Information Sciences最新文献_第2页

Towards heterogeneous federated graph learning via structural entropy and prototype aggregation 基于结构熵和原型聚合的异构联邦图学习

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-27 DOI: 10.1016/j.ins.2025.122338

Zhehao Dai , Guojiang Shen , Haopeng Yuan , Shangfei Zheng , Yuyue Hu , Jiaxin Du , Xiangjie Kong , Feng Xia

{"title":"Towards heterogeneous federated graph learning via structural entropy and prototype aggregation","authors":"Zhehao Dai , Guojiang Shen , Haopeng Yuan , Shangfei Zheng , Yuyue Hu , Jiaxin Du , Xiangjie Kong , Feng Xia","doi":"10.1016/j.ins.2025.122338","DOIUrl":"10.1016/j.ins.2025.122338","url":null,"abstract":"<div><div>In today's data-driven landscape, Federated Graph Learning (FGL) facilitates collaborative training between distributed data while providing robust privacy protections. However, FGL faces significant challenges in practical application: data heterogeneity owing to divergent node distributions and graph structures across clients, coupled with model heterogeneity caused by heterogeneous GNN architectures, substantially impedes the aggregation efficacy and generalization capabilities of global models. Existing FGL frameworks often overlook the unique impact of graph topology, inherent to graph data, between data and model heterogeneity. We propose an innovative framework called Structural Entropy Federated Graph Learning (SEFGL) that leverages structural entropy to simultaneously address data and model heterogeneity. At the client level, structural entropy-based virtual node generation and graph reconstruction techniques are applied to strengthen minority class node representations and optimize local graph topology while maintaining the original data distribution. At the server level, a prototype learning approach based on structural entropy aggregates data from clients with similar entropy characteristics. This enables each client to acquire a more diverse global representation, fostering the development of a personalized and robust prototype. Experiments conducted on three graph datasets demonstrate that the SEFGL framework achieves superior performance in terms of generalizability, efficiency, and effectiveness in high-heterogeneity scenarios.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"718 ","pages":"Article 122338"},"PeriodicalIF":8.1,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144168676","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}

引用次数: 0

A novel virtual sample generation method based on dual correlation generative adversarial nets for soft sensing application 基于对偶相关生成对抗网络的虚拟样本生成方法在软测量中的应用

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-26 DOI: 10.1016/j.ins.2025.122334

Qun-Xiong Zhu , Jie-Long Zhang , Xiao-Lu Song , Yan-Lin He , Yuan Xu

{"title":"A novel virtual sample generation method based on dual correlation generative adversarial nets for soft sensing application","authors":"Qun-Xiong Zhu , Jie-Long Zhang , Xiao-Lu Song , Yan-Lin He , Yuan Xu","doi":"10.1016/j.ins.2025.122334","DOIUrl":"10.1016/j.ins.2025.122334","url":null,"abstract":"<div><div>In industrial production processes, due to the limitation of the environment and economy, the soft sensor models faced the challenge of poor prediction effect caused by the small sample. To solve the above challenges, this paper puts forward a novel virtual sample generation method based on dual correlation generative adversarial nets (DCRGAN-VSG). In DCRGAN, Wasserstein GAN (WGAN) is used to generate the process variables, the improved Conditional GAN (CGAN) is used to generate the quality variables, and the mean square error is added to the value function of CGAN to improve the mapping relationship between the process variables and quality variables. In addition, the Pielou index is used to select process variables that can fill the sparse region from the generated samples. Finally, training soft sensors using the dataset containing virtual samples. The benchmark function and industrial case called the purified terephthalic acid (PTA) data are used to verify the virtual sample generation performance of the DCRGAN-VSG. Simulation results show that DCRGAN-VSG improves the accuracy of soft sensing models by approximately 70% in benchmark experiments and by nearly 40% in industrial case studies.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122334"},"PeriodicalIF":8.1,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144154436","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}

引用次数: 0

Efficient multi-party privacy preserving federated k-means based on homomorphic encryption 基于同态加密的高效多方隐私保护联合k-均值

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-26 DOI: 10.1016/j.ins.2025.122335

Zeng-Ao Tang , Xue-Feng Duan , Rong-Hua Liang , Yong Ding

{"title":"Efficient multi-party privacy preserving federated k-means based on homomorphic encryption","authors":"Zeng-Ao Tang , Xue-Feng Duan , Rong-Hua Liang , Yong Ding","doi":"10.1016/j.ins.2025.122335","DOIUrl":"10.1016/j.ins.2025.122335","url":null,"abstract":"<div><div>Nowadays, the collection and storage of data is increasingly decentralized, and the demand for data mining on distributed data is growing. Traditional k-means risks privacy leaks through direct data sharing. Existing privacy-preserving methods still expose intermediate cluster details during iterations. This paper introduces DTK-means (distributed privacy-preserving k-means) clustering algorithm, a distributed privacy-preserving k-means method addressing these issues. It involves multiple users and two non-colluding servers. The user locally computes centroids, encrypts them, and submits to the server. The two servers then aggregate these encrypted centroids into global cluster centroids with multiplicative perturbation, ensuring that neither the participants nor the servers have knowledge of the specific details. The scheme includes four algorithms for compute local centroids, data aggregation, compute global centroids and output final cluster centroids, implemented using Paillier homomorphic encryption. An extensive performance analysis is carried out to show that DTK-means ensures that intermediate data and private data are concealed from all parties involved. Participants can accurately perform k-means clustering using these hidden global centroids without any information loss. Furthermore, it can resist collusion attacks, even if one server colludes with all participants except one. Complexity analysis and numerical experiments show that our algorithm has good efficiency.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122335"},"PeriodicalIF":8.1,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144139598","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}

引用次数: 0

A rate allocation model for VVC intercoding using a quality dependency 基于质量依赖的VVC互编码的速率分配模型

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-26 DOI: 10.1016/j.ins.2025.122333

Heqiang Wang , Xuekai Wei , Mingliang Zhou , Horace Ho Shing Ip , Sam Kwong

引用次数: 0

Supply network disruption: A framework for assessing vulnerability and implementing resilience strategies 供应网络中断：评估脆弱性和实施弹性战略的框架

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-26 DOI: 10.1016/j.ins.2025.122336

Michael-Sam Vidza , Marcin Budka , Wei Koong Chai , Mark Thrush , Mickaël Teixeira Alves

{"title":"Supply network disruption: A framework for assessing vulnerability and implementing resilience strategies","authors":"Michael-Sam Vidza , Marcin Budka , Wei Koong Chai , Mark Thrush , Mickaël Teixeira Alves","doi":"10.1016/j.ins.2025.122336","DOIUrl":"10.1016/j.ins.2025.122336","url":null,"abstract":"<div><div>Disruptions to food supply chains can have significant impacts on food security and economic stability. This study investigates the resilience of supply networks to such disruptions, focusing on the distribution of live fish between farms in England and Wales as a case study. A decision support framework is developed to assess network vulnerability and ensure operational continuity in the face of disruptions to the supply and demand balance. The framework incorporates a novel rewiring algorithm that dynamically reconfigures network connections to maintain the flow of goods. The algorithm predicts supply-demand pairs and adjusts connections to preserve functionality during disruptions. To evaluate the performance of the framework and algorithm, a combination of topological metrics, such as connectivity and redundancy, and operational measures, including supply fulfilment and distribution efficiency, is utilised. Through simulations of random and targeted node removals, the rewiring algorithm is shown to effectively mitigate the impact of disruptions, preserve network functionality, and help ensure a consistent supply of live fish. These findings offer valuable insights for managing disruptions in aquaculture supply chains and highlight the broader applicability of the framework to enhance the resilience of other supply networks.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122336"},"PeriodicalIF":8.1,"publicationDate":"2025-05-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144154435","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}

引用次数: 0

You only adapt once: An adaptive transformer for dynamic multivariate time series forecasting across time-varying topologies and multi-patterns 您只需适应一次：用于跨时变拓扑和多模式的动态多元时间序列预测的自适应转换器

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-24 DOI: 10.1016/j.ins.2025.122350

Shuai Zhang, Jiyuan Xu, Wenyu Zhang, Yixiang Zhang, Chengjie Ni

{"title":"You only adapt once: An adaptive transformer for dynamic multivariate time series forecasting across time-varying topologies and multi-patterns","authors":"Shuai Zhang, Jiyuan Xu, Wenyu Zhang, Yixiang Zhang, Chengjie Ni","doi":"10.1016/j.ins.2025.122350","DOIUrl":"10.1016/j.ins.2025.122350","url":null,"abstract":"<div><div>Although deep learning has been effective in capturing spatiotemporal patterns within Multivariate Time Series (MTS) data, dynamic shifts in sensor network topologies present substantial challenges. These challenges include adapting to new topologies, managing missing data, and preserving the integrity of spatial dependencies. Existing methods, such as model retraining, are resource-intensive and time-consuming while lacking the robustness required for dynamic urban environments. Hence, a novel You Only Adapt Once (YOAO) model comprising a dynamic MTS encoder and multi-scale decoder is proposed to adapt to time-varying topologies and diverse spatiotemporal multi-patterns. A new dynamic MTS encoder with variable input dimensions on node-level is proposed to adapt seamlessly to time-varying topologies, and efficiently handle severe missing values. A new multi-scale decoder is proposed to self-adaptively abstract complex spatiotemporal multi-patterns in MTS data, and integrate fine-grained representations across scales for comprehensive multi-pattern analysis. In addition, a new node-wise curriculum learning method is proposed to enhance the training efficiency and model performance. Extensive experiments across three real-world datasets and two subtasks, both of which simulate time-varying topologies, demonstrate that YOAO outperforms seven state-of-the-art baselines by an average of 18.75%. Moreover, YOAO reduces training time by an average of 41.52% compared with the five transformer-based models.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122350"},"PeriodicalIF":8.1,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144154439","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}

引用次数: 0

Game-based consensus model considering individual effort 考虑个体努力的博弈共识模型

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-23 DOI: 10.1016/j.ins.2025.122332

Dong Cheng, Jiali Qin, Zhe Hou, Yong Wu

{"title":"Game-based consensus model considering individual effort","authors":"Dong Cheng, Jiali Qin, Zhe Hou, Yong Wu","doi":"10.1016/j.ins.2025.122332","DOIUrl":"10.1016/j.ins.2025.122332","url":null,"abstract":"<div><div>In the consensus-reaching process (CRP), decision-makers (DMs) need to make efforts to adjust their inconsistent opinions, and the degree of effort is affected by the compensation offered by the moderator. However, existing consensus models often ignore DMs' efforts in the CRP. To address this problem, we propose a game-based consensus model considering individual effort and investigate the influence of effort cost on the CRP. First, the perceived utility function considering individual effort of each DM is defined. Then, we propose a feedback adjustment mechanism incorporating DM's effort degree and unit effort cost. The optimal adjustment strategy could be obtained by constructing a game-based consensus model and an improved particle swarm optimization algorithm. The validity of the model is illustrated by an example of the carbon reduction negotiation in the supply chain. Results show that: (1) Effort degree of each DM is related to the compensation strategy, which determines whether the DM accepts the moderator's suggested opinion. (2) An increase in unit effort cost will significantly reduce the effort degree of each DM, thus reducing the compensation cost of the moderator. It can promote consensus-reaching by making reasonable compensation strategies according to the effort cost of each DM.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122332"},"PeriodicalIF":8.1,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144139536","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}

引用次数: 0

GR Loss+: A unified perspective for weakly supervised multi-label learning GR Loss+：弱监督多标签学习的统一视角

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-22 DOI: 10.1016/j.ins.2025.122326

Yanxi Chen , Chunxiao Li , Xiaofei Zhou , Bo Wang

{"title":"GR Loss+: A unified perspective for weakly supervised multi-label learning","authors":"Yanxi Chen , Chunxiao Li , Xiaofei Zhou , Bo Wang","doi":"10.1016/j.ins.2025.122326","DOIUrl":"10.1016/j.ins.2025.122326","url":null,"abstract":"<div><div>Multi-label Learning (MLL) often requires fully supervised label information, resulting in high annotation costs and motivating the development of weakly supervised MLL. Previous studies typically address specific weakly supervised scenarios independently, such as Multi-label Learning with Missing Labels (MLML) and Partial Multi-label Learning (PML), which limits their generality. In this paper, we adopt a unified perspective across diverse weak supervision settings and propose a general loss function framework that is broadly applicable. Furthermore, we argue that a well-designed loss function not only mitigates label noise with low computational cost but also demonstrates strong extensibility. Specifically, we focus on the most challenging Single Positive Multi-label Learning (SPML) scenario, where each sample is annotated with only one positive label. The properties and hyperparameter update strategy of our SPML loss are investigated, and theoretical insights are provided through pseudo-label risk and gradient analysis. Extensive experiments on benchmark datasets and the real-world RFMiD 2.0 dataset demonstrate the superiority and practical effectiveness of our method, achieving significant improvements over existing approaches across various weak supervision settings. Our code is available at <span><span>https://github.com/yan4xi1/GRLoss</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122326"},"PeriodicalIF":8.1,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134752","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}

引用次数: 0

Kolmogorov-Arnold network-based enhanced fusion transformer for hyperspectral image classification 基于Kolmogorov-Arnold网络的高光谱图像分类增强融合变压器

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-22 DOI: 10.1016/j.ins.2025.122323

Xingyu Han , Feng Jiang , Shiping Wen , Tianhai Tian

{"title":"Kolmogorov-Arnold network-based enhanced fusion transformer for hyperspectral image classification","authors":"Xingyu Han , Feng Jiang , Shiping Wen , Tianhai Tian","doi":"10.1016/j.ins.2025.122323","DOIUrl":"10.1016/j.ins.2025.122323","url":null,"abstract":"<div><div>Hyperspectral Image (HSI) classification is a crucial research area with wide-ranging applications in precision agriculture, military reconnaissance, and geospatial analysis. Although deep learning has achieved significant progress, two key limitations persist. First, these models often suffer from overfitting due to the limited availability of training samples and the high-dimensional nature of HSIs. Second, existing deep learning methods fail to extract irregular multi-scale local features from HSIs and fail to enhance accuracy by integrating complementary information. To address these challenges, we propose the Kolmogorov-Arnold Network-based Enhanced Fusion Transformer (KANEFT). Specifically, we construct both linear and convolutional structures using Kolmogorov-Arnold Network (KAN), effectively reducing the model complexity and eliminating the need for manually selecting activation functions. Moreover, we design a Multi-scale Fusion Learning Transformer (MSFLT) module to extract different features from HSIs. In MSFLT, parallel convolutional-KAN layers are employed to extract multi-scale local pixel features. In the meantime, spatial attention modules capture global spatial information and cross-attention enables the complementary integration of diverse features. Finally, we integrate multi-scale spectral information with semantic boundaries, leading to more refined classification blocks. KANEFT effectively extracts multi-scale local information and global spatial features from HSI and leverages complementary feature learning to enhance model accuracy. The proposed model achieves remarkable overall accuracy on three public datasets: 96.8% on Indian Pines, 99.98% on Pavia University, and 99.34% on Houston University, outperforming several state-of-the-art models.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122323"},"PeriodicalIF":8.1,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134753","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}

引用次数: 0

A contrast based feature selection algorithm for high-dimensional datasets in machine learning 机器学习中基于对比度的高维数据集特征选择算法

IF 8.1 1区计算机科学

Information Sciences Pub Date : 2025-05-22 DOI: 10.1016/j.ins.2025.122308

Chunxu Cao , Qiang Zhang , Yuhui Deng

{"title":"A contrast based feature selection algorithm for high-dimensional datasets in machine learning","authors":"Chunxu Cao , Qiang Zhang , Yuhui Deng","doi":"10.1016/j.ins.2025.122308","DOIUrl":"10.1016/j.ins.2025.122308","url":null,"abstract":"<div><div>Feature selection plays a pivotal role in enhancing machine learning models by identifying relevant features and eliminating redundancies. However, existing methods often face challenges with high computational costs and inefficiencies, particularly when applied to large-scale, high-dimensional datasets. To address these issues, we propose ContrastFS, a novel contrast-based feature selection method that evaluates feature importance by analyzing discrepancies in feature distributions across different classes. By leveraging dimensionless surrogate of class-wise feature statistics, ContrastFS enables efficient assessment of both feature relevance and redundancy. Comprehensive experiments on diverse benchmark datasets demonstrate that ContrastFS achieves computational efficiency that is several orders of magnitude higher than state-of-the-art methods while maintaining competitive accuracy. Furthermore, it effectively reduces feature redundancy, enhancing both model interpretability and performance. With its efficiency, scalability, and robustness, ContrastFS offers a powerful solution for feature selection in high-dimensional datasets, making it particularly suited for large-scale artificial intelligence applications where speed and accuracy are critical.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"717 ","pages":"Article 122308"},"PeriodicalIF":8.1,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134751","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}

引用次数: 0