Knowledge-Based Systems最新文献

{"title":"Particle swarm optimization algorithm based on teaming behavior","authors":"Yu-Feng Yu ,&nbsp;Ziwei Wang ,&nbsp;Xinjia Chen ,&nbsp;Qiying Feng","doi":"10.1016/j.knosys.2025.113555","DOIUrl":"10.1016/j.knosys.2025.113555","url":null,"abstract":"<div><div>The traditional particle swarm optimization algorithms have some shortcomings, such as low convergence precision, slow convergence speed, and susceptibility to falling into local optima when solving complex optimization problems. To address these issues, this paper proposes a new particle swarm optimization algorithm that incorporates teamwork. Specifically, we introduce the concept of teamwork, and divide the particles into multiple teams and selecting team leaders. The particles can fully utilize the team’s prompt information to guide the search process. The team leader updates the search direction of its particles through the generation of information factors, thus giving the algorithm better global search capabilities. The position and behavior of the team leader affect the search behavior of other particles, reducing the risk of falling into local optimal solutions. In addition, to further improve the algorithm’s efficiency, we propose adaptive adjustment of information factors and learning factors. This adaptive adjustment mechanism enables the algorithm to adjust parameters flexibly according to the characteristics of the problem and the current search state, thereby accelerating convergence speed and improving convergence precision. To verify the performance of the proposed algorithm, we make an empirical analysis on 27 different test functions, the shortest path problem and the optimal SINR value problem for UAV deployment. The experimental results show that the proposed algorithm has obvious advantages in convergence accuracy and convergence speed. Compared with other algorithms, this algorithm can find a better solution faster and converge to the global optimal solution more stably.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113555"},"PeriodicalIF":7.2,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143870739","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DSNet: Predicting drug-side effect frequencies via Dual-Graph Ensemble and Similarity Learning","authors":"Qiuyu Long ,&nbsp;Nan Zhao ,&nbsp;Haifeng Liu","doi":"10.1016/j.knosys.2025.113537","DOIUrl":"10.1016/j.knosys.2025.113537","url":null,"abstract":"<div><div>Drug safety remains a critical concern in healthcare, making the accurate prediction of drug-side effect frequencies essential for risk-benefit assessments. Recent advancements in graph neural network-based methods for predicting drug-side effect frequencies have shown significant promise. However, the inherent complexity of drug molecular structures, often characterized by multi-ring and long-chain substructures, poses a challenge. Mainstream graph-based approaches are limited in expressive power and suffer from low information transmission efficiency, which hampers the ability to capture deep structural features. Additionally, the sparsity of drug-side effect interaction networks restricts the effective utilization of similarity information between drugs and side effects, substantially degrading prediction quality.</div><div>To address these challenges, we propose a novel framework for predicting drug-side effect frequencies, termed DSNet. By integrating multi-source heterogeneous features to construct embedding representations, and designing a Dual-Graph Ensemble Network with residual connections, DSNet enhances the capture of local, subtle features of drug molecules while preserving global structural consistency. To mitigate the sparsity limitations of drug-side effect interaction networks, we introduce a Structural Consistency Preservation Loss, which ensures that critical information is retained in the low-dimensional space. Additionally, we propose a Temperature-Adaptive Similarity Loss to dynamically adjust the sharpness of the similarity distribution between drugs and side effects. Experimental results on the SIDER dataset demonstrate that DSNet significantly improves prediction performance in both warm-start and cold-start scenarios. Furthermore, molecular docking experiments targeting tigecycline further validate the effectiveness of DSNet in predicting drug-side effect frequencies.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113537"},"PeriodicalIF":7.2,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143877154","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Source-Free Unsupervised Domain Adaptation through Trust-Guided Partitioning and Worst-Case Aligning","authors":"Qing Tian ,&nbsp;Lulu Kang","doi":"10.1016/j.knosys.2025.113493","DOIUrl":"10.1016/j.knosys.2025.113493","url":null,"abstract":"<div><div>In source-free unsupervised domain adaptation (SFUDA) tasks, adapting to the target domain without directly accessing the source domain data and relying solely on a pre-trained source domain model and the target domain data is a common challenge. Existing approaches often rely on pseudo-labeling techniques for intra-class clustering to achieve global alignment of classes. However, the presence of noise can lead to incorrect clustering results. In this paper, we introduce a novel approach referred to as Trust-guided Partitioning and Worst-case Aligning (TPWA). We assess the reliability of pseudo-labels by computing the similarity difference between the class centers corresponding to the pseudo-labels and the centers of the most similar classes. Based on this, we perform partitioning and then conduct intra-class clustering only on high-trustworthy samples. We also train a worst-case classifier to predict correctly on high-trustworthy samples and make as many mistakes as possible on low-trustworthy samples, and then adversarially trains feature extractors to align low-trustworthy samples to high-trustworthy samples. This approach leverages the structural information present in the high-trustworthy sample set, improving the robustness of the adaptation. Additionally, we also consider enforcing prediction consistency among neighboring samples to further constrain the pseudo-labels. Extensive experiments demonstrate the superiority of our method in SFUDA tasks.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113493"},"PeriodicalIF":7.2,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143865177","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimal distributed subsampling for expected shortfall regression via Neyman-orthogonal score","authors":"Xing Li ,&nbsp;Lei Wang ,&nbsp;Heng Lian","doi":"10.1016/j.knosys.2025.113529","DOIUrl":"10.1016/j.knosys.2025.113529","url":null,"abstract":"<div><div>Massive data bring a big challenge for analysis, and subsampling as an effective solution can significantly reduce the computational burden and maintain estimation efficiency. Expected Shortfall Regression (ESR) studies the impact of covariates on the tail expectation of response and explores the heterogeneous effects of the covariates. For joint linear quantile and expected shortfall regression models, we study the optimal subsampling method for ESR based on the Neyman-orthogonal score to reduce sensitivity with respect to nuisance parameters in quantile regression. When the massive data are stored in different sites, we further propose a distributed optimal subsampling method for the ESR. Asymptotic properties of the resultant estimators are established and the two-step algorithms are proposed for practical implementation. Extensive simulations and applications to Protein Tertiary Structure and Beijing Air Quality datasets show satisfactory performance of the proposed estimators.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113529"},"PeriodicalIF":7.2,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143874097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TFC: Time–frequency contrasting network for wearable-based human activity recognition","authors":"Zhenzhen Huang ,&nbsp;Jiukai Deng ,&nbsp;Shengzhi Wang ,&nbsp;Chaogang Tang ,&nbsp;Shuo Xiao","doi":"10.1016/j.knosys.2025.113373","DOIUrl":"10.1016/j.knosys.2025.113373","url":null,"abstract":"<div><div>Human Activity Recognition (HAR) using sensor data has significantly progressed with various supervised learning architectures, including both traditional CNN/LSTM models and the more recent Transformer-based models. A primary challenge in supervised learning is the requirement for extensive, accurately labeled training data. Self-supervised methods, particularly those employing contrastive learning, offer an innovative solution to this challenge by leveraging unlabeled data. In this study, we introduce a novel self-supervised learning method named Time–Frequency Contrasting (TFC) for limited labeled data in HAR, rooted in the principles of contrastive learning and Bayesian structural time series. This approach interprets sensor data as a combination of time-domain trends, frequency-domain cycles, and error noise. Our objective is to learn a universal representation so as to enhance the performance of human activity recognition in downstream tasks. This is achieved by minimizing the impact of redundant noise and leveraging time-domain prior knowledge to learn time-domain trend features and utilizing frequency-domain prior knowledge to acquire frequency-domain cycle features, respectively. After fine-tuning, TFC achieved Macro F1-scores of 86.39, 95.44, and 80.27 on three publicly available datasets, namely MotionSense, USC-HAD, and UCI-HAR. Additionally, it obtained a F1-score of 97.64 on a custom-built dataset called BARD. Our extensive experiment demonstrate that TFC markedly improves self-supervised activity recognition tasks, especially in scenarios with limited labeled data.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"319 ","pages":"Article 113373"},"PeriodicalIF":7.2,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143881446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DLAN: A dual attention network for effective land cover classification in remote sensing","authors":"Muhammad Fayaz ,&nbsp;L. Minh Dang ,&nbsp;Hyeonjoon Moon","doi":"10.1016/j.knosys.2025.113620","DOIUrl":"10.1016/j.knosys.2025.113620","url":null,"abstract":"<div><div>In the era of remote sensing (RS), the demand for accurate land cover classification (LCC) has intensified due to various environmental challenges such as deforestation and urbanization. Conventional approaches often rely on shallow features for classification, limiting their effectiveness in capturing spatial patterns and diverse land cover types. In response, this study introduces a novel LCC approach utilizing a convolutional neural network (CNN) equipped with a dual land cover attention segment. The proposed module integrates channel attention (CA) and spatial attention mechanisms (SA) to enhance the discriminative capabilities of deep models. Leveraging inter-channel and inter-spatial relationships, the dual attention module enables the identification of various land cover types, spatial patterns, and color variations. Through thorough experimentation, the InceptionV3 feature extractor was identified as the optimal backbone for the proposed network architecture. Furthermore, to address the challenge of diverse land cover types, highly curated datasets are utilized. Additionally, to optimize model efficiency and reduce size, an improved model compression approach is employed. The effectiveness of the proposed Dual Land Cover Attention Network (DLAN) was evaluated through extensive experimentation, demonstrating superior performance compared to conventional methods. The results indicate the potential of DLAN in advancing LCC tasks, facilitating detailed agricultural zoning, environmental monitoring, and urban planning at a regional scale.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"319 ","pages":"Article 113620"},"PeriodicalIF":7.2,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143895420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"LGDAAN-Nets: A local and global domain adversarial attention neural networks for EEG emotion recognition","authors":"Yanling An ,&nbsp;Shaohai Hu ,&nbsp;Shuaiqi Liu ,&nbsp;Xinrui Wang ,&nbsp;Zhihui Gu ,&nbsp;Yudong Zhang","doi":"10.1016/j.knosys.2025.113613","DOIUrl":"10.1016/j.knosys.2025.113613","url":null,"abstract":"<div><div>Extensive research is being conducted worldwide on emotion recognition, which is a crucial technology in affective computing. Electroencephalogram (EEG) signals are widely employed in emotion recognition owing to their ease of discernibility and high accuracy. Effectively harnessing the spatial-temporal-spectral features of EEG signals is essential for realizing accurate emotion classification due to their low signal-to-noise ratio. In this study, we proposed an EEG emotion recognition algorithm based on local and global domain adversarial attention neural networks, called LGDAAN-Nets, to address the problems of cross-subject EEG emotion recognition. Firstly, we constructed a ConvLSTM block with residual structures as a spatial-temporal-spectral feature to fully exploit the temporal relationship, spatial structure, and spectral information of the input spatial-temporal matrix and spatial-spectral matrix in the network. We then introduced a self-attention module as a supplementary component to the feature extractor, which integrates the long-range and multilevel dependencies of the cross-modal emotion features. This facilitates the learning of complementary information from different feature patterns and enhances the emotion recognition capability of the model. Lastly, we built a local-global domain discriminator using two local domain discriminators that reduce the distribution differences under different feature patterns and capture the locally invariant features of the EEG signals. The global domain discriminator minimizes the global differences in the fused features between the source and target domains, which improves the robustness and generalization performance of the model. The proposed method was comprehensively tested on the SEED, SEED-IV, and DEAP datasets and demonstrated superior performance over most existing emotion recognition methods. Additionally, experiments were also conducted on a self-collected EEG-based emotion dataset that included 20 subjects, which further validated the proposed model's performance in cross-dataset emotion recognition. The source code is available at: <span><span>https://github.com/cvmdsp/LGDAAN-Nets</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113613"},"PeriodicalIF":7.2,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143865179","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An efficient grid-based path planning approach using improved artificial bee colony algorithm","authors":"Mustafa Yusuf Yildirim ,&nbsp;Rustu Akay","doi":"10.1016/j.knosys.2025.113528","DOIUrl":"10.1016/j.knosys.2025.113528","url":null,"abstract":"<div><div>Grid-based path planning with large solution spaces, is considered computationally hard because of the computational time required to examine all possible paths. Many algorithms have been developed to solve this problem, one of which is the artificial bee colony (ABC) algorithm, known for its strong search capabilities. In this paper, an improved artificial bee colony algorithm (IABC), designed to achieve a balance between exploitation and exploration by integrating two mechanisms, is proposed. First, a path linearization strategy that eliminates unnecessary corners in the planned path within the grid environment is integrated. Second, a local search strategy is employed to enhance the convergence speed of ABC and improve its ability to find the global optimum solution. To evaluate the performance of IABC, it is first compared with the basic ABC in environments of the same size and demonstrates improvements in the range of 7%–14% in terms of path length. Secondly, the contributions of the two improvement strategies are analyzed through ablation studies. Thirdly, IABC is tested for scalability by running it in environments of varying sizes, achieving improvements in the range of 19%–20%. Fourthly, IABC is compared with the advanced ABC variants, achieving improvements in the range of 2%–32%. Fifthly, IABC is compared with the well-known and recent advanced algorithms, achieving improvements starting from 2%. Finally, IABC is evaluated against the results from recent studies in the literature, showing improvements of up to 37%. These results demonstrate that IABC is an effective method for solving grid-based path planning problems.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113528"},"PeriodicalIF":7.2,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143860518","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Systems of implications obtained using the Carve decomposition of a formal context","authors":"Domingo López-Rodríguez ,&nbsp;Manuel Ojeda-Hernández ,&nbsp;Tim Pattison","doi":"10.1016/j.knosys.2025.113475","DOIUrl":"10.1016/j.knosys.2025.113475","url":null,"abstract":"<div><div>The <span>Carve</span> algorithm uses a divide-and-conquer strategy to compute the concept lattice of a formal context. The decomposition phase of the <span>Carve</span> algorithm discovers hierarchical structure in an amenable formal context, which the synthesis phase then exploits to construct the concept lattice from those of the component sub-contexts. In this paper, the problem of computing a sound and complete set of attribute implications via a refinement of the <span>Carve</span> decomposition is studied. Indeed, a set of rules is devised to obtain a set of valid implications which is proved to be complete. The refined decomposition and these rules are implemented in the novel <span>Carve+</span> algorithm, whose runtime compares favorably with direct computation of the Duquenne–Guigues base of implications via the <span>NextClosure</span> algorithm.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"318 ","pages":"Article 113475"},"PeriodicalIF":7.2,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143860515","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Anchor graph based connectivity peaks clustering","authors":"Mingjie Cai ,&nbsp;Jiangyuan Wang ,&nbsp;Feng Xu ,&nbsp;Qiong Liu ,&nbsp;Hamido Fujita","doi":"10.1016/j.knosys.2025.113498","DOIUrl":"10.1016/j.knosys.2025.113498","url":null,"abstract":"<div><div>Clustering by fast search and find of density peaks (DPC), a classic density-based algorithm, excels in identifying clusters of arbitrary shape. However, it struggles in recognizing complex structures due to challenges in selecting density peaks and allocating non-central points. To address these issues, we propose an anchor graph based connectivity peaks clustering method which is the connection between anchor graph and density-based clustering, called AG-CPC. Firstly, it introduces a new concept of connectivity by analyzing the divergence and discreteness of neighborhood adjacency graph to detect low-density clusters and border points. Secondly, a robust two-stage assignment strategy using an adaptive parent–child relationships based on data distribution characteristics, is proposed to reduce the wrong allocation of non-central points. Lastly, a local method for constructing anchor graphs is introduced, combined with fuzzy connectivity and boundary domains of clusters, to scale down the anchor graphs and establish the connection among anchor points. The experiments demonstrate the efficiency and stability of the proposed algorithm compared to state-of-the-art algorithms on synthetic, real-world, and image datasets.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"319 ","pages":"Article 113498"},"PeriodicalIF":7.2,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143886763","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}