Utility-driven pattern analysis is a fundamental method for analyzing noteworthy patterns with high utility for diverse quantitative transactional databases. Recently, various approaches have emerged to handle large, dynamic database environments more efficiently by reducing the number of data scans and pattern expansion operations with the pre-large concept. However, existing pre-large-based high utility pattern mining methods either fail to handle real-time transaction modifications or require additional data scans to validate candidate patterns. In this paper, we propose a novel efficient utility-driven pattern mining algorithm using the pre-large concept for transaction modifications. Our method incorporates a single-scan-based framework through the management of actual utility values and discovers high utility patterns without candidate generation for efficient utility-driven dynamic data analysis in the modification environment. We compared the performance of the proposed method with state-of-the-art methods through extensive performance evaluation utilizing real and synthetic datasets. According to the evaluation results and a case study, the suggested method performs a minimum of 1.5 times faster than state-of-the-art methods alongside minimal compromise in memory, and it scaled well with increases in database size. Further statistical analyses indicate that the proposed method reduces the pattern search space compared to the previous method while delivering a complete set of accurate results without loss.
{"title":"Utility-Driven Data Analytics Algorithm for Transaction Modifications Using Pre-Large Concept With Single Database Scan","authors":"Unil Yun;Hanju Kim;Myungha Cho;Taewoong Ryu;Seungwan Park;Doyoon Kim;Doyoung Kim;Chanhee Lee;Witold Pedrycz","doi":"10.1109/TBDATA.2025.3556615","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3556615","url":null,"abstract":"Utility-driven pattern analysis is a fundamental method for analyzing noteworthy patterns with high utility for diverse quantitative transactional databases. Recently, various approaches have emerged to handle large, dynamic database environments more efficiently by reducing the number of data scans and pattern expansion operations with the pre-large concept. However, existing pre-large-based high utility pattern mining methods either fail to handle real-time transaction modifications or require additional data scans to validate candidate patterns. In this paper, we propose a novel efficient utility-driven pattern mining algorithm using the pre-large concept for transaction modifications. Our method incorporates a single-scan-based framework through the management of actual utility values and discovers high utility patterns without candidate generation for efficient utility-driven dynamic data analysis in the modification environment. We compared the performance of the proposed method with state-of-the-art methods through extensive performance evaluation utilizing real and synthetic datasets. According to the evaluation results and a case study, the suggested method performs a minimum of 1.5 times faster than state-of-the-art methods alongside minimal compromise in memory, and it scaled well with increases in database size. Further statistical analyses indicate that the proposed method reduces the pattern search space compared to the previous method while delivering a complete set of accurate results without loss.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2792-2808"},"PeriodicalIF":5.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934407","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-04-01DOI: 10.1109/TBDATA.2025.3556637
Weihua Xu;Di Jiang
Concept-cognitive learning (CCL) is the process of enabling machines to simulate the concept learning of the human brain. Existing CCL models focus on formal context while neglecting the importance of skill context. Furthermore, CCL models, which solely focus on positive information, restrict the learning capacity by neglecting negative information, and greatly impeding the acquisition of knowledge. To overcome these issues, we proposes a novel concept-cognitive learning model oriented to three-way concept for knowledge acquisition. First, this paper explains and investigates the relationship between skills and knowledge based on the three-way concept and its properties. Then, in order to simultaneously consider positive and negative information, describe more detailed information, learn more skills, and acquire accurate knowledge, a three-way information granule is described from the perspective of cognitive learning. Then, a transformation method is proposed to transform between different three-way information granules, allowing for the transformation of arbitrary three-way information granule into necessary, sufficient, sufficient and necessary three-way information granules. Finally, algorithm corresponding to the transformation method is designed, and subsequently tested across diverse UCI datasets. The experimental outcomes affirm the effectiveness and excellence of the suggested model and algorithm.
{"title":"A Novel Concept-Cognitive Learning Model Oriented to Three-Way Concept for Knowledge Acquisition","authors":"Weihua Xu;Di Jiang","doi":"10.1109/TBDATA.2025.3556637","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3556637","url":null,"abstract":"Concept-cognitive learning (CCL) is the process of enabling machines to simulate the concept learning of the human brain. Existing CCL models focus on formal context while neglecting the importance of skill context. Furthermore, CCL models, which solely focus on positive information, restrict the learning capacity by neglecting negative information, and greatly impeding the acquisition of knowledge. To overcome these issues, we proposes a novel concept-cognitive learning model oriented to three-way concept for knowledge acquisition. First, this paper explains and investigates the relationship between skills and knowledge based on the three-way concept and its properties. Then, in order to simultaneously consider positive and negative information, describe more detailed information, learn more skills, and acquire accurate knowledge, a three-way information granule is described from the perspective of cognitive learning. Then, a transformation method is proposed to transform between different three-way information granules, allowing for the transformation of arbitrary three-way information granule into necessary, sufficient, sufficient and necessary three-way information granules. Finally, algorithm corresponding to the transformation method is designed, and subsequently tested across diverse UCI datasets. The experimental outcomes affirm the effectiveness and excellence of the suggested model and algorithm.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2779-2791"},"PeriodicalIF":5.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The development of a robust and adaptive deep learning technique for the diagnosis of pneumonia and the assessment of its severity was a major challenge. Indeed, both chest X-rays (CXR) and CT scans have been widely studied for the diagnosis, detection and quantification of pneumonia. In this paper, a novel approach (PViTGAtt-IP) based on a parallel array of vision transformers is presented, in which the input image is divided into regions of interest. Each region is fed into an individual model and the collective output gives the severity score. Three parallel architectures were also derived and tested. The proposed models were subjected to rigorous tests on two different datasets: RALO CXRs and Per COVID-19 CT scans. The experimental results showed that the proposed models exhibited high performance in accurately predicting scores for both datasets. In particular, the parallel transformers with multi-gate attention proved to be the best performing model. Furthermore, a comparative analysis using state-of-the-art methods showed that our proposed approach consistently achieved competitive or even better performance in terms of the Mean Absolute Error (MAE) and the Pearson Correlation Coefficient (PC). This emphasizes the effectiveness and superiority of our models in the context of diagnosing and assessing the severity of pneumonia.
{"title":"PViTGAtt-IP: Severity Quantification of Lung Infections in Chest X-Rays and CT Scans via Parallel and Cross-Attended Encoders","authors":"Bouthaina Slika;Fadi Dornaika;Fares Bougourzi;Karim Hammoudi","doi":"10.1109/TBDATA.2025.3556612","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3556612","url":null,"abstract":"The development of a robust and adaptive deep learning technique for the diagnosis of pneumonia and the assessment of its severity was a major challenge. Indeed, both chest X-rays (CXR) and CT scans have been widely studied for the diagnosis, detection and quantification of pneumonia. In this paper, a novel approach (PViTGAtt-IP) based on a parallel array of vision transformers is presented, in which the input image is divided into regions of interest. Each region is fed into an individual model and the collective output gives the severity score. Three parallel architectures were also derived and tested. The proposed models were subjected to rigorous tests on two different datasets: RALO CXRs and Per COVID-19 CT scans. The experimental results showed that the proposed models exhibited high performance in accurately predicting scores for both datasets. In particular, the parallel transformers with multi-gate attention proved to be the best performing model. Furthermore, a comparative analysis using state-of-the-art methods showed that our proposed approach consistently achieved competitive or even better performance in terms of the Mean Absolute Error (MAE) and the Pearson Correlation Coefficient (PC). This emphasizes the effectiveness and superiority of our models in the context of diagnosing and assessing the severity of pneumonia.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2736-2748"},"PeriodicalIF":5.7,"publicationDate":"2025-03-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934547","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-31DOI: 10.1109/TBDATA.2025.3556639
Ling Hu;Tongqing Zhou;Zhihuang Liu;Fang Liu;Zhiping Cai
Sequential data learning is vital to harnessing the encompassed rich knowledge for diverse downstream tasks, particularly in healthcare (e.g., disease prediction). Considering data sensitiveness, privacy-preserving learning methods, based on federated learning (FL) and split learning (SL), have been widely investigated. Yet, this work identifies, for the first time, existing methods overlook that sequential data are generated by different patients at different times and stored in different hospitals, failing to learn the sequential correlations between different temporal segments. To fill this void, a novel distributed learning framework STSL is proposed by training a model on the segments in order. Considering that patients have different visit sequences, STSL first implements privacy-preserving visit ordering based on a secure multi-party computation mechanism. Then batch scheduling participates patients with similar visit (sub-)sequences into the same training batch, facilitating subsequent split learning on batches. The scheduling process is formulated as an NP-hard optimization problem on balancing learning loss and efficiency and a greedy-based solution is presented. Theoretical analysis proves the privacy preservation property of STSL. Experimental results on real-world eICU data show its superior performance compared with FL and SL ($5% sim 28%$ better accuracy) and effectiveness (a remarkable 75% reduction in communication costs).
{"title":"Split Learning on Segmented Healthcare Data","authors":"Ling Hu;Tongqing Zhou;Zhihuang Liu;Fang Liu;Zhiping Cai","doi":"10.1109/TBDATA.2025.3556639","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3556639","url":null,"abstract":"Sequential data learning is vital to harnessing the encompassed rich knowledge for diverse downstream tasks, particularly in healthcare (e.g., disease prediction). Considering data sensitiveness, privacy-preserving learning methods, based on federated learning (FL) and split learning (SL), have been widely investigated. Yet, this work identifies, for the first time, existing methods overlook that sequential data are generated by different patients at different times and stored in different hospitals, failing to learn the sequential correlations between different temporal segments. To fill this void, a novel distributed learning framework <monospace>STSL</monospace> is proposed by training a model on the segments in order. Considering that patients have different visit sequences, <monospace>STSL</monospace> first implements privacy-preserving visit ordering based on a secure multi-party computation mechanism. Then batch scheduling participates patients with similar visit (sub-)sequences into the same training batch, facilitating subsequent split learning on batches. The scheduling process is formulated as an NP-hard optimization problem on balancing learning loss and efficiency and a greedy-based solution is presented. Theoretical analysis proves the privacy preservation property of <monospace>STSL</monospace>. Experimental results on real-world eICU data show its superior performance compared with FL and SL (<inline-formula><tex-math>$5% sim 28%$</tex-math></inline-formula> better accuracy) and effectiveness (a remarkable 75% reduction in communication costs).","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2749-2763"},"PeriodicalIF":5.7,"publicationDate":"2025-03-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934398","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-26DOI: 10.1109/TBDATA.2025.3552344
Weidong Zhao;Xiaoyu Wang;Liqing Qiu
Chinese Spelling Correction (CSC) is designed to detect and correct spelling errors that occur in Chinese text. In real life, most keyboard input scenarios use the pinyin input method. Researching spelling errors in this scenario is practical and valuable. However, there is currently no research that has truly proposed a model suitable for this scenario. Considering this concern, this paper proposes a model IPCK-IME, which incorporates confused phraseological knowledge based on the pinyin input method. The model integrates its own phonetic features with external similarity knowledge to guide the model to output more correct characters. Furthermore, to mitigate the influence of spelling errors on the semantics of sentences, a Gaussian bias is introduced into the self-attention network of the model. This approach aims to reduces the focus on typos and improve attention to local context. Empirical evidence indicates that our method surpasses existing models in correcting spelling errors generated by the pinyin input method. And, it is more appropriate for correcting Chinese spelling errors in real input scenarios.
{"title":"Incorporating Confused Phraseological Knowledge Based on Pinyin Input Method for Chinese Spelling Correction","authors":"Weidong Zhao;Xiaoyu Wang;Liqing Qiu","doi":"10.1109/TBDATA.2025.3552344","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3552344","url":null,"abstract":"Chinese Spelling Correction (CSC) is designed to detect and correct spelling errors that occur in Chinese text. In real life, most keyboard input scenarios use the pinyin input method. Researching spelling errors in this scenario is practical and valuable. However, there is currently no research that has truly proposed a model suitable for this scenario. Considering this concern, this paper proposes a model IPCK-IME, which incorporates confused phraseological knowledge based on the pinyin input method. The model integrates its own phonetic features with external similarity knowledge to guide the model to output more correct characters. Furthermore, to mitigate the influence of spelling errors on the semantics of sentences, a Gaussian bias is introduced into the self-attention network of the model. This approach aims to reduces the focus on typos and improve attention to local context. Empirical evidence indicates that our method surpasses existing models in correcting spelling errors generated by the pinyin input method. And, it is more appropriate for correcting Chinese spelling errors in real input scenarios.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2724-2735"},"PeriodicalIF":5.7,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934449","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Multivariate time series forecasting has extensive applications in urban computing, such as financial analysis, weather prediction, and traffic forecasting. Using graph structures to model the complex correlations among variables in time series, and leveraging graph neural networks and recurrent neural networks for temporal aggregation and spatial propagation stage, has shown promise. However, traditional methods’ graph structure node learning and discrete neural architecture are not sensitive to issues such as sudden changes, time variance, and irregular sampling often found in real-world data. To address these challenges, we propose a method called Adaptive Graph structure Learning neural Rough Differential Equations (AGLRDE). Specifically, we combine dynamic and static graph structure learning to adaptively generate a more robust graph representation. Then we employ a spatio-temporal encoder-decoder based on Neural Rough Differential Equations (Neural RDE) to model spatio-temporal dependencies. Additionally, we introduce a path reconstruction loss to constrain the path generation stage. We conduct experiments on six benchmark datasets, demonstrating that our proposed method outperforms existing state-of-the-art methods. The results show that AGLRDE effectively handles aforementioned challenges, significantly improving the accuracy of multivariate time series forecasting.
{"title":"Adaptive Graph Structure Learning Neural Rough Differential Equations for Multivariate Time Series Forecasting","authors":"Yuming Su;Tinghuai Ma;Huan Rong;Mohamed Magdy Abdel Wahab","doi":"10.1109/TBDATA.2025.3552334","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3552334","url":null,"abstract":"Multivariate time series forecasting has extensive applications in urban computing, such as financial analysis, weather prediction, and traffic forecasting. Using graph structures to model the complex correlations among variables in time series, and leveraging graph neural networks and recurrent neural networks for temporal aggregation and spatial propagation stage, has shown promise. However, traditional methods’ graph structure node learning and discrete neural architecture are not sensitive to issues such as sudden changes, time variance, and irregular sampling often found in real-world data. To address these challenges, we propose a method called <underline>A</u>daptive <underline>G</u>raph structure <underline>L</u>earning neural <underline>R</u>ough <underline>D</u>ifferential <underline>E</u>quations (AGLRDE). Specifically, we combine dynamic and static graph structure learning to adaptively generate a more robust graph representation. Then we employ a spatio-temporal encoder-decoder based on Neural Rough Differential Equations (Neural RDE) to model spatio-temporal dependencies. Additionally, we introduce a path reconstruction loss to constrain the path generation stage. We conduct experiments on six benchmark datasets, demonstrating that our proposed method outperforms existing state-of-the-art methods. The results show that AGLRDE effectively handles aforementioned challenges, significantly improving the accuracy of multivariate time series forecasting.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2710-2723"},"PeriodicalIF":5.7,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934381","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-18DOI: 10.1109/TBDATA.2025.3552341
Lei Zhang;Mingren Ke;Likang Wu;Wuji Zhang;Zihao Chen;Hongke Zhao
Recently, numerous studies have integrated self-supervised contrastive learning with Graph Convolutional Networks (GCNs) to address the data sparsity and popularity bias to enhance recommendation performance. While such studies have made breakthroughs in accuracy metric, they often neglect non-accuracy objectives such as diversity, novelty and percentage of long-tail items, which greatly reduces the user experience in real-world applications. To this end, we propose a novel graph collaborative filtering model named Multi-Objective Graph Contrastive Learning for recommendation (MOGCL), designed to provide more comprehensive recommendations by considering multiple objectives. Specifically, MOGCL comprises three modules: a multi-objective embedding generation module, an embedding fusion module and a transfer learning module. In the multi-objective embedding generation module, we employ two GCN encoders with different goal orientations to generate node embeddings targeting accuracy and non-accuracy objectives, respectively. These embeddings are then effectively fused with complementary weights in the embedding fusion module. In the transfer learning module, we suggest an auxiliary self-supervised task to promote the maximization of the mutual information of the two sets of embeddings, so that the obtained final embeddings are more stable and comprehensive. The experimental results on three real-world datasets show that MOGCL achieves optimal trade-offs between multiple objectives comparing to the state-of-the-arts.
{"title":"Multi-Objective Graph Contrastive Learning for Recommendation","authors":"Lei Zhang;Mingren Ke;Likang Wu;Wuji Zhang;Zihao Chen;Hongke Zhao","doi":"10.1109/TBDATA.2025.3552341","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3552341","url":null,"abstract":"Recently, numerous studies have integrated self-supervised contrastive learning with Graph Convolutional Networks (GCNs) to address the data sparsity and popularity bias to enhance recommendation performance. While such studies have made breakthroughs in accuracy metric, they often neglect non-accuracy objectives such as diversity, novelty and percentage of long-tail items, which greatly reduces the user experience in real-world applications. To this end, we propose a novel graph collaborative filtering model named Multi-Objective Graph Contrastive Learning for recommendation (MOGCL), designed to provide more comprehensive recommendations by considering multiple objectives. Specifically, MOGCL comprises three modules: a multi-objective embedding generation module, an embedding fusion module and a transfer learning module. In the multi-objective embedding generation module, we employ two GCN encoders with different goal orientations to generate node embeddings targeting accuracy and non-accuracy objectives, respectively. These embeddings are then effectively fused with complementary weights in the embedding fusion module. In the transfer learning module, we suggest an auxiliary self-supervised task to promote the maximization of the mutual information of the two sets of embeddings, so that the obtained final embeddings are more stable and comprehensive. The experimental results on three real-world datasets show that MOGCL achieves optimal trade-offs between multiple objectives comparing to the state-of-the-arts.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2696-2709"},"PeriodicalIF":5.7,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Multiplex Graph Contrastive Learning (MGCL) has attracted significant attention. However, existing MGCL methods often struggle with suboptimal graph structures and fail to fully capture intricate interdependencies across multiplex views. To address these issues, we propose a novel self-supervised framework, Multiplex Graph Refinement with progressive fusion (MGRefine), for multiplex graph contrastive representation learning. Specifically, MGRefine introduces a multi-view learning module to extract a structural guidance matrix by exploring the underlying relationships between nodes. Then, a progressive fusion module is employed to progressively enhance and fuse representations from different views, capturing and leveraging nuanced interdependencies and comprehensive information across the multiplex graphs. The fused representation is then used to construct a consensus guidance matrix. A self-enhanced refinement module continuously refines the multiplex graphs using these guidance matrices while providing effective supervision signals. MGRefine achieves mutual reinforcement between graph structures and representations, ensuring continuous optimization of the model throughout the learning process in a self-enhanced manner. Extensive experiments demonstrate that MGRefine outperforms state-of-the-art methods and also verify the effectiveness of MGRefine across various downstream tasks on several benchmark datasets.
多元图对比学习（MGCL）已经引起了广泛的关注。然而，现有的MGCL方法经常与次优图结构作斗争，并且不能完全捕获多路视图之间复杂的相互依赖关系。为了解决这些问题，我们提出了一种新的自监督框架，multi - plex Graph Refinement with progressive fusion (MGRefine)，用于多路图对比表示学习。具体来说，MGRefine引入了一个多视图学习模块，通过探索节点之间的潜在关系来提取结构指导矩阵。然后，采用渐进融合模块逐步增强和融合来自不同视图的表示，捕获和利用多路图中细微的相互依赖关系和综合信息。然后使用融合表示构造一致指导矩阵。自增强的细化模块利用这些引导矩阵不断地细化多路图，同时提供有效的监督信号。MGRefine实现了图结构和表示之间的相互强化，以自我增强的方式确保模型在整个学习过程中不断优化。大量的实验表明，MGRefine优于最先进的方法，并且在几个基准数据集上验证了MGRefine在各种下游任务中的有效性。
{"title":"Self-Guided Graph Refinement With Progressive Fusion for Multiplex Graph Contrastive Representation Learning","authors":"Qi Dai;Yu Gu;Xiaofeng Zhu;Xiaohua Li;Fangfang Li;Ge Yu","doi":"10.1109/TBDATA.2025.3552331","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3552331","url":null,"abstract":"Multiplex Graph Contrastive Learning (MGCL) has attracted significant attention. However, existing MGCL methods often struggle with suboptimal graph structures and fail to fully capture intricate interdependencies across multiplex views. To address these issues, we propose a novel self-supervised framework, Multiplex Graph Refinement with progressive fusion (MGRefine), for multiplex graph contrastive representation learning. Specifically, MGRefine introduces a multi-view learning module to extract a structural guidance matrix by exploring the underlying relationships between nodes. Then, a progressive fusion module is employed to progressively enhance and fuse representations from different views, capturing and leveraging nuanced interdependencies and comprehensive information across the multiplex graphs. The fused representation is then used to construct a consensus guidance matrix. A self-enhanced refinement module continuously refines the multiplex graphs using these guidance matrices while providing effective supervision signals. MGRefine achieves mutual reinforcement between graph structures and representations, ensuring continuous optimization of the model throughout the learning process in a self-enhanced manner. Extensive experiments demonstrate that MGRefine outperforms state-of-the-art methods and also verify the effectiveness of MGRefine across various downstream tasks on several benchmark datasets.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2669-2680"},"PeriodicalIF":5.7,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934433","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-18DOI: 10.1109/TBDATA.2025.3552340
Li Li;Junjun Si;Jinna Lv;Junting Lu;Jianyu Zhang;Shuaifu Dai
Computing trajectory similarity is a fundamental task in trajectory analysis. Traditional heuristic methods suffer from quadratic computational complexity, which limits their scalability to large datasets. Recently, Trajectory Representation Learning (TRL) has been extensively studied to address this limitation. However, most existing TRL algorithms face two key challenges. First, they prioritize spatial similarity while neglecting the intricate spatio-temporal dynamics of trajectories, particularly temporal regularities. Second, these methods are often constrained by predefined single spatial or temporal scales, which can significantly impact performance, since the measurement of trajectory similarity depends on spatial and temporal resolution. To address these issues, we propose MSST, a Multi-Scale Self-supervised Trajectory Representation Learning framework. MSST simultaneously processes spatial and temporal information by generating 3D spatial-temporal tokens, thereby capturing spatio-temporal characteristics of trajectories more effectively. Further, MSST explore the multi-scale characteristics of trajectories. Finally, self-supervised contrastive learning is employed to enhance the consistency between the trajectory representations from different views. Experimental results on three real-world datasets for similarity trajectory computation provide insight into the design properties of our approach and demonstrate the superiority of our approach over existing TRL methods. MSST significantly surpasses all state-of-the-art competitors in terms of effectiveness, efficiency, and robustness. We explore the multi-scale characteristics of trajectories. To the best of our knowledge, this is the first effort in the TRL literature. Compared to previous TRL research, the proposed method can balance the noise and the details of trajectories, enabling a more comprehensive analysis by accounting for the variability inherent in trajectory data across different scales.
{"title":"MSST: Multi-Scale Spatial-Temporal Representation Learning for Trajectory Similarity Computation","authors":"Li Li;Junjun Si;Jinna Lv;Junting Lu;Jianyu Zhang;Shuaifu Dai","doi":"10.1109/TBDATA.2025.3552340","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3552340","url":null,"abstract":"Computing trajectory similarity is a fundamental task in trajectory analysis. Traditional heuristic methods suffer from quadratic computational complexity, which limits their scalability to large datasets. Recently, Trajectory Representation Learning (TRL) has been extensively studied to address this limitation. However, most existing TRL algorithms face two key challenges. First, they prioritize spatial similarity while neglecting the intricate spatio-temporal dynamics of trajectories, particularly temporal regularities. Second, these methods are often constrained by predefined single spatial or temporal scales, which can significantly impact performance, since the measurement of trajectory similarity depends on spatial and temporal resolution. To address these issues, we propose MSST, a Multi-Scale Self-supervised Trajectory Representation Learning framework. MSST simultaneously processes spatial and temporal information by generating 3D spatial-temporal tokens, thereby capturing spatio-temporal characteristics of trajectories more effectively. Further, MSST explore the multi-scale characteristics of trajectories. Finally, self-supervised contrastive learning is employed to enhance the consistency between the trajectory representations from different views. Experimental results on three real-world datasets for similarity trajectory computation provide insight into the design properties of our approach and demonstrate the superiority of our approach over existing TRL methods. MSST significantly surpasses all state-of-the-art competitors in terms of effectiveness, efficiency, and robustness. We explore the multi-scale characteristics of trajectories. To the best of our knowledge, this is the first effort in the TRL literature. Compared to previous TRL research, the proposed method can balance the noise and the details of trajectories, enabling a more comprehensive analysis by accounting for the variability inherent in trajectory data across different scales.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2657-2668"},"PeriodicalIF":5.7,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934392","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-18DOI: 10.1109/TBDATA.2025.3552342
Linna Zhao;Jianqiang Li;Li Li;Xi Xu
Glaucoma is a chronic and irreversible eye disease. Early detection and treatment can effectively prevent severe consequences. Deep transfer learning is widely used in fundus imaging analysis to remedy the shortage of training data of glaucoma. The model trained on the source domain may struggle to predict glaucoma in the target domain due to distribution differences. Several limitations cannot be ignored: (1) Image matching: enhancing global and local image consistency through bidirectional matching; (2) Hierarchical transfer: developing a strategy for transferring different hierarchical features. To this end, we propose a novel Matched Hierarchical Transfer Network (MHT-Net) to achieve automatic glaucoma detection. We initially create a fundus structure detector to match global and local images using intermediate layers of a pre-trained diagnostic model with source domain data. Next, a hierarchical transfer network is implemented, sharing parameters for general features and using a domain discriminator for specific features. By integrating adversarial and classification losses, the model acquires domain-invariant features, facilitating precise and seamless transfer of fundus information from source to target domains. Extensive experiments demonstrate the effectiveness of our proposed method, outperforming existing glaucoma detection methods. These advantages endow our algorithm as a promising efficient assisted tool in the glaucoma screening.
{"title":"MHT-Net: A Matching-Based Hierarchical Transfer Network for Glaucoma Detection From Fundus Images","authors":"Linna Zhao;Jianqiang Li;Li Li;Xi Xu","doi":"10.1109/TBDATA.2025.3552342","DOIUrl":"https://doi.org/10.1109/TBDATA.2025.3552342","url":null,"abstract":"Glaucoma is a chronic and irreversible eye disease. Early detection and treatment can effectively prevent severe consequences. Deep transfer learning is widely used in fundus imaging analysis to remedy the shortage of training data of glaucoma. The model trained on the source domain may struggle to predict glaucoma in the target domain due to distribution differences. Several limitations cannot be ignored: (1) Image matching: enhancing global and local image consistency through bidirectional matching; (2) Hierarchical transfer: developing a strategy for transferring different hierarchical features. To this end, we propose a novel Matched Hierarchical Transfer Network (MHT-Net) to achieve automatic glaucoma detection. We initially create a fundus structure detector to match global and local images using intermediate layers of a pre-trained diagnostic model with source domain data. Next, a hierarchical transfer network is implemented, sharing parameters for general features and using a domain discriminator for specific features. By integrating adversarial and classification losses, the model acquires domain-invariant features, facilitating precise and seamless transfer of fundus information from source to target domains. Extensive experiments demonstrate the effectiveness of our proposed method, outperforming existing glaucoma detection methods. These advantages endow our algorithm as a promising efficient assisted tool in the glaucoma screening.","PeriodicalId":13106,"journal":{"name":"IEEE Transactions on Big Data","volume":"11 5","pages":"2681-2695"},"PeriodicalIF":5.7,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144934486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：