IMPLEMENTATION: The source code is publicly available at https://github.com/1075793472/AFTGAN. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Abstract As a special case of common object removal, image person removal is playing an increasingly important role in social media and criminal investigation domains. Due to the integrity of person area and the complexity of human posture, person removal has its own dilemmas. In this paper, a novel idea is proposed to tackle these problems from the perspective of data synthesis. Concerning the lack of a dedicated dataset for image person removal, two dataset production methods are proposed to automatically generate images, masks and ground truths, respectively. Then, a learning framework similar to local image degradation is proposed so that the masks can be used to guide the feature extraction process and more texture information can be gathered for final prediction. A coarse‐to‐fine training strategy is further applied to refine the details. The data synthesis and learning framework combine well with each other. Experimental results verify the effectiveness of the method quantitatively and qualitatively, and the trained network proves to have good generalization ability either on real or synthetic images.

The perception layer of the Internet of Things (IoT) needs to respond to service requests rapidly. In the existing IoT perception methods, it remains challenging to concurrently ensure high energy maintenance and perception rate. To address this issue, inspired from the human visual direction-sensitive system, this article proposes an intelligent directional sensitivity-based perception algorithm (DSPA) for IoT service. First, the perception range of each node is divided into multiple regions. The perception direction of the node is represented by an arrow. The perception orientation of the node indicates a region to which the direction belongs. Then, by imitating the human visual direction-sensitive mechanism, direction optimization is proposed for IoT service perception. Meanwhile, region weight is designed to assist in determining the optimal direction to prevent DSPA from getting stuck in local optimums. Finally, simulation results demonstrate that DSPA achieves better energy maintenance and obtains a high perception rate faster than the compared algorithms.

This article proposes a CNN-based born-again Takagi–Sugeno–Kang (TSK) fuzzy classifier denoted as CNNBaTSK. CNNBaTSK achieves the following distinctive characteristics: 1) CNNBaTSK provides a new perspective of knowledge distillation with a noniterative learning method (least learning machine with knowledge distillation, LLM-KD) to solve the consequent parameters of fuzzy rule, where consequent parameters are trained jointly on the ground-truth label loss, knowledge distillation loss, and regularization term; 2) with the inherent advantage of the fuzzy rule, CNNBaTSK has the capability to express the dark knowledge acquired from the CNN in an interpretable manner. Specifically, the dark knowledge (soft label information) is partitioned into five fixed antecedent fuzzy spaces. The centers of each soft label information in different fuzzy rules are {0, 0.25, 0.5, 0.75, 1}, which may have corresponding linguistic explanations: {very low, low, medium, high, very high}. For the consequent part of the fuzzy rule, the original features are employed to train the consequent parameters that ensure the direct interpretability in the original feature space. The experimental results on the benchmark datasets and the CHB-MIT EEG dataset demonstrate that CNNBaTSK can simultaneously improve the classification performance and model interpretability.

Zeroing neural network (ZNN) with different activation functions (AFs) for finding zero-result of time-varying quadratic programming (TVQP) with no noises are revisited. To improve the convergent speed of the ZNN and resist various noises occurred in the real application, two robust terminal recurrent neural network (RTRNN) models by adding two different AFs are presented for the exact solution of the TVQP problem facing various noises. The appearing advantage of the prespecified time of the RTRNN model is independent of the initial status of a generated system and the convergent time can be accelerated in advance, which is much superior than the finite-time performance with regard to the initial status. In addition, the prespecified convergent time of the RTRNN is mathematically discussed in detail under external noises. Simulated comparisons between the proposed RTRNN and the state-of-the-art neural networks substantiate the predefined time performance and strong robustness.

The adaptive neurofuzzy inference system (ANFIS) is a structured multioutput learning machine that has been successfully adopted in learning problems without noise or outliers. However, it does not work well for learning problems with noise or outliers. High-accuracy real-time forecasting of traffic flow is extremely difficult due to the effect of noise or outliers from complex traffic conditions. In this study, a novel probabilistic learning system, probabilistic regularized extreme learning machine combined with ANFIS (probabilistic R-ELANFIS), is proposed to capture the correlations among traffic flow data and, thereby, improve the accuracy of traffic flow forecasting. The new learning system adopts a fantastic objective function that minimizes both the mean and the variance of the model bias. The results from an experiment based on real-world traffic flow data showed that, compared with some kernel-based approaches, neural network approaches, and conventional ANFIS learning systems, the proposed probabilistic R-ELANFIS achieves competitive performance in terms of forecasting ability and generalizability.

Aiming at low efficiency of agricultural machinery scheduling, this paper proposes an adaptive immune-following algorithm (AIFA) based on immune algorithm and artificial fish swarm algorithm. The adaptive crossover operator is used to accelerate convergence, and adaptive mutation operator ensures good diversity of population. After the adaptive evolution operations are performed, the following operator based on the following behavior of artificial fish swarm algorithm is embedded into the algorithm, which improves the convergence precision and obtains the promising optimization results. Experiments on scheduling considering the breakdown of agricultural machinery are performed based on multiple regions and multiple agricultural machineries. Compared with the immune algorithm and genetic algorithm, the simulation results demonstrate that AIFA can converge faster and achieve a better optimal solution.

Accurate estimation of short-term traffic flow, which can help to assist travelers make better route choices, is a significant research field of intelligent transportation system. In order to extract complex spatiotemporal features from a small amount of available traffic data, in this paper we propose a novel Deep Graph Gaussian Processes (DGGPs) for short-term traffic flow prediction. First, in order to accurately describe the relationship between vertices in time series, this paper proposes an attention kernel. Based on this, the Aggregation Gaussian Process uses attention kernel as the covariance function, which overcomes the problem that the existing Gaussian processes and the deep Gaussian processes cannot effectively obtain dynamic spatial features. Second, DGGPs are constructed by the Aggregation Gaussian Process (AGP), the Temporal Convolutional Gaussian Process (TCGP) and the Gaussian process with linear kernel, to solve the existing short-term traffic flow forecasting models cannot obtain complex spatiotemporal features from a small amount of available data. We verify that the attention kernel helps to the proposed model convergence on the three data sets. At the same time, the proposed DGGP can obtain spatiotemporal features from the situation with less available spatial information or temporal information, accurately predict short-term traffic flow, and quantify temporal uncertainty.

Small object detection remains an unsolved challenge because it is hard to extract the information of small objects with only a few pixels. While scale-level corresponding detection in feature pyramid network alleviates this problem, we find feature coupling of various scales still impairs the performance of small objects. In this paper, we propose an extended feature pyramid network (EFPN) with an extra high-resolution pyramid level specialized for small object detection. Specifically, we design a novel module, named feature texture transfer (FTT), which is used to super-resolve features and extract credible regional details simultaneously. Moreover, we introduce a cross resolution distillation mechanism to transfer the ability of perceiving details across the scales of the network, where a foreground-background-balanced loss function is designed to alleviate area imbalance of foreground and background. In our experiments, the proposed EFPN is efficient on both computation and memory, and yields state-of-the-art results on small traffic-sign dataset Tsinghua-Tencent 100 K and small category of general object detection dataset MS COCO.

Time-varying matrix reciprocal problems are widely appeared in different matrix computations and engineering fields. Neural networks as a powerful tool have been developed to solve the time-varying problems. Recurrent neural networks (RNNs) are designed considering mainly for two aspects: 1) convergent precision and 2) convergent time. The core part of the existed neural methods is to design various kinds of activation function for time-varying matrix solving. However, most of the activation functions of neural networks are with infinite value, which demands long convergent time and are not applicable in practical engineering fields. This note proposes theoretical analyses and simulation results on the performance of terminal RNN (TRNN) and accelerated TRNN (ATRNN) with finite-time convergence, which is not only designed for constant matrix inversions but also for time-varying reciprocal matrix. Compared to the traditional RNNs, TRNNs are of limit-valued activation function and possess a finite time convergence property. The simulation results for time-varying reciprocal solving validate the perfect performance solved by TRNN and ATRNN. In addition, a quadratic program (QP) of velocity minimization based on TRNN is proposed to solve the trajectory tracking problems without considering the initial position error of the redundant manipulators. Finally, practical experiments of the redundant manipulators based on PUMA560 show the effectiveness and accuracy of the proposed approaches.

To obtain the solution for time-varying quadratic programming (QP), a time controlling neural network (TCNN) is presented and discussed. The traditional recurrent neural networks provide a prospect for real-time calculations and repeatable trajectory control of the mobile manipulators due to its high executing processing and nonlinear disposal ability. However, the convergent time is still a considerable point for the solution of a dynamic system dealing with synchronism and robustness. In this note, a TCNN model by incorporating an initial rectified term is applied to solve the online calculation problems and the convergent time can be controlled in advance. Theoretical analyses on stability, prespecified time and convergence are rigorously clarified. Finally, effectiveness and precision of the TCNN model for the solution of a QP example have been verified. In addition, a repetitive trajectory planning for a three-wheel manipulator is introduced to demonstrate the superiority of the TCNN.

Blind hyperspectral unmixing (BHSU) is ill-posedness. It aims to obtain accurate and robust endmember signatures and the corresponding abundances simultaneously. Nonnegative matrix factorization (NMF)-based sparsity-regularized algorithms have been widely employed for the BHSU. However, the existing unmixing approaches are sensitive to the multifarious intrinsic interferences and noises, which are caused because of the utilization of the inappropriate loss function to measure the quality of the hyperspectral data (HD) reconstruction and regularization. In this article, we propose a noise-free graph regularized model (NFGRM) by applying the dual graph regularized robust nonnegative matrix tri-factorization (NMTF), which leads to a novel reliable reconstruction of the HD. In the NFGRM, all the challenging interferences are addressed as noises. Consequently, a more faithful approximation is expected to recover from the highly noisy mixed data set and achieve robust regularization by controlling the heteroscedastic noises and the ill-posedness of the BHSU problem simultaneously. Experimental results on synthetic and several benchmark HD sets demonstrate the effectiveness and robustness of the proposed model and algorithm.

Traditional spectral unmixing methods are usually based on the linear mixture model (LMM) or nonlinear mixture model (NLMM), in which only the additive noise is considered. However, in hyperspectral applications, the additive, multiplicative, and mixed noises play important roles. In this paper, we propose an antinoise model for hyperspectral unmixing. In the antinoise model, all the additive, multiplicative and mixed noises are addressed. To deal with the problems faced by LMM or NLMM and to tackle the antinoise model, an antinoise model based hyperspectral unmixing method is presented, where block coordinate descent is employed to solve an approximatedL0norm constraint, then a nonnegative matrix factorization (NMF) method is presented, which is based on the bounded Itakura-Saito divergence. The experimental results on both synthetic and real hyperspectral data sets demonstrate the efficacy of the proposed model and the corresponding method.

SUMMARY The mobile agent technique has been broadly used in next generation distributed systems. The system performance measurement and simulation are required before the system can be deployed on a large scale. In this paper, we address performance analysis on a finite state mobile agent prototype on the basis ofVirtual Hierarchical TreeGridOrganizations (VIRGO). The finite states refer to the migration, execution, and searching of the mobile agent. We introduce a novel evaluation model for the finite state mobile agent. The experimental results based on this evaluation model show that the finite mobile agents can perform well under multiple agent conditions and are superior to the traditional client/server approach. Copyright © 2013 John Wiley & Sons, Ltd.

The prevalence of Web 2.0 techniques has led to the boom of various online communities, where topics spread ubiquitously among user-generated documents. Working together with this diffusion process is the evolution of topic content, where novel contents are introduced by documents which adopt the topic. Unlike explicit user behavior (e.g., buying a DVD), both the diffusion paths and the evolutionary process of a topic are implicit, making their discovery challenging. In this paper, we track the evolution of an arbitrary topic and reveal the latent diffusion paths of that topic in a social community. A novel and principled probabilistic model is proposed which casts our task as an joint inference problem, which considers textual documents, social influences, and topic evolution in a unified way. Specifically, a mixture model is introduced to model the generation of text according to the diffusion and the evolution of the topic, while the whole diffusion process is regularized with user-level social influences through a Gaussian Markov Random Field. Experiments on both synthetic data and real world data show that the discovery of topic diffusion and evolution benefits from this joint inference, and the probabilistic model we propose performs significantly better than existing methods.

We propose granular computing (GrC) on ontology as a solution to the problem of modeling complex architectures. We expressed the architectures formally as ontology domains, which include two components: the set of basic vocabularies and a knowledge library of rules. The set of basic vocabularies contains elements or basic architecture components. The knowledge library comprises rules that control the combination and construction of the basic elements. As the rules are often given by architectural experts subjectively, they may contain redundant, conflicting, and overlapping rules, especially in certain styles of ancient southeast Chinese architecture. It is difficult to distinguish or identify these rules; therefore, we apply the multilevel approach on ontology [Y. Liu, C. Xu, Q. Zhang, and Y. Pan, ¿Smart architect: Scalable ontology-based modeling for ancient chinese architecture,¿ IEEE Intell. Syst., vol. 23, no. 1, pp. 49-56, Jan./Feb. 2008] and approximation theory of GrC. In this process, we present a measurement that is based on roughness functions to evaluate the degrees of approximation between the selected set and certain architecture domains. With the monotonicity characteristic of roughness functions, we can design a heuristic algorithm to select a suitable knowledge base (rule set) to assist in integrating the parts into final architectures, via several levels. Experiments with a real architectural project, i.e., modeling ancient southeast Chinese architectures, show that our method is effective and may simplify the design of the automodeling system and enhance its performance.

While keyword queries have become the "standard" query language of web search and many other database applications, their brevity and unstructuredness make it difficult to detect what users really want. In this demonstration, we aim to detect such hidden query intents, which we define as the frequent phrases that users co-ask with the query term, by exploring query logs. Toward building an online search system AIDE, we offer users the function to detect general and unique intents using arbitrary ad-hoc queries at run time. We will also demonstrate the effectiveness of the system which achieves indexing and searching over 14M MSN query log records.

The attribute reduction of information system can enhance accuracy and efficiency of knowledge discovery, machine learning, etc. After studying reduction strategy in rough set theory, the concept of approximate reduction and an approximate reduction algorithm are proposed in this paper. This algorithm can retain minimal attributes in the basic style of information system, i.e. reduce as many attributes as possible. That can save much time for the system's later disposal. The algorithm's time complexity hasnpsilat been improved, but attributes after reduction are reduced greatly. The original information system has a certain loss, but this can be accepted under a certain significance level. Lastly, the reduction strategy is compared with approximate reduction strategy by nine attributes which belong to the information system.

Although the data fusion and data mining belong to the data processing technology, former researchers combine these technologies quite a few. In fact these technologies are nearly correlative, and both serve the goal of the knowledge detects. The target of data fusion is forming the data foundation of data mining, and the target of data mining is extracting the useful knowledge in the foundation of above data, to complete the knowledge detection. Therefore, our research's aim is combining the two technologies, designs the valid algorithm to carry on the data mining in the foundation of data fusion, mainly on the artificial intelligence theory and the Bayes method, excavates the valid information in the data of environmental monitoring as far as possible.