Information Fusion最新文献_第6页

Incomplete multi-view clustering based on hypergraph 基于超图的不完全多视图聚类

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-23 DOI: 10.1016/j.inffus.2024.102804

Jin Chen , Huafu Xu , Jingjing Xue , Quanxue Gao , Cheng Deng , Ziyu Lv

The graph-based incomplete multi-view clustering aims at integrating information from multiple views and utilizes graph models to capture the global and local structure of the data for reconstructing missing data, which is suitable for processing complex data. However, ordinary graph learning methods usually only consider pairwise relationships between data points and cannot unearth higher-order relationships latent in the data. And existing graph clustering methods often divide the process of learning the representations and the clustering process into two separate steps, which may lead to unsatisfactory clustering results. Besides, they also tend to consider only intra-view similarity structures and overlook inter-view ones. To this end, this paper introduces an innovative one-step incomplete multi-view clustering based on hypergraph (IMVC_HG). Specifically, we use a hypergraph to reconstruct missing views, which can better explore the local structure and higher-order information between sample points. Moreover, we use non-negative matrix factorization with orthogonality constraints to equate K-means, which eliminates post-processing operations and avoids the problem of suboptimal results caused by the two-step approach. In addition, the tensor Schatten

p

-norm is used to better capture the complementary information and low-rank structure between the cluster label matrices of multiple views. Numerous experiments verify the superiority of IMVC_HG.

基于图的不完全多视图聚类旨在整合来自多个视图的信息，利用图模型捕捉数据的全局和局部结构来重建缺失数据，适用于处理复杂数据。然而，普通的图学习方法通常只考虑数据点之间的配对关系，无法发掘数据中潜藏的高阶关系。而且现有的图聚类方法往往将学习表示过程和聚类过程分为两个独立的步骤，这可能会导致聚类结果不尽人意。此外，它们还往往只考虑视图内的相似性结构，而忽略了视图间的相似性结构。为此，本文引入了一种创新的基于超图的一步不完全多视图聚类（IMVC_HG）。具体来说，我们使用超图来重建缺失视图，这样可以更好地探索样本点之间的局部结构和高阶信息。此外，我们使用带有正交性约束的非负矩阵因式分解来等效 K-means，省去了后处理操作，避免了两步法造成的结果不理想的问题。此外，还使用了张量 Schatten p-norm，以更好地捕捉多视图聚类标签矩阵之间的互补信息和低秩结构。大量实验验证了 IMVC_HG 的优越性。

{"title":"Incomplete multi-view clustering based on hypergraph","authors":"Jin Chen , Huafu Xu , Jingjing Xue , Quanxue Gao , Cheng Deng , Ziyu Lv","doi":"10.1016/j.inffus.2024.102804","DOIUrl":"10.1016/j.inffus.2024.102804","url":null,"abstract":"<div><div>The graph-based incomplete multi-view clustering aims at integrating information from multiple views and utilizes graph models to capture the global and local structure of the data for reconstructing missing data, which is suitable for processing complex data. However, ordinary graph learning methods usually only consider pairwise relationships between data points and cannot unearth higher-order relationships latent in the data. And existing graph clustering methods often divide the process of learning the representations and the clustering process into two separate steps, which may lead to unsatisfactory clustering results. Besides, they also tend to consider only intra-view similarity structures and overlook inter-view ones. To this end, this paper introduces an innovative one-step <em>incomplete multi-view clustering based on hypergraph (IMVC_HG)</em>. Specifically, we use a hypergraph to reconstruct missing views, which can better explore the local structure and higher-order information between sample points. Moreover, we use non-negative matrix factorization with orthogonality constraints to equate K-means, which eliminates post-processing operations and avoids the problem of suboptimal results caused by the two-step approach. In addition, the tensor Schatten <span><math><mi>p</mi></math></span>-norm is used to better capture the complementary information and low-rank structure between the cluster label matrices of multiple views. Numerous experiments verify the superiority of IMVC_HG.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"117 ","pages":"Article 102804"},"PeriodicalIF":14.7,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142721315","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

Self-supervised learning-based multi-source spectral fusion for fruit quality evaluation:a case study in mango fruit ripeness prediction 基于自监督学习的水果质量评估多源光谱融合：芒果果实成熟度预测案例研究

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-23 DOI: 10.1016/j.inffus.2024.102814

Liu Zhang , Jincun Liu , Yaoguang Wei , Dong An , Xin Ning

Rapid and non-destructive techniques for fruit quality evaluation are widely concerned in modern agro-industry. Spectroscopy is one of the most commonly used techniques in this field. With the growing popularity of various spectroscopic instruments, it is indeed worthwhile to explore modeling with multi-source spectral data to achieve more accurate predictions. Nonetheless, a major challenge is acquiring enough labeled samples, as measuring fruit chemical values is laborious, expensive, and time-consuming, which hinders the development of a reliable prediction model. Therefore, this study aims to develop a model for predicting the internal chemical composition of fruits by integrating multi-source spectral fusion combined with self-supervised learning (SSL). A visible (Vis) and near-infrared (NIR) spectral dataset related to dry matter content (DMC) prediction in mango fruit is used as an example to validate the effectiveness of the proposed method. To obtain multi-source spectral data, the Vis and NIR portions are processed as two separate spectral ranges. An SSL pre-training is performed utilizing a large amount of raw unlabeled spectral data to extract general knowledge, which is subsequently migrated to a downstream task for fine-tuning. The experimental results indicate that the multi-source spectral fusion model performs better than the single-source spectral model. Moreover, SSL solves the data scarcity problem and outperforms non-pre-trained models in downstream DMC prediction tasks with less computational overhead. Remarkably, utilizing only less than 10% of the total samples is sufficient to achieve a performance close to 99% of the best results. The presented method has great potential in spectral analysis of food and agro-products.

快速、非破坏性的水果质量评估技术在现代农产品加工业中受到广泛关注。光谱技术是这一领域最常用的技术之一。随着各种光谱仪器的日益普及，利用多源光谱数据建模以实现更准确的预测确实值得探索。然而，获取足够的标记样本是一大挑战，因为测量水果化学值费力、昂贵且耗时，这阻碍了可靠预测模型的开发。因此，本研究旨在通过将多源光谱融合与自我监督学习（SSL）相结合，开发一种预测水果内部化学成分的模型。以预测芒果果实干物质含量（DMC）的可见光（Vis）和近红外（NIR）光谱数据集为例，验证了所提方法的有效性。为了获得多源光谱数据，可见光和近红外部分被作为两个独立的光谱范围进行处理。利用大量未标记的原始光谱数据进行 SSL 预训练，以提取一般知识，然后将其迁移到下游任务中进行微调。实验结果表明，多源光谱融合模型的性能优于单源光谱模型。此外，在下游 DMC 预测任务中，SSL 解决了数据稀缺问题，并以更少的计算开销超越了非预训练模型。值得注意的是，只需利用不到总样本量的 10%，就能获得接近 99% 的最佳结果。该方法在食品和农产品的光谱分析中具有巨大潜力。

{"title":"Self-supervised learning-based multi-source spectral fusion for fruit quality evaluation:a case study in mango fruit ripeness prediction","authors":"Liu Zhang , Jincun Liu , Yaoguang Wei , Dong An , Xin Ning","doi":"10.1016/j.inffus.2024.102814","DOIUrl":"10.1016/j.inffus.2024.102814","url":null,"abstract":"<div><div>Rapid and non-destructive techniques for fruit quality evaluation are widely concerned in modern agro-industry. Spectroscopy is one of the most commonly used techniques in this field. With the growing popularity of various spectroscopic instruments, it is indeed worthwhile to explore modeling with multi-source spectral data to achieve more accurate predictions. Nonetheless, a major challenge is acquiring enough labeled samples, as measuring fruit chemical values is laborious, expensive, and time-consuming, which hinders the development of a reliable prediction model. Therefore, this study aims to develop a model for predicting the internal chemical composition of fruits by integrating multi-source spectral fusion combined with self-supervised learning (SSL). A visible (Vis) and near-infrared (NIR) spectral dataset related to dry matter content (DMC) prediction in mango fruit is used as an example to validate the effectiveness of the proposed method. To obtain multi-source spectral data, the Vis and NIR portions are processed as two separate spectral ranges. An SSL pre-training is performed utilizing a large amount of raw unlabeled spectral data to extract general knowledge, which is subsequently migrated to a downstream task for fine-tuning. The experimental results indicate that the multi-source spectral fusion model performs better than the single-source spectral model. Moreover, SSL solves the data scarcity problem and outperforms non-pre-trained models in downstream DMC prediction tasks with less computational overhead. Remarkably, utilizing only less than 10% of the total samples is sufficient to achieve a performance close to 99% of the best results. The presented method has great potential in spectral analysis of food and agro-products.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"117 ","pages":"Article 102814"},"PeriodicalIF":14.7,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142721316","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

Graph convolutional network for compositional data 组合数据的图卷积网络

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-22 DOI: 10.1016/j.inffus.2024.102798

Shan Lu , Huiwen Wang , Jichang Zhao

Graph convolutional network (GCN) has garnered significant attention and become a powerful tool for learning graph representations. However, when dealing with compositional data prevalent in various fields, the traditional GCN faces theoretical challenges due to the intrinsic constraints of such data. This paper generalizes the spectral graph theory in simplex space, aiming to address the graph structures among observations for compositional data analysis and to extend GCN by assigning mathematical objects of compositions to each vertex of a graph. We propose the graph Fourier transformation in simplex space, based on which a compositional graph convolutional network (CGCN) layer is introduced. This novel layer enables a GCN to appropriately capture the sample space of compositional data, allowing it to handle compositional features as model inputs. We then propose a new GCN architecture called COMP-GCN, incorporating the CGCN layer at the initial stage. We evaluate the effectiveness of COMP-GCN through simulation studies and two real-world applications: stock networks derived from co-investors in the Chinese stock market and student social networks based on co-locations in campus activities. The results demonstrate its superior performance over competitive methods with modest additional computational cost compared to traditional GCN. Our findings suggest the potential of the proposed model to inspire a new class of powerful algorithms for graph inference on compositional data in virtue of the generalization of graph convolution on simplex space.

图卷积网络（GCN）已成为学习图表示的一个强大工具。然而，在处理各个领域普遍存在的成分数据时，由于这些数据的内在约束，传统的GCN在理论上面临挑战。本文在单纯形空间中对谱图理论进行了推广，旨在解决组成数据分析中观测值之间的图结构问题，并通过将组成的数学对象分配到图的每个顶点来扩展GCN。提出了单纯形空间中的图傅里叶变换，并在此基础上引入了复合图卷积网络层。这个新颖的层使GCN能够适当地捕获组合数据的样本空间，允许它处理组合特征作为模型输入。然后，我们提出了一个新的GCN架构，称为COMP-GCN，在初始阶段结合了CGCN层。我们通过模拟研究和两个实际应用来评估COMP-GCN的有效性：来自中国股票市场共同投资者的股票网络和基于校园活动共同地点的学生社交网络。结果表明，与传统GCN相比，该方法的性能优于竞争对手的方法，且计算成本较低。我们的研究结果表明，所提出的模型有潜力激发出一类新的强大算法，利用单纯形空间上的图卷积的泛化来对组成数据进行图推理。

{"title":"Graph convolutional network for compositional data","authors":"Shan Lu , Huiwen Wang , Jichang Zhao","doi":"10.1016/j.inffus.2024.102798","DOIUrl":"10.1016/j.inffus.2024.102798","url":null,"abstract":"<div><div>Graph convolutional network (GCN) has garnered significant attention and become a powerful tool for learning graph representations. However, when dealing with compositional data prevalent in various fields, the traditional GCN faces theoretical challenges due to the intrinsic constraints of such data. This paper generalizes the spectral graph theory in simplex space, aiming to address the graph structures among observations for compositional data analysis and to extend GCN by assigning mathematical objects of compositions to each vertex of a graph. We propose the graph Fourier transformation in simplex space, based on which a compositional graph convolutional network (CGCN) layer is introduced. This novel layer enables a GCN to appropriately capture the sample space of compositional data, allowing it to handle compositional features as model inputs. We then propose a new GCN architecture called COMP-GCN, incorporating the CGCN layer at the initial stage. We evaluate the effectiveness of COMP-GCN through simulation studies and two real-world applications: stock networks derived from co-investors in the Chinese stock market and student social networks based on co-locations in campus activities. The results demonstrate its superior performance over competitive methods with modest additional computational cost compared to traditional GCN. Our findings suggest the potential of the proposed model to inspire a new class of powerful algorithms for graph inference on compositional data in virtue of the generalization of graph convolution on simplex space.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"117 ","pages":"Article 102798"},"PeriodicalIF":14.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142744310","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

When multi-view meets multi-level: A novel spatio-temporal transformer for traffic prediction 当多视角遇上多层次：一种新的交通预测时空转换器

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-22 DOI: 10.1016/j.inffus.2024.102801

Jiaqi Lin, Qianqian Ren, Xingfeng Lv, Hui Xu, Yong Liu

Traffic prediction is a vital aspect of Intelligent Transportation Systems with widespread applications. The main challenge is accurately modeling the complex spatial and temporal relationships in traffic data. Spatial–temporal Graph Neural Networks (GNNs) have emerged as one of the most promising methods to solve this problem. However, several key issues have not been well addressed in existing studies. Firstly, traffic patterns have significant periodic trends, existing methods often overlook the importance of periodicity. Secondly, most methods model spatial dependencies in a static manner, which limits the ability to learn dynamic traffic patterns. Lastly, achieving satisfactory results for both long-term and short-term forecasting remains a challenge. To tackle the above problems, this paper proposes a Multi-level Multi-view Augmented Spatio-temporal Transformer (LVSTformer) for traffic prediction, which captures spatial dependencies from three different levels: local geographic, global semantic, and pivotal nodes, along with long- and short-term temporal dependencies. Specifically, we design three spatial augmented views to delve into the spatial information from above three levels. By combining three spatial augmented views with three parallel spatial self-attention mechanisms, the model can comprehensively captures spatial dependencies at different levels. We design a gated temporal self-attention mechanism to dynamically capture long- and short-term temporal dependencies. Furthermore, a spatio-temporal context broadcasting module is introduced between two spatio-temporal layers to ensure a well-distributed allocation of attention scores, alleviating overfitting and information loss, and enhancing the generalization ability and robustness of the model. A comprehensive set of experiments are conducted on six well-known traffic benchmarks, the experimental results demonstrate that LVSTformer achieves state-of-the-art performance compared to competing baselines, with the maximum improvement reaching up to 4.32%.

交通预测是智能交通系统的一个重要方面，有着广泛的应用。主要的挑战是对交通数据中复杂的时空关系进行准确建模。时空图神经网络（GNNs）已成为解决这一问题的最有前途的方法之一。然而，在现有的研究中，有几个关键问题没有得到很好的解决。首先，交通模式具有明显的周期性趋势，现有的方法往往忽略了周期性的重要性。其次，大多数方法以静态方式建模空间依赖关系，这限制了学习动态交通模式的能力。最后，在长期和短期预报方面取得令人满意的结果仍然是一项挑战。为了解决上述问题，本文提出了一种用于交通预测的多级多视图增强时空转换器（LVSTformer），它从三个不同的层面捕获空间依赖关系：局部地理、全局语义和关键节点，以及长期和短期时间依赖关系。具体来说，我们设计了三个空间增强视图，从以上三个层次深入研究空间信息。该模型通过将三种空间增强视图与三种平行的空间自注意机制相结合，可以全面捕获不同层次的空间依赖关系。我们设计了一个封闭的时间自注意机制来动态捕捉长期和短期的时间依赖性。此外，在两个时空层之间引入了时空上下文广播模块，保证了注意力分数的均匀分配，减轻了过拟合和信息丢失，增强了模型的泛化能力和鲁棒性。在6个知名的流量基准上进行了全面的实验，实验结果表明，LVSTformer与竞争基准相比，性能达到了最先进的水平，最大提升幅度可达4.32%。

{"title":"When multi-view meets multi-level: A novel spatio-temporal transformer for traffic prediction","authors":"Jiaqi Lin, Qianqian Ren, Xingfeng Lv, Hui Xu, Yong Liu","doi":"10.1016/j.inffus.2024.102801","DOIUrl":"10.1016/j.inffus.2024.102801","url":null,"abstract":"<div><div>Traffic prediction is a vital aspect of Intelligent Transportation Systems with widespread applications. The main challenge is accurately modeling the complex spatial and temporal relationships in traffic data. Spatial–temporal Graph Neural Networks (GNNs) have emerged as one of the most promising methods to solve this problem. However, several key issues have not been well addressed in existing studies. Firstly, traffic patterns have significant periodic trends, existing methods often overlook the importance of periodicity. Secondly, most methods model spatial dependencies in a static manner, which limits the ability to learn dynamic traffic patterns. Lastly, achieving satisfactory results for both long-term and short-term forecasting remains a challenge. To tackle the above problems, this paper proposes a Multi-level Multi-view Augmented Spatio-temporal Transformer (LVSTformer) for traffic prediction, which captures spatial dependencies from three different levels: local geographic, global semantic, and pivotal nodes, along with long- and short-term temporal dependencies. Specifically, we design three spatial augmented views to delve into the spatial information from above three levels. By combining three spatial augmented views with three parallel spatial self-attention mechanisms, the model can comprehensively captures spatial dependencies at different levels. We design a gated temporal self-attention mechanism to dynamically capture long- and short-term temporal dependencies. Furthermore, a spatio-temporal context broadcasting module is introduced between two spatio-temporal layers to ensure a well-distributed allocation of attention scores, alleviating overfitting and information loss, and enhancing the generalization ability and robustness of the model. A comprehensive set of experiments are conducted on six well-known traffic benchmarks, the experimental results demonstrate that LVSTformer achieves state-of-the-art performance compared to competing baselines, with the maximum improvement reaching up to 4.32%.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"117 ","pages":"Article 102801"},"PeriodicalIF":14.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142744363","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

Fusion of probabilistic linguistic term sets for enhanced group decision-making: Foundations, survey and challenges 融合概率语言术语集，加强群体决策：基础、调查与挑战

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-20 DOI: 10.1016/j.inffus.2024.102802

Xueling Ma , Xinru Han , Zeshui Xu , Rosa M. Rodríguez , Jianming Zhan

Probabilistic linguistic term set (PLTS) provides a flexible and comprehensive approach to reflecting qualitative information about decision makers (DMs) by fusing linguistic terms and probability distributions. This fusion makes PLTS an important focus of fuzzy decision theory. Dealing with uncertainty and ambiguity has always been a major challenge in the group decision-making (GDM) process, and PLTS provides a versatile and effective approach to address these issues. PLTS is able to more accurately represent the preferences and opinions of the DMs, thus improving the accuracy and consistency of decision-making, thereby improving the accuracy and consistency of decision-making. Therefore, the application of PLTSs in GDM (PLTS-GDM) has attracted more and more attention and shown great potential. In this paper, we provide a comprehensive overview of the underlying theories of PLTS-GDM, the existing approaches and the challenges they face. Specifically, we explore how the PLTS utilizes fuzzy information systems to manage imprecise and ambiguous data to enhance the effectiveness of decision-making. In addition, through an extensive review and analysis of the current literature, we summarize the major advances in the field and identify important gaps in the existing research. Finally, we point out future research directions aimed at addressing these challenges and further advancing the application and development of PLTS-GDM. In summary, this paper provides a valuable resource for scholars and practitioners to help them understand and promote the practical applications of PLTS-GDM.

概率语言术语集（PLTS）通过融合语言术语和概率分布，为反映决策者（DM）的定性信息提供了一种灵活而全面的方法。这种融合使 PLTS 成为模糊决策理论的一个重要焦点。处理不确定性和模糊性一直是群体决策（GDM）过程中的一大挑战，而 PLTS 则为解决这些问题提供了一种通用而有效的方法。PLTS 能够更准确地代表 DMs 的偏好和意见，从而提高决策的准确性和一致性，进而提高决策的准确性和一致性。因此，PLTS 在 GDM 中的应用（PLTS-GDM）引起了越来越多的关注，并显示出巨大的潜力。本文全面概述了 PLTS-GDM 的基础理论、现有方法及其面临的挑战。具体而言，我们探讨了 PLTS 如何利用模糊信息系统来管理不精确和模糊的数据，以提高决策的有效性。此外，通过对现有文献的广泛回顾和分析，我们总结了该领域的主要进展，并找出了现有研究中的重要空白。最后，我们指出了未来的研究方向，旨在应对这些挑战，进一步推动 PLTS-GDM 的应用和发展。总之，本文为学者和从业人员提供了宝贵的资源，帮助他们了解和促进 PLTS-GDM 的实际应用。

{"title":"Fusion of probabilistic linguistic term sets for enhanced group decision-making: Foundations, survey and challenges","authors":"Xueling Ma , Xinru Han , Zeshui Xu , Rosa M. Rodríguez , Jianming Zhan","doi":"10.1016/j.inffus.2024.102802","DOIUrl":"10.1016/j.inffus.2024.102802","url":null,"abstract":"<div><div>Probabilistic linguistic term set (PLTS) provides a flexible and comprehensive approach to reflecting qualitative information about decision makers (DMs) by fusing linguistic terms and probability distributions. This fusion makes PLTS an important focus of fuzzy decision theory. Dealing with uncertainty and ambiguity has always been a major challenge in the group decision-making (GDM) process, and PLTS provides a versatile and effective approach to address these issues. PLTS is able to more accurately represent the preferences and opinions of the DMs, thus improving the accuracy and consistency of decision-making, thereby improving the accuracy and consistency of decision-making. Therefore, the application of PLTSs in GDM (PLTS-GDM) has attracted more and more attention and shown great potential. In this paper, we provide a comprehensive overview of the underlying theories of PLTS-GDM, the existing approaches and the challenges they face. Specifically, we explore how the PLTS utilizes fuzzy information systems to manage imprecise and ambiguous data to enhance the effectiveness of decision-making. In addition, through an extensive review and analysis of the current literature, we summarize the major advances in the field and identify important gaps in the existing research. Finally, we point out future research directions aimed at addressing these challenges and further advancing the application and development of PLTS-GDM. In summary, this paper provides a valuable resource for scholars and practitioners to help them understand and promote the practical applications of PLTS-GDM.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"116 ","pages":"Article 102802"},"PeriodicalIF":14.7,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142696895","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

Flare-aware cross-modal enhancement network for multi-spectral vehicle Re-identification 用于多光谱车辆再识别的耀斑感知跨模态增强网络

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-20 DOI: 10.1016/j.inffus.2024.102800

Aihua Zheng , Zhiqi Ma , Yongqi Sun , Zi Wang , Chenglong Li , Jin Tang

Multi-spectral vehicle Re-identification (Re-ID) aims to incorporate complementary visible and infrared information to tackle the challenge of re-identifying vehicles in complex lighting conditions. However, in harsh environments, the discriminative cues in RGB (visible) and NI (near infrared) modalities are significantly lost by the strong flare from vehicle lamps or the sunlight. To handle this problem, we propose a Flare-Aware Cross-modal Enhancement Network (FACENet) to adaptively restore the flare-corrupted RGB and NI features with the guidance from the flare-immunized TI (thermal infrared) spectra. First, to reduce the influence of locally degraded appearance by the intense flare, we propose a Mutual Flare Mask Prediction (MFMP) module to jointly obtain the flare-corrupted masks in RGB and NI modalities in a self-supervised manner. Second, to utilize the flare-immunized TI information to enhance the masked RGB and NI, we propose a Flare-aware Cross-modal Enhancement module (FCE) to adaptively guide feature extraction of masked RGB and NI spectra with the prior flare-immunized knowledge from the TI spectra. Third, to mine the common semantic information of RGB and NI, and alleviate the severe semantic loss in the NI spectra using TI, we propose a Multi-modality Consistency (MC) loss to enhance the semantic consistency among the three modalities. Finally, to evaluate the proposed FACENet while handling the intense flare problem, we contribute a new multi-spectral vehicle Re-ID dataset, named WMVEID863 with additional challenges, such as motion blur, huge background changes, and especially intense flare degradation. Comprehensive experiments on both the newly collected dataset and public benchmark multi-spectral vehicle Re-ID datasets verify the superior performance of the proposed FACENet compared to the state-of-the-art methods, especially in handling the strong flares. The codes and dataset will be released at this link.

多光谱车辆再识别（Re-ID）旨在结合互补的可见光和红外信息，解决在复杂照明条件下重新识别车辆的难题。然而，在恶劣的环境中，RGB（可见光）和 NI（近红外）模式的分辨线索会因车灯或阳光的强烈耀斑而严重丢失。为了解决这个问题，我们提出了一种耀斑感知的跨模态增强网络（FACENet），通过耀斑免疫的 TI（热红外）光谱的引导，自适应地恢复被耀斑破坏的 RGB 和 NI 特征。首先，为了减少强烈耀斑造成的局部外观劣化的影响，我们提出了一个相互耀斑掩码预测（MFMP）模块，以自我监督的方式联合获取 RGB 和 NI 模式中被耀斑破坏的掩码。其次，为了利用耀斑免疫的 TI 信息来增强 RGB 和 NI 掩膜，我们提出了耀斑感知跨模态增强模块 (FCE)，利用来自 TI 光谱的先验耀斑免疫知识自适应地指导 RGB 和 NI 掩膜光谱的特征提取。第三，为了挖掘 RGB 和 NI 的共同语义信息，并减轻使用 TI 时 NI 光谱中严重的语义损失，我们提出了多模态一致性（MC）损失，以增强三种模态之间的语义一致性。最后，为了在处理强烈耀斑问题的同时评估所提出的 FACENet，我们提供了一个新的多光谱车辆再识别数据集，名为 WMVEID863，该数据集面临更多挑战，如运动模糊、巨大的背景变化，尤其是强烈的耀斑退化。在新收集的数据集和公共基准多光谱车辆再识别数据集上进行的综合实验验证了与最先进的方法相比，所提出的 FACENet 性能优越，尤其是在处理强耀斑方面。代码和数据集将在此链接发布。

{"title":"Flare-aware cross-modal enhancement network for multi-spectral vehicle Re-identification","authors":"Aihua Zheng , Zhiqi Ma , Yongqi Sun , Zi Wang , Chenglong Li , Jin Tang","doi":"10.1016/j.inffus.2024.102800","DOIUrl":"10.1016/j.inffus.2024.102800","url":null,"abstract":"<div><div>Multi-spectral vehicle Re-identification (Re-ID) aims to incorporate complementary visible and infrared information to tackle the challenge of re-identifying vehicles in complex lighting conditions. However, in harsh environments, the discriminative cues in RGB (visible) and NI (near infrared) modalities are significantly lost by the strong flare from vehicle lamps or the sunlight. To handle this problem, we propose a Flare-Aware Cross-modal Enhancement Network (FACENet) to adaptively restore the flare-corrupted RGB and NI features with the guidance from the flare-immunized TI (thermal infrared) spectra. First, to reduce the influence of locally degraded appearance by the intense flare, we propose a Mutual Flare Mask Prediction (MFMP) module to jointly obtain the flare-corrupted masks in RGB and NI modalities in a self-supervised manner. Second, to utilize the flare-immunized TI information to enhance the masked RGB and NI, we propose a Flare-aware Cross-modal Enhancement module (FCE) to adaptively guide feature extraction of masked RGB and NI spectra with the prior flare-immunized knowledge from the TI spectra. Third, to mine the common semantic information of RGB and NI, and alleviate the severe semantic loss in the NI spectra using TI, we propose a Multi-modality Consistency (MC) loss to enhance the semantic consistency among the three modalities. Finally, to evaluate the proposed FACENet while handling the intense flare problem, we contribute a new multi-spectral vehicle Re-ID dataset, named WMVEID863 with additional challenges, such as motion blur, huge background changes, and especially intense flare degradation. Comprehensive experiments on both the newly collected dataset and public benchmark multi-spectral vehicle Re-ID datasets verify the superior performance of the proposed FACENet compared to the state-of-the-art methods, especially in handling the strong flares. The codes and dataset will be released at <span><span>this link.</span><svg><path></path></svg></span></div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"116 ","pages":"Article 102800"},"PeriodicalIF":14.7,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142705622","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 hybrid opinion dynamics model with leaders and followers fusing dynamic social networks in large-scale group decision-making 大规模群体决策中融合动态社交网络的领导者与追随者混合舆论动力学模型

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-20 DOI: 10.1016/j.inffus.2024.102799

Yufeng Shen , Xueling Ma , Muhammet Deveci , Enrique Herrera-Viedma , Jianming Zhan

Objectives:

In this study, our goal is to enhance consensus efficiency in complex decision-making scenarios by constructing a large-scale group decision-making (LSGDM) method that integrates dynamic social network (DSN) and opinion dynamics. To this end, we design a model that can effectively cluster experts and dynamically adjust the network structure to more accurately reflect the diversity and complexity of the actual decision-making process.

Methods:

Specifically, we first design an improved Louvain algorithm based on social influence to effectively cluster participants with similar opinions into the same community. Then, we utilize structural hole theory to distinguish opinion leaders and followers in the community, and construct a DSN updating mechanism based on opinion disagreement and trust relationship. Finally, we combine the advantages of the DeGroot and Hegselmann–Krause (HK) models and propose a hybrid opinion dynamics (HOD) model in the LSGDM framework, referred to as DSN-HOD-LSGDM.

Findings:

Experimental results demonstrate that the DSN-HOD-LSGDM model significantly enhances consensus-building efficiency across diverse decision-making communities. The model effectively tracks opinion evolution in complex networks, outperforming conventional methods in both adaptability and scalability.

Novelty:

In this study, we propose an improved Louvain algorithm and dynamic weight allocation mechanism based on influence index, and design a personalized opinion evolution mechanism combined with structural hole theory. By fusing opinion evolution and dynamic trust, we construct a new LSGDM consensus model that realizes the dynamic adjustment of the trust relationship between individuals.

在本研究中，我们的目标是通过构建一种整合了动态社会网络（DSN）和意见动态的大规模群体决策（LSGDM）方法，提高复杂决策场景中的共识效率。为此，我们设计了一个模型，可以有效地对专家进行聚类，并动态调整网络结构，以更准确地反映实际决策过程的多样性和复杂性。

{"title":"A hybrid opinion dynamics model with leaders and followers fusing dynamic social networks in large-scale group decision-making","authors":"Yufeng Shen , Xueling Ma , Muhammet Deveci , Enrique Herrera-Viedma , Jianming Zhan","doi":"10.1016/j.inffus.2024.102799","DOIUrl":"10.1016/j.inffus.2024.102799","url":null,"abstract":"<div><h3>Objectives:</h3><div>In this study, our goal is to enhance consensus efficiency in complex decision-making scenarios by constructing a large-scale group decision-making (LSGDM) method that integrates dynamic social network (DSN) and opinion dynamics. To this end, we design a model that can effectively cluster experts and dynamically adjust the network structure to more accurately reflect the diversity and complexity of the actual decision-making process.</div></div><div><h3>Methods:</h3><div>Specifically, we first design an improved Louvain algorithm based on social influence to effectively cluster participants with similar opinions into the same community. Then, we utilize structural hole theory to distinguish opinion leaders and followers in the community, and construct a DSN updating mechanism based on opinion disagreement and trust relationship. Finally, we combine the advantages of the DeGroot and Hegselmann–Krause (HK) models and propose a hybrid opinion dynamics (HOD) model in the LSGDM framework, referred to as DSN-HOD-LSGDM.</div></div><div><h3>Findings:</h3><div>Experimental results demonstrate that the DSN-HOD-LSGDM model significantly enhances consensus-building efficiency across diverse decision-making communities. The model effectively tracks opinion evolution in complex networks, outperforming conventional methods in both adaptability and scalability.</div></div><div><h3>Novelty:</h3><div>In this study, we propose an improved Louvain algorithm and dynamic weight allocation mechanism based on influence index, and design a personalized opinion evolution mechanism combined with structural hole theory. By fusing opinion evolution and dynamic trust, we construct a new LSGDM consensus model that realizes the dynamic adjustment of the trust relationship between individuals.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"116 ","pages":"Article 102799"},"PeriodicalIF":14.7,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142696898","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

Multimodal sentiment analysis with unimodal label generation and modality decomposition 利用单模态标签生成和模态分解进行多模态情感分析

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-20 DOI: 10.1016/j.inffus.2024.102787

Linan Zhu , Hongyan Zhao , Zhechao Zhu , Chenwei Zhang , Xiangjie Kong

Multimodal sentiment analysis aims to combine information from different modalities to enhance the understanding of emotions and achieve accurate prediction. However, existing methods face issues of information redundancy and modality heterogeneity during the fusion process, and common multimodal sentiment analysis datasets lack unimodal labels. To address these issues, this paper proposes a multimodal sentiment analysis approach based on unimodal label generation and modality decomposition (ULMD). This method employs a multi-task learning framework, dividing the multimodal sentiment analysis task into a multimodal task and three unimodal tasks. Additionally, a modality representation separator is introduced to decompose modality representations into modality-invariant representations and modality-specific representations. This approach explores the fusion between modalities and generates unimodal labels to enhance the performance of the multimodal sentiment analysis task. Extensive experiments on two public benchmark datasets demonstrate the effectiveness of this method.

多模态情感分析旨在结合来自不同模态的信息，加强对情感的理解并实现准确预测。然而，现有方法在融合过程中面临信息冗余和模态异构的问题，而且常见的多模态情感分析数据集缺乏单模态标签。为解决这些问题，本文提出了一种基于单模态标签生成和模态分解（ULMD）的多模态情感分析方法。该方法采用多任务学习框架，将多模态情感分析任务分为一个多模态任务和三个单模态任务。此外，还引入了模态表征分离器，将模态表征分解为模态不变表征和特定模态表征。这种方法探索了模态之间的融合，并生成了单模态标签，从而提高了多模态情感分析任务的性能。在两个公共基准数据集上进行的广泛实验证明了这种方法的有效性。

{"title":"Multimodal sentiment analysis with unimodal label generation and modality decomposition","authors":"Linan Zhu , Hongyan Zhao , Zhechao Zhu , Chenwei Zhang , Xiangjie Kong","doi":"10.1016/j.inffus.2024.102787","DOIUrl":"10.1016/j.inffus.2024.102787","url":null,"abstract":"<div><div>Multimodal sentiment analysis aims to combine information from different modalities to enhance the understanding of emotions and achieve accurate prediction. However, existing methods face issues of information redundancy and modality heterogeneity during the fusion process, and common multimodal sentiment analysis datasets lack unimodal labels. To address these issues, this paper proposes a multimodal sentiment analysis approach based on unimodal label generation and modality decomposition (ULMD). This method employs a multi-task learning framework, dividing the multimodal sentiment analysis task into a multimodal task and three unimodal tasks. Additionally, a modality representation separator is introduced to decompose modality representations into modality-invariant representations and modality-specific representations. This approach explores the fusion between modalities and generates unimodal labels to enhance the performance of the multimodal sentiment analysis task. Extensive experiments on two public benchmark datasets demonstrate the effectiveness of this method.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"116 ","pages":"Article 102787"},"PeriodicalIF":14.7,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142696899","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

Has multimodal learning delivered universal intelligence in healthcare? A comprehensive survey 多模态学习是否为医疗保健提供了通用智能？全面调查

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-19 DOI: 10.1016/j.inffus.2024.102795

Qika Lin , Yifan Zhu , Xin Mei , Ling Huang , Jingying Ma , Kai He , Zhen Peng , Erik Cambria , Mengling Feng

The rapid development of artificial intelligence has constantly reshaped the field of intelligent healthcare and medicine. As a vital technology, multimodal learning has increasingly garnered interest because of data complementarity, comprehensive information fusion, and great application potential. Currently, numerous researchers are dedicating their attention to this field, conducting extensive studies and constructing abundant intelligent systems. Naturally, an open question arises that has multimodal learning delivered universal intelligence in healthcare? To answer this question, we adopt three unique viewpoints for a holistic analysis. Firstly, we conduct a comprehensive survey of the current progress of medical multimodal learning from the perspectives of datasets, task-oriented methods, and universal foundation models. Based on them, we further discuss the proposed question from five issues to explore the real impacts of advanced techniques in healthcare, from data and technologies to performance and ethics. The answer is that current technologies have NOT achieved universal intelligence and there remains a significant journey to undertake. Finally, in light of the above reviews and discussions, we point out ten potential directions for exploration to promote multimodal fusion technologies in the domain, towards the goal of universal intelligence in healthcare.

人工智能的快速发展不断重塑着智能医疗和医药领域。作为一项重要技术，多模态学习因其数据互补性、信息融合全面性以及巨大的应用潜力而日益受到关注。目前，众多研究人员正致力于这一领域，开展广泛研究，构建丰富的智能系统。自然，一个开放性的问题也随之而来：多模态学习是否为医疗保健领域带来了通用智能？为了回答这个问题，我们采用了三个独特的视角进行全面分析。首先，我们从数据集、面向任务的方法和通用基础模型等角度全面考察了当前医学多模态学习的进展。在此基础上，我们从数据、技术、性能和伦理五个方面进一步讨论了提出的问题，探讨先进技术在医疗领域的实际影响。答案是，当前的技术尚未实现普遍智能化，仍有很长的路要走。最后，根据上述回顾和讨论，我们指出了十个潜在的探索方向，以促进该领域的多模态融合技术，实现医疗保健领域的普遍智能化目标。

{"title":"Has multimodal learning delivered universal intelligence in healthcare? A comprehensive survey","authors":"Qika Lin , Yifan Zhu , Xin Mei , Ling Huang , Jingying Ma , Kai He , Zhen Peng , Erik Cambria , Mengling Feng","doi":"10.1016/j.inffus.2024.102795","DOIUrl":"10.1016/j.inffus.2024.102795","url":null,"abstract":"<div><div>The rapid development of artificial intelligence has constantly reshaped the field of intelligent healthcare and medicine. As a vital technology, multimodal learning has increasingly garnered interest because of data complementarity, comprehensive information fusion, and great application potential. Currently, numerous researchers are dedicating their attention to this field, conducting extensive studies and constructing abundant intelligent systems. Naturally, an open question arises that <em>has multimodal learning delivered universal intelligence in healthcare?</em> To answer this question, we adopt three unique viewpoints for a holistic analysis. Firstly, we conduct a comprehensive survey of the current progress of medical multimodal learning from the perspectives of datasets, task-oriented methods, and universal foundation models. Based on them, we further discuss the proposed question from five issues to explore the real impacts of advanced techniques in healthcare, from data and technologies to performance and ethics. The answer is that current technologies have <strong>NOT</strong> achieved universal intelligence and there remains a significant journey to undertake. Finally, in light of the above reviews and discussions, we point out ten potential directions for exploration to promote multimodal fusion technologies in the domain, towards the goal of universal intelligence in healthcare.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"116 ","pages":"Article 102795"},"PeriodicalIF":14.7,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142696902","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

FedFR-ADP: Adaptive differential privacy with feedback regulation for robust model performance in federated learning FedFR-ADP：联合学习中具有反馈调节功能的自适应差分隐私，以实现稳健的模型性能

IF 14.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Information Fusion

Pub Date : 2024-11-19 DOI: 10.1016/j.inffus.2024.102796

Debao Wang, Shaopeng Guan

Privacy preservation is a critical concern in Federated Learning (FL). However, traditional Local Differential Privacy (LDP) methods face challenges in balancing FL model accuracy with noise strength. To address this, we propose a novel adaptive differential privacy method with feedback regulation, FedFR-ADP. First, we employ Earth Mover’s Distance (EMD) to measure the data heterogeneity of each client and adaptively apply Gaussian noise based on the degree of heterogeneity, making the noise addition more targeted and effective. Second, we introduce a feedback regulation mechanism to dynamically tune the privacy budget according to the global model’s error feedback, further enhancing model performance. Finally, we validate our approach through experiments on two commonly used image classification datasets. The experimental results demonstrate that FedFR-ADP outperforms three benchmark algorithms, including DP-FedAvg, in terms of model training accuracy and Mean Squared Error (MSE) under varying degrees of heterogeneity. Compared to these benchmarks, FedFR-ADP achieves at least a 3.05% and 1.76% improvement in training accuracy across both datasets, with significantly reduced MSE fluctuations. This not only boosts model accuracy but also provides more stable noise control.

隐私保护是联合学习（FL）的一个关键问题。然而，传统的局部差分隐私（LDP）方法在平衡 FL 模型准确性和噪声强度方面面临挑战。为了解决这个问题，我们提出了一种新颖的具有反馈调节功能的自适应差分隐私保护方法--FedFR-ADP。首先，我们采用地球移动距离（EMD）来测量每个客户端的数据异质性，并根据异质性程度自适应地应用高斯噪声，从而使噪声添加更有针对性、更有效。其次，我们引入了反馈调节机制，根据全局模型的误差反馈动态调整隐私预算，进一步提高模型性能。最后，我们在两个常用的图像分类数据集上进行了实验，验证了我们的方法。实验结果表明，在不同程度的异质性条件下，FedFR-ADP 在模型训练精度和均方误差（MSE）方面优于包括 DP-FedAvg 在内的三种基准算法。与这些基准算法相比，FedFR-ADP 在两个数据集上的训练准确率分别提高了至少 3.05% 和 1.76%，MSE 波动也显著降低。这不仅提高了模型的准确性，还提供了更稳定的噪声控制。

{"title":"FedFR-ADP: Adaptive differential privacy with feedback regulation for robust model performance in federated learning","authors":"Debao Wang, Shaopeng Guan","doi":"10.1016/j.inffus.2024.102796","DOIUrl":"10.1016/j.inffus.2024.102796","url":null,"abstract":"<div><div>Privacy preservation is a critical concern in Federated Learning (FL). However, traditional Local Differential Privacy (LDP) methods face challenges in balancing FL model accuracy with noise strength. To address this, we propose a novel adaptive differential privacy method with feedback regulation, FedFR-ADP. First, we employ Earth Mover’s Distance (EMD) to measure the data heterogeneity of each client and adaptively apply Gaussian noise based on the degree of heterogeneity, making the noise addition more targeted and effective. Second, we introduce a feedback regulation mechanism to dynamically tune the privacy budget according to the global model’s error feedback, further enhancing model performance. Finally, we validate our approach through experiments on two commonly used image classification datasets. The experimental results demonstrate that FedFR-ADP outperforms three benchmark algorithms, including DP-FedAvg, in terms of model training accuracy and Mean Squared Error (MSE) under varying degrees of heterogeneity. Compared to these benchmarks, FedFR-ADP achieves at least a 3.05% and 1.76% improvement in training accuracy across both datasets, with significantly reduced MSE fluctuations. This not only boosts model accuracy but also provides more stable noise control.</div></div>","PeriodicalId":50367,"journal":{"name":"Information Fusion","volume":"116 ","pages":"Article 102796"},"PeriodicalIF":14.7,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142696901","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