In Internet of Things (IoT), artificial intelligence (AI) agents have become a core paradigm for edge intelligence, supporting on-device sensor data processing and privacy-preserving analytics. In practice, however, the sensor data streams evolve continuously and often exhibit nonindependent and identically distributed (non-IID) characteristics. Federated class-incremental learning (FCIL) has thus emerged as a promising approach to endow edge AI agents with long-tail learning capabilities, allowing models to adapt autonomously to persistent changes in the data. Despite recent progress, most existing FCIL methods assume balanced data distributions, which rarely hold in real-world deployments. Under realistic long-tailed distributions, the local optimization objectives of individual AI agents diverge from the desired global convergence, leading to head-class bias and rapid forgetting of tail classes. To overcome these limitations, we propose gradient-aligned and imbalance-aware FCIL model (FedGRAIL), a unified FCIL framework that resolves the conflict between local optimization bias and global consistency in dynamic IoT environments. FedGRAIL integrates local training and global aggregation within a structurally aligned learning process, rather than treating them as separate stages. It further introduces a gradient-correction mechanism that coordinates dispersed local optima and stabilizes global optimization via diversity-driven generative replay. This design supports effective retention of minority concepts while respecting the zero-storage constraints of edge agents. Extensive experiments on long-tailed benchmarks show that FedGRAIL substantially improves tail-class accuracy under both independent and identically distributed (IID) and non-IID conditions - achieving a 59% average gain over the strongest baseline, while preserving the competitive accuracy on head classes. These results demonstrate that FedGRAIL provides a storage-efficient, privacy-preserving, and robust solution for continual learning in IoT edge intelligence.

Visual object tracking plays a critical role in various applications, such as autonomous driving, video surveillance, and human-computer interaction. However, it remains a challenging task due to the complexities of target motion, appearance variations, and occlusions in video sequences. To address these challenges, this paper proposes TIFTrack, a single-object tracker based on spatiotemporal information. We explore the target’s motion trajectory in video sequences and the visual features between adjacent frames, introducing an inter-frame information extraction module to adaptively capture the required temporal information during tracking. This includes changes in the target’s trajectory, appearance, and size. Furthermore, to fully leverage temporal information, we propose a novel trajectory token that maps the target’s motion within a specific frame range. Additionally, we design a Temporal Decoder to process the captured temporal information and effectively combine static visual features with dynamic temporal features, enhancing tracking robustness. Experimental results on the GOT-10k, TrackingNet, LaSOT, and LaSOTextdatasets demonstrate that TIFTrack outperforms existing trackers across multiple benchmarks.

Due to the rare occurrence of abnormal targets in substations, detecting these anomalies can be framed as a few-shot learning problem, for which recent few-shot object detection (FSOD) methods offer promising solutions. However, the complexity of substation scenarios makes current FSOD approaches struggle to control class centers, resulting in trained models typically exhibiting bias towards base classes and tending to confuse novel classes with base classes. To address these issues, we propose a feature aggregation method based on Variational Autoencoder (VAE) that transforms instance-level features into class distributions, thereby enhancing the robustness of feature aggregation. Additionally, to improve the quality of region proposal generation in sample-limited and complex environments, we integrate a multi-head attention mechanism into the Region Proposal Network (RPN), enabling it to produce more relevant region proposals. Extensive experiments on the Substation Abnormal Target Dataset (SATD2024) demonstrate that our approach consistently outperforms existing methods across various settings. Furthermore, we validate the effectiveness and generalization of our method on the public Pascal VOC dataset.

The integration of Intelligent Welding Systems (IWS) in smart manufacturing leverages advancements in sensors, robotics, and artificial intelligence to optimize welding processes. However, in industry practice, we still face challenges such as sufficient data is not available for every manufacturing task, the costs associated with welding data annotation quality, and the risk of knowledge forgetting during the continual welding process. To tackle these issues, we developed an Adaptive Domain-Enhanced Transfer Learning (ADETL) framework that integrates self-supervised and continual learning strategies. This framework is adept at using incremental and unlabeled data for pre-training, in which we analyze the parameter space, loss landscape, and make the model understand the behaviour of knowledge transfer from diverse source domains. The ADETL framework improves the performance of defect classification, offering a promising solution to the challenges inherent in automatic, continuous welding operations.

Abstract The negative herding effect tends to cause the spread of negative emotions among netizens, resulting in significant public opinion crises. Therefore, linking the negative herding effect and online public opinion, we aim to exploring the connection between netizens' emotions, opinions, and the negative herding effect. To better distinguish the emotions in texts, a new BiGADG ( Bi directional G ated recurrent unit based on A ttention with D istance‐based dependency syntax G raph convolutional network) model is proposed in this article. This model includes primarily the proposed dependency syntax based on dependency distance, Bidirectional Gated Recurrent Unit (BiGRU), multi‐head self‐attention and Graph Convolutional Network (GCN). Particularly, the dependency syntax based on dependency distances incorporates word‐to‐word dependency distances into the traditional dependency syntax, which enriches the syntactic structural features of the text. After the experiments, the accuracy of BiGADG model is 86.53%, which is 3.11% higher compared to BERT model. It is evident that our model for sentiment classification has been improved. Based on the results of sentiment classification, we investigate the existence of negative herding behaviour and its underlying motivation in online public opinion. We have found that the negative herding effect, if not effectively managed, may trigger large‐scale and uncontrollable public opinion crises.

Cluster ensemble improves the robustness and stability of clustering performances by utilizing multiple solutions. Although traditional cluster ensemble methods have achieved promising performances, they are not adaptive enough to cope with data sets that have multiple levels of complexities. Besides, these methods may contain noisy and redundancy members which have negative effects. To mitigate the above issues, in this paper, we propose a multi-objective filter refinement scheme (MOFRS). First, we perform various clustering methods on different representations of data to generate diverse solutions. Second, we propose a solution filter to select a proper method and reduce the number of initial partitions for a given data set. Third, four stability indices are designed to split instances into stable and unstable groups. Fourth, objective functions based on diversity and quality are utilized to quantify the goodness of base clustering solutions. Finally, we design an improvement oriented multi-objective evolutionary algorithm to optimize these objective functions. Extensive experimental results conducted on 27 real-world data sets show that MOFRS outperforms most cluster ensemble selection methods, and achieves statistically significant improvements, compared with full ensemble methods.

Despite the tremendous success in computer vision, deep convolutional networks suffer from serious computation costs and redundancies. Although previous works address that by enhancing the diversities of filters, they have not considered the complementarity and the completeness of the internal convolutional structure. To respond to this problem, we propose a novel inner-imaging (InI) architecture, which allows relationships between channels to meet the above requirement. Specifically, we organize the channel signal points in groups using convolutional kernels to model both the intragroup and intergroup relationships simultaneously. A convolutional filter is a powerful tool for modeling spatial relations and organizing grouped signals, so the proposed methods map the channel signals onto a pseudoimage, like putting a lens into the internal convolution structure. Consequently, not only is the diversity of channels increased but also the complementarity and completeness can be explicitly enhanced. The proposed architecture is lightweight and easy to be implement. It provides an efficient self-organization strategy for convolutional networks to improve their performance. Extensive experiments are conducted on multiple benchmark datasets, including CIFAR, SVHN, and ImageNet. Experimental results verify the effectiveness of the InI mechanism with the most popular convolutional networks as the backbones.

Zero-shot learning uses semantic attributes to connect the search space of unseen objects. In recent years, although the deep convolutional network brings powerful visual modeling capabilities to the ZSL task, its visual features have severe pattern inertia and lack of representation of semantic relationships, which leads to severe bias and ambiguity. In response to this, we propose the Graph-based Visual-Semantic Entanglement Network to conduct graph modeling of visual features, which is mapped to semantic attributes by using a knowledge graph, it contains several novel designs: 1. it establishes a multi-path entangled network with the convolutional neural network (CNN) and the graph convolutional network (GCN), which input the visual features from CNN to GCN to model the implicit semantic relations, then GCN feedback the graph modeled information to CNN features; 2. it uses attribute word vectors as the target for the graph semantic modeling of GCN, which forms a self-consistent regression for graph modeling and supervise GCN to learn more personalized attribute relations; 3. it fuses and supplements the hierarchical visual-semantic features refined by graph modeling into visual embedding. Our method outperforms state-of-the-art approaches on multiple representative ZSL datasets: AwA2, CUB, and SUN by promoting the semantic linkage modelling of visual features.

Healthcare question answering (HQA) system plays a vital role in encouraging patients to inquire for professional consultation. However, there are some challenging factors in learning and representing the question corpus of HQA datasets, such as high dimensionality, sparseness, noise, nonprofessional expression, etc. To address these issues, we propose an inception convolutional autoencoder model for Chinese healthcare question clustering (ICAHC). First, we select a set of kernels with different sizes using convolutional autoencoder networks to explore both the diversity and quality in the clustering ensemble. Thus, these kernels encourage to capture diverse representations. Second, we design four ensemble operators to merge representations based on whether they are independent, and input them into the encoder using different skip connections. Third, it maps features from the encoder into a lower-dimensional space, followed by clustering. We conduct comparative experiments against other clustering algorithms on a Chinese healthcare dataset. Experimental results show the effectiveness of ICAHC in discovering better clustering solutions. The results can be used in the prediction of patients' conditions and the development of an automatic HQA system.

The tongue image provides important physical information of humans. It is of great importance for diagnoses and treatments in clinical medicine. Herbal prescriptions are simple, noninvasive, and have low side effects. Thus, they are widely applied in China. Studies on the automatic construction technology of herbal prescriptions based on tongue images have great significance for deep learning to explore the relevance of tongue images for herbal prescriptions, it can be applied to healthcare services in mobile medical systems. In order to adapt to the tongue image in a variety of photographic environments and construct herbal prescriptions, a neural network framework for prescription construction is designed. It includes single/double convolution channels and fully connected layers. Furthermore, it proposes the auxiliary therapy topic loss mechanism to model the therapy of Chinese doctors and alleviate the interference of sparse output labels on the diversity of results. The experiment use the real-world tongue images and the corresponding prescriptions and the results can generate prescriptions that are close to the real samples, which verifies the feasibility of the proposed method for the automatic construction of herbal prescriptions from tongue images. Also, it provides a reference for automatic herbal prescription construction from more physical information.

Classification of high-dimensional data with very limited labels is a challenging task in the field of data mining and machine learning. In this paper, we propose the multiobjective semisupervised classifier ensemble (MOSSCE) approach to address this challenge. Specifically, a multiobjective subspace selection process (MOSSP) in MOSSCE is first designed to generate the optimal combination of feature subspaces. Three objective functions are then proposed for MOSSP, which include the relevance of features, the redundancy between features, and the data reconstruction error. Then, MOSSCE generates an auxiliary training set based on the sample confidence to improve the performance of the classifier ensemble. Finally, the training set, combined with the auxiliary training set, is used to select the optimal combination of basic classifiers in the ensemble, train the classifier ensemble, and generate the final result. In addition, diversity analysis of the ensemble learning process is applied, and a set of nonparametric statistical tests is adopted for the comparison of semisupervised classification approaches on multiple datasets. The experiments on 12 gene expression datasets and two large image datasets show that MOSSCE has a better performance than other state-of-the-art semisupervised classifiers on high-dimensional data.

In this paper, we present an experimental evaluation of the Cartesian stiffness of a novel parallel wrist under redundant and antagonistic actuation. The mechanism in consideration is Omni-Wrist V (OW5), a two degrees-of-freedom (DoF) parallel mechanism redundantly actuated by three subchains. We first give a brief review of its kineto-statics and derive its reduced Cartesian stiffness model. To illustrate the stiffness enhancement of OW5 under redundant and antagonistic actuation, its Cartesian stiffness is measured and evaluated under four control schemes: the non-redundant control, the minimum 2-norm torque control without or with redundant encoder, and the antagonistic actuation control. Measurement data are represented using stiffness matrices and stiffness ellipses. Our study offers a quick quantitative evaluation of stiffness enhancement of OW5 under redundant and antagonistic actuation.

In many applications, such as online dating or job hunting websites, users often need to search for potential matches based on the requirements or preferences imposed by both sides.We refer to this type of queries as matching queries. In spite of their wide applicabilities, there has been little attention devoted to improve their performance. As matching queries often appear in various forms even within a single application, we, in this paper, propose a general processing framework, which can efficiently process various forms of matching queries. Moreover, we elaborate the detailed processing algorithms for two particular forms of matching queries to illustrate the applicability of this framework. We conduct an extensive experimental study with both synthetic and real datasets. The results indicate that, for various matching queries, our techniques can dramatically improve the query performance, especially when the dimensionality is high.

In this paper, dynamic feedback compensator design is proposed for the purpose of anti-windup. The design structure is based on coprime factorization of the pre-defined linear controller in the general unified anti-windup framework. Stability and performance objectives by considering the mismatch between the plant input and the controller output are represented in the form of linear matrix inequalities (LMIs). Sub-optimal LMI-based dynamic feedback compensator is got and the simulation example shows its effectiveness and demonstrates graceful performance degradation in comparison with other existing methods.