Applied Intelligence最新文献_第6页

A gradient inversion attack defense method based on data augmentation 一种基于数据增强的梯度反转攻击防御方法

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-09 DOI: 10.1007/s10489-025-06533-y

Yingge Li, Xianlin Wu, Yuwen Chen, Haiyang Yu, Zhen Yang

{"title":"A gradient inversion attack defense method based on data augmentation","authors":"Yingge Li, Xianlin Wu, Yuwen Chen, Haiyang Yu, Zhen Yang","doi":"10.1007/s10489-025-06533-y","DOIUrl":"10.1007/s10489-025-06533-y","url":null,"abstract":"<div><p>The gradient inversion attack presents a significant threat to the data privacy in federated learning, enabling malicious adversaries to reconstruct private training data from gradients. Among the various protection strategies, data augmentation-based approaches have emerged as particularly promising. These methods can be seamlessly incorporated into existing federated learning frameworks, offering both efficiency and minimal impact on model accuracy. In this paper, we propose a novel data protection technique that leverages data augmentation methods, specifically CutMix and SaliencyMix. These techniques work by mixing images, which allows for more efficient utilization of training pixels. This, in turn, aids the model in learning more robust and meaningful feature representations, thereby enhancing both the model performance and its resilience to adversarial attacks. To further strengthen data privacy, we integrate these data augmentation methods with data pruning techniques. Our empirical results demonstrate that the proposed approach not only improves the accuracy of federated learning models but also reduces the quality of reconstructed images, offering a higher level of data privacy protection.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145011827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

PC-UNet: a pure convolutional UNet with channel shuffle average for medical image segmentation PC-UNet：用于医学图像分割的具有信道洗牌平均的纯卷积UNet

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-09 DOI: 10.1007/s10489-025-06887-3

Wei Liu, Qian Dong, Shiren Li, Cong Wang, Yongliang Xiong, Guangguang Yang

{"title":"PC-UNet: a pure convolutional UNet with channel shuffle average for medical image segmentation","authors":"Wei Liu, Qian Dong, Shiren Li, Cong Wang, Yongliang Xiong, Guangguang Yang","doi":"10.1007/s10489-025-06887-3","DOIUrl":"10.1007/s10489-025-06887-3","url":null,"abstract":"<div><p>In this study, a pure convolutional UNet with channel shuffle average, abbreviated as PC-UNet, has been proposed for medical image segmentation. Notably, the proposed PC-UNet is suitable for extracting context features, which is useful for model improvement. PC-UNet operates as an encoder-decoder network, where both the encoder and decoder are stacked with the proposed Pure Convolution (PC) modules. The PC module, containing a Channel Shuffle Average (CSA) component, is efficient in capturing context features without significant computational overhead. The CSA component transfers feature information from the channel dimension to the spatial dimension, enabling efficient computation. The effectiveness of the proposed PC-UNet has been rigorously validated on four widely used datasets, which are ISIC 2018, BUSI, GlaS, and Kvasir-SEG. Experimental results demonstrate that PC-UNet yields outstanding performance without imposing a significant computational load or increasing floating-point operations (FLOPs). When compared with eight mainstream models across all datasets, PC-UNet achieves the highest scores in both Dice and IoU metrics. The source code is available at: https://github.com/lwwant2sleep/PC-UNet.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145011851","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Using composite attribute similarity multi-graph convolutional network for recommendation 采用复合属性相似度多图卷积网络进行推荐

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-08 DOI: 10.1007/s10489-025-06840-4

Weichao He, Yi Zhu, Mei Song, Yuheng Su, Guosheng Hao

{"title":"Using composite attribute similarity multi-graph convolutional network for recommendation","authors":"Weichao He, Yi Zhu, Mei Song, Yuheng Su, Guosheng Hao","doi":"10.1007/s10489-025-06840-4","DOIUrl":"10.1007/s10489-025-06840-4","url":null,"abstract":"<div><p>Graph Convolutional Networks (GCNs) are frequently utilized and havel a significant role in recommender systems. This is attributed to their ability to capture signals of collaboration between higher-order neighbors using graph structures. GCN-based recommendation models have been greatly improved in improving recommendation performance, but continue to face serious data sparsity problems. Data sparsity can be effectively alleviated by introducing attribute information. However, current GCN-based models face challenges in effectively handling the diverse attribute information of users and items and capturing the complex relationships among users, items, and attributes. With the purpose of addressing aforementioned problems, this research proposes a Using Composite Attribute Similarity Multi-Graph Convolutional Network (UCASM-GCN) for recommendation. In concrete terms, an attribute fusion strategy based on the attention mechanism is first utilized to construct the composite attributes of users or items. Then, the user-user graph and the item-item graph are constructed using the composite attributes of nodes to mine the relationships between users and between items. Finally, two isomorphic graphs are injected into the user-item interaction graph as auxiliary information through a multi-graph convolution strategy to generate optimized embedding representations, which ultimately improve the recommendation performance. Extensive experiments on three public datasets demonstrate the effectiveness of the proposed UCASM-GCN, achieving performance gains of 2.48%, 8.20% and 5.52% over a competitive graph-based collaborative filtering model on the Movielens 100k, Movielens 1M and DoubanBook datasets, respectively.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145011759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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 self-attention multisource precipitation fusion model for improving long-sequence precipitation estimation accuracy 提高长序列降水估计精度的自关注多源降水融合模型

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-08 DOI: 10.1007/s10489-025-06832-4

Shaojie You, Xiaodan Zhang, Hongyu Wang, Chen Quan, Tong Zhao, Yongkun Zhang, Chang Liu

{"title":"A self-attention multisource precipitation fusion model for improving long-sequence precipitation estimation accuracy","authors":"Shaojie You, Xiaodan Zhang, Hongyu Wang, Chen Quan, Tong Zhao, Yongkun Zhang, Chang Liu","doi":"10.1007/s10489-025-06832-4","DOIUrl":"10.1007/s10489-025-06832-4","url":null,"abstract":"<div><p>Accurate precipitation estimation is essential in agricultural production, water resource management, and flood forecasting. However, high-precision precipitation data remain very hard to obtain due to the complex spatio-temporal distribution of precipitation. Most existing methods considering spatio-temporal correlations in precipitation rely on a convolutional neural network for spatial feature extraction. However, these methods are less efficient in capturing global spatial features due to the local receptive fields of convolutional operators. In this study, we designed a Self-LSTM cell structure capable of effectively capturing temporal and global spatial features. Based on this, a self-attention precipitation fusion model (SAPFM) is proposed. The results demonstrate that SAPFM outperforms basic models and the original precipitation products. SAPFM improves by 28.8% and 21.8% on the Kling-Gupta efficiency (KGE) and Correlation Coefficient (CC) compared to the best-performing precipitation product (GsMap), respectively. Additionally, SAPFM reduces the Root Mean Square Error (RMSE) by 12.5%.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10489-025-06832-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145007898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Cluster-infused low-rank subspace learning for robust multi-label classification 基于聚类注入的低秩子空间学习的鲁棒多标签分类

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-08 DOI: 10.1007/s10489-025-06837-z

Ziyue Zhu, Conghua Zhou, Shijie Sun, Emmanuel Ntaye, Xiang-Jun Shen, Zhifeng Liu

{"title":"Cluster-infused low-rank subspace learning for robust multi-label classification","authors":"Ziyue Zhu, Conghua Zhou, Shijie Sun, Emmanuel Ntaye, Xiang-Jun Shen, Zhifeng Liu","doi":"10.1007/s10489-025-06837-z","DOIUrl":"10.1007/s10489-025-06837-z","url":null,"abstract":"<div><p>Multi-label learning in high-dimensional spaces Suffers from the curse of dimensionality, noisy labels, and complex feature-label dependencies. Traditional deep learning solutions for multi-label classification employ multi-layer networks but overfit and generalize poorly owing to ineffective high-order data dependencies. In this paper, we introduce a cluster-infused low-rank subspace learning framework that integrates low-rank subspace learning with cluster infusion to solve these issues. Our model resolves sensitivity to noise, overfitting and poor generalization in high-dimensional data by using low-rank subspace representation decomposition of the classifier for dimension reduction and low-rank classifier for discriminative classification. To enhance robustness, we reconstruct each data sample as a Linear combination of its neighbours, infusing clustering-derived features into the model. These facilitate feature robustness via local correlations, thereby improving noise resilience and discriminative power. Extensive experiments on benchmark high-dimensional datasets, compared against state-of-the-art approaches, indicate that our approach significantly improves classification accuracy and robustness, making it a good solution for noisy, high-dimensional multi-label classification tasks. This effectiveness is evidenced across datasets of various scales, including a 3.04% improvement in Example-F1 over CNN-RNN on the smaller 20NG dataset and a significant 9.9% gain in Micro-F1 against RethinkNet on the large-scale NUS-WIDE dataset, highlighting DL-CS’s superiority for diverse multi-label classification tasks.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145007897","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Multi-3D pose tracking based on multi-view fusion feature correlation 基于多视角融合特征关联的多三维姿态跟踪

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-08 DOI: 10.1007/s10489-025-06774-x

Kai Chen, Yujie Huang, Xiaodong Zhao, Guoyu Fang, Ziyuan Wang

{"title":"Multi-3D pose tracking based on multi-view fusion feature correlation","authors":"Kai Chen, Yujie Huang, Xiaodong Zhao, Guoyu Fang, Ziyuan Wang","doi":"10.1007/s10489-025-06774-x","DOIUrl":"10.1007/s10489-025-06774-x","url":null,"abstract":"<div><p>Pedestrian 3D pose tracking in multi-view scenarios has extensive practical applications. However, existing methods often overlook the overall tracking accuracy of pedestrians, particularly the issues of missing and erroneous tracking caused by severe occlusions, disappearances, and reappearances. It further affects the accuracy of pose point association. To address these limitations, a two-stage method is proposed, involving tracking with an exceptionally low error rate, followed by obtaining higher precision 3D pose points. Firstly, a multi-object tracking model is introduced, which integrates feature association-validation-updating and employs dynamic thresholding strategy to achieve high-accuracy matching of multiple individuals in multi-view scenarios by computing similarity with feature pool templates. Additionally, a Gaussian Mixture-based feature pool updating model ensures the universality of stored features to solve the reappearance problem. Secondly, a pedestrian 2D pose detection and 3D pose reprojection method based on SMPL (Skinned Multi-Person Linear model) is proposed, which detects more complete pose points than OpenPose in complex scenes and better conforms to the distribution principles of human skeletal pose points. To validate the advancedness of the proposed method, the Shelf and Campus public datasets are re-annotated. Experimental results demonstrate the excellent performance of the proposed method in overall error control in complex environments, outperforming existing methods in multi-object tracking and pose point estimation accuracy and completeness.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145007963","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

EMERALD-O: efficient multi-agent reinforcement learning framework for optimised deep learning hyperparameter tuning and selection EMERALD-O：高效的多智能体强化学习框架，用于优化深度学习超参数调整和选择

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-08 DOI: 10.1007/s10489-025-06878-4

Akhila VH, Anu Mary Chacko, Ponnurangam Kumaraguru

{"title":"EMERALD-O: efficient multi-agent reinforcement learning framework for optimised deep learning hyperparameter tuning and selection","authors":"Akhila VH, Anu Mary Chacko, Ponnurangam Kumaraguru","doi":"10.1007/s10489-025-06878-4","DOIUrl":"10.1007/s10489-025-06878-4","url":null,"abstract":"<div><p>Traditional hyperparameter tuning methods, such as Bayesian Optimization and Grid Search, often prove computationally expensive and inefficient for complex deep learning architectures. This paper introduces the Multi-Agent Reinforcement Learning (MARL) framework EMERALD-O to optimize deep learning networks. The MARL-based approach utilizes two specialized agents, Agent1 focuses on data augmentation and Agent 2 on managing the learning rate and optimizer selection. The agents operate within an environment that simulates the model’s training dynamics and uses validation accuracy as the reward signal. Agent performance is enhanced through epsilon-greedy exploration and experience replay mechanisms. EMERALD-O performs favorably 88.59 % with improved classification accuracy and training efficiency. The framework exhibits adaptability to diverse dataset characteristics, underscoring scalability and robustness. The framework was validated on different models built for image classification problem on Efficientnet, VGG16 and VGG19. The results highlight the potential of reinforcement learning to fine-tune complex neural network architectures and suggest that MARL can serve as a powerful tool to improve the performance of deep learning models. EMERALD-O can contribute by advancing the frontier of deep neural optimization, demonstrating that reinforcement learning can fundamentally transform the model-tuning approach. This framework establishes a new paradigm for automated hyperparameter optimization and provides a systematic lens for analyzing the behavior of the deep learning model across various hyperparametric configurations. By quantifying model responsiveness to parameter variations, this approach enables deeper insights into architectural characteristics and performance dynamics, facilitating both the theoretical understanding and practical optimization of deep learning systems.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145011748","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

DMSE: An efficient malicious traffic detection model based on deep multi-stacking ensemble learning DMSE：一种基于深度多堆叠集成学习的高效恶意流量检测模型

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-06 DOI: 10.1007/s10489-025-06819-1

Saihua Cai, Yang Zhang, Yanghang Li, Yupeng Wang, Jiayao Li, Xiang Zhou

{"title":"DMSE: An efficient malicious traffic detection model based on deep multi-stacking ensemble learning","authors":"Saihua Cai, Yang Zhang, Yanghang Li, Yupeng Wang, Jiayao Li, Xiang Zhou","doi":"10.1007/s10489-025-06819-1","DOIUrl":"10.1007/s10489-025-06819-1","url":null,"abstract":"<div><p>In the context of increasing cyber threats, developing an efficient malicious traffic detection model to recognize the cyber attacks has become an urgent demand in the field of cyber security. This paper proposes an efficient malicious traffic detection model called DMSE based on deep multi-stacking ensemble learning, it is primarily consisted of feature representation module, base model detection module and multi-stacking ensemble learning module. In the feature representation phase, we propose a novel RGB image representation method, which hierarchically represents the global and local features of network traffic by allocating the information to three channels of RGB images. In the base model detection phase, we adopt five different deep learning models—CNN, TCN, LSTM, BiLSTM and BiTCN—as base models for the first-stage prediction. In the multi-stacking ensemble learning phase, we adopt the best-performing BiTCN from extensive experiments as the meta-learner to perform a second prediction using the results from the first stage, thereby obtaining the final detection result. Experiments conducted on USTC-TFC2016, CTU and ISAC datasets demonstrate that the proposed DMSE model significantly outperforms existing ensemble learning-based detection models in terms of accuracy, F1-score, false positive rate (FPR), true positive rate (TPR) and stability. The experimental results indicate that the proposed DMSE model can effectively identify and defend against network attacks, providing the new perspectives and technical support for maintaining a secure network environment.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144998533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

An unsupervised domain adaptation method for cross-domain deceptive reviews detection 一种跨域欺骗性评论检测的无监督域自适应方法

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-06 DOI: 10.1007/s10489-025-06825-3

Ning Cao, Shujuan Ji, Fuzhen Zhuang, Dickson K. W. Chiu, Yajie Guo, Maoguo Gong

{"title":"An unsupervised domain adaptation method for cross-domain deceptive reviews detection","authors":"Ning Cao, Shujuan Ji, Fuzhen Zhuang, Dickson K. W. Chiu, Yajie Guo, Maoguo Gong","doi":"10.1007/s10489-025-06825-3","DOIUrl":"10.1007/s10489-025-06825-3","url":null,"abstract":"<div><p>Deceptive reviews seriously affect the interests of consumers, honest sellers, and e-commerce platforms. As e-commerce platforms often involve multiple domains (i.e., different products or services), in-domain deceptive review detection models trained and tested on a specific dataset may not perform well on other domains. Moreover, obtaining annotated data for so many individual domains is unrealistic. Cross-domain deceptive review detection aims to leverage labeled source domain data to improve the model’s performance on unlabeled target domain data. However, existing cross-domain deceptive review detection methods require labels for target domain data or do not consider domain-specific information. To further advance research, this paper proposes an unsupervised domain adaptation method for detecting cross-domain deceptive reviews. First, we propose a multiple mask views generation method to enhance domain-specific information to obtain different mask views of reviews. Secondly, the BERT and mask attention mechanisms are used sequentially to obtain contextual representations of the mask views and the original view of reviews. Thirdly, to maintain the consistency between the mask views and the original view of reviews, we use the intra-domain Kullback-Leibler divergence to constrain their learning process. Moreover, we use inter-domain dynamic maximum mean discrepancy and conditional maximum mean discrepancy to reduce differences between the distribution of source and target domains. Three sets of experiments on two datasets show that our method is superior to the baselines. In particular, the impact of domain differences on domain adaptability is further analyzed according to the quantified metric named domain distance defined in this paper.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144998532","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Multi-view contrastive learning with Static attributes and Dynamic interests for Sequential Recommendation 基于静态属性和动态兴趣的序列推荐多视图对比学习

IF 3.5 2区计算机科学

Applied Intelligence Pub Date : 2025-09-06 DOI: 10.1007/s10489-025-06816-4

Mukun Chen, Jia Wu, Shirui Pan, Xiantao Cai, Bo Du, Wenbin Hu, Huiting Xu

{"title":"Multi-view contrastive learning with Static attributes and Dynamic interests for Sequential Recommendation","authors":"Mukun Chen, Jia Wu, Shirui Pan, Xiantao Cai, Bo Du, Wenbin Hu, Huiting Xu","doi":"10.1007/s10489-025-06816-4","DOIUrl":"10.1007/s10489-025-06816-4","url":null,"abstract":"<div><p>Sequential recommendation plays a critical role in preference prediction by capturing the temporal evolution of user behavior. However, a key challenge lies in effectively integrating static attributes, such as stable user traits and item properties, with dynamic interests, which reflect the users’ transient and evolving preferences during interactions with various items. Current approaches typically focus on static attributes or recent interactions, neglecting the nuanced interplay between long-term stability and short-term variability. Additionally, the disparate encoding strategies for various data structures—such as bipartite interaction graphs, heterogeneous knowledge graphs, and sequential data streams—lead to fragmented user and item representations, hindering the development of a unified framework and reducing the system’s ability to holistically model user preferences. To address these challenges, we propose the multi-view contrastive learning with <b>S</b>tatic attributes and Dynamic interests for Sequential Recommendation (SDSR), a novel framework that integrates static and dynamic characteristics to enhance recommendation systems. SDSR employs graph-based encoders to capture static user and item features, while a sequence encoder models temporal changes in user behavior. By leveraging contrastive learning, SDSR aligns representations across multiple data views—such as interaction graphs, knowledge graphs, and sequential data—creating a unified user-item model that bridges long-term preferences with short-term trends. It also ensures consistency across various representations, yielding a cohesive and robust framework for synthesizing multi-perspective data. Empirical evaluations on benchmark datasets demonstrate that SDSR significantly outperforms state-of-the-art models, validating its effectiveness in integrating multi-view data and capturing both static and dynamic user preferences.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 14","pages":""},"PeriodicalIF":3.5,"publicationDate":"2025-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144998535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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