Information Sciences最新文献

{"title":"Weighted dynamic network link prediction based on graph autoencoder","authors":"Peng Mei,&nbsp;Yuhong Zhao,&nbsp;Jingyu Wang,&nbsp;Yefei Liang","doi":"10.1016/j.ins.2025.122507","DOIUrl":"10.1016/j.ins.2025.122507","url":null,"abstract":"<div><div>With the development of deep learning, Graph Autoencoders (GAE) within unsupervised learning frameworks have been widely applied to representation learning in dynamic networks. However, existing methods typically assume that the node set remains fixed across all time slices and ignore edge weight information, which limits the ability to capture network dynamics and distinguish the strength of node relationships. To address these issues, this paper proposes a weighted dynamic network link prediction framework based on GAE, called GAE_GGLA. This framework introduces an alignment module that can handle non-fixed node sets to adapt to dynamic network environments. Additionally, the edge weight matrix is used as a bias term in the graph attention network to calculate attention coefficients, guiding the learning of node features and enhancing their representational capacity. Furthermore, the GAE encoder employs graph convolution network (GCN) and long short-term memory (LSTM) networks to capture, respectively, structural features and temporal evolution. The alignment module connects different node sets through adjacent time slices, ensuring the continuity and consistency of network information. Finally, the GAE decoder reconstructs the adjacency matrix of the original graph to achieve link prediction. Experiments conducted on five different datasets demonstrate that the proposed method significantly outperforms state-of-the-art baselines.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122507"},"PeriodicalIF":8.1,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144670485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predefined-time distributed constrained multi-conflicting objective optimization with nonlinear uncertainty over directed graph","authors":"He Jiang ,&nbsp;Junlong He ,&nbsp;Sen Chen ,&nbsp;Zhenhua Deng","doi":"10.1016/j.ins.2025.122512","DOIUrl":"10.1016/j.ins.2025.122512","url":null,"abstract":"<div><div>This paper addresses the constrained multi-conflicting objective optimization (MCOO) problem for multi-agent systems under strong nonlinear uncertainty over directed graph. Each agent is subject to multiple conflicting local objectives. To enable agents to autonomously seek the Pareto optimality of the MCOO problem, three distributed algorithms are developed. First, by utilizing the online updating weighted <span><math><msub><mrow><mi>L</mi></mrow><mrow><mi>p</mi></mrow></msub></math></span> preference index, the MCOO problem is reformulated into a single-objective optimization problem, and two essential parameters are determined by solving auxiliary optimization subproblems. Next, to actively eliminate and compensate for the impact of strong nonlinear uncertainty in three optimization problems, three reduced-order extended vector observers are utilized. By the three proposed algorithms employing time-based generator, state feedback, and disturbance compensation, all agents converge to an arbitrarily small neighborhood of the Pareto optimality within predefined time, although strong nonlinear uncertainty exists and the predefined time can be set arbitrarily. Furthermore, simulation example verifies the effectiveness of the proposed algorithms.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122512"},"PeriodicalIF":8.1,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144653830","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Towards a moving target defense based on stochastic games and honeypots","authors":"Di Li ,&nbsp;Shirui Tian ,&nbsp;Wenqiang Jin ,&nbsp;Jiwu Peng ,&nbsp;Mingxing Duan","doi":"10.1016/j.ins.2025.122488","DOIUrl":"10.1016/j.ins.2025.122488","url":null,"abstract":"<div><div>Honeypots, which serve as active defense mechanisms, have historically played pivotal roles in cyberspace offensive and defensive countermeasure scenarios. However, with the advancement of honeypot recognition technologies, their effectiveness in real-world network defense has gradually diminished. In response, moving target defense (MTD) has recently solidified its position as a proactive cybersecurity strategy and a critical research frontier. MTD leverages heterogeneous, redundant deployments of service resources and randomization techniques to disrupt attack methods. However, despite their advantages, MTD systems face challenges related to high resource consumption. To address these limitations, we propose a moving target defense based on stochastic games and honeypots (GH-MTD) framework. This framework consists of four key modules: traffic detection, gaming, MTD, and honeynet. Firstly, malicious traffic is identified through a deep learning-based detection method. Secondly, a zero-sum game model is constructed to capture the decision-making dynamics between defenders and attackers in the context of moving target defense. Subsequently, a cross-scenario adaptive MTD module is designed to route different types of traffic to corresponding virtual server groups. Finally, a honeypot module is implemented to capture and analyze the specific attack behaviors of malicious actors. By integrating honeynet probes with real services and employing attack behavior analysis alongside internet protocol (IP) address redirection techniques, the GH-MTD system achieves a defense response that is both cost efficient and highly effective. Empirical evaluation reveals a 5.5-fold enhancement in attack diversion probability through benchmarking with service-oriented MTD architectures, while the capture rate surpasses that of conventional honeypots by 3.4 times. Particularly against real attackers, GH-MTD exhibits 5.6 times more captured packets and extends the time consumed by attackers by 1.5 times over that of standalone honeypots. In our experiments, we evaluate the architecture's performance against various attack methods, including automated scripts, manual attacks, and assaults by high-level penetration testers. The results demonstrate that the GH-MTD architecture performs exceptionally well, particularly in mitigating and countering advanced, sophisticated attacks, thereby demonstrating its effectiveness in modern network defense strategies.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122488"},"PeriodicalIF":8.1,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144634420","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A doctor recommendation model based on multidimensional feature extraction of doctors and patients from online medical platform","authors":"Minghui Qian ,&nbsp;Mengchun Zhao ,&nbsp;Meng Pan ,&nbsp;Yuchen Pan ,&nbsp;Desheng Wu ,&nbsp;David L. Olson ,&nbsp;Weiping Ding","doi":"10.1016/j.ins.2025.122500","DOIUrl":"10.1016/j.ins.2025.122500","url":null,"abstract":"<div><div>To address the challenge of efficiently allocating limited medical resources in China, this study proposes a similarity-driven online doctor recommendation model (SimRec) to improve healthcare accessibility and resource utilization. The model was developed using object-oriented methods to analyze the current service mode of online consultation platforms, incorporating the actual needs of doctors and patients into its design. The framework consists of two layers: the object layer, which represents patient and doctor models abstractly, and the function layer, which implements recommendation technology. The function layer divides the process into two stages—department prediction and doctor-patient matching—to guide patients to appropriate departments, recommend suitable doctors, and allocate doctors based on patient needs. Tests on real-world data demonstrate that SimRec achieves better performance compared to baseline models in both department prediction and doctor-patient matching, indicating its effectiveness in optimizing medical resource allocation.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122500"},"PeriodicalIF":8.1,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144670480","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"RaCo: Reliable multi-augmentation complementary label learning","authors":"Yiwei You,&nbsp;Zan Chen,&nbsp;Meng Xu,&nbsp;Bo Wang","doi":"10.1016/j.ins.2025.122509","DOIUrl":"10.1016/j.ins.2025.122509","url":null,"abstract":"<div><div>Complementary Label Learning (CLL) represents a fundamental weakly-supervised learning paradigm where training instances are annotated with complementary labels - each indicating a class to which the instance does not belong. While existing approaches have attempted to incorporate advanced augmentation alignment techniques to overcome the challenges of this extremely weak supervision, they fail to address a critical aspect: How to effectively leverage information from multiple augmented views to guide model learning. In this paper, we present the first integration of Evidential Deep Learning (EDL) into CLL, introducing a principled framework to quantify and utilize view-specific uncertainty in CLL. Specifically, we propose a novel framework termed RaCo, i.e., <em><strong>R</strong>eliable Multi-<strong>a</strong>ugmentation <strong>Co</strong>mplementary Label Learning</em>, which leverages EDL to estimate second-order uncertainty for each augmented view and then calculate a stable metric, named augmented view certainty (AVC), to assess the quality of each view. Using the proposed AVC, RaCo formulates a reliable target distribution by combining the predictions of multiple augmented views and further employs Kullback-Leibler (KL) divergence to align the prediction probability of each augmented view with this reliable distribution. Extensive experiments on the benchmark datasets under various CLL settings validate the effectiveness and superiority of the proposed RaCo. Our code will be released upon acceptance.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"719 ","pages":"Article 122509"},"PeriodicalIF":8.1,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144631471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"OUTO-Miner: Detecting outlying occurrences in maximal frequent order-preserving patterns in time series","authors":"Youxi Wu ,&nbsp;Siqi Lou ,&nbsp;Yan Li ,&nbsp;Lei Guo ,&nbsp;Philippe Fournier-Viger ,&nbsp;Xindong Wu","doi":"10.1016/j.ins.2025.122497","DOIUrl":"10.1016/j.ins.2025.122497","url":null,"abstract":"<div><div>Order-preserving pattern (OPP) mining primarily focuses on the frequent trends of time series, and frequent OPPs have potential crucial value. However, the results of OPP mining ignore the significance of numerical values, especially in the field of outlier detection. In addition, OPP mining often generates redundant patterns, leading to high memory consumption or low operational efficiency in outlier detection. To address these problems, this paper focuses on detecting outlying occurrences (OUTO) in maximal frequent order-preserving patterns, which employs the dynamic time warping method to calculate the distance between two sub-time series, and proposes OUTO-Miner to detect outlying occurrences. In data preprocessing, a linear fitting method is employed to extract key points, compressing the data and preserving the main features. To mitigate the generation of redundant patterns, OUTO-Miner utilizes maximal frequent OPPs for outlier detection. To avoid excessive computations, OUTO-Miner uses the interquartile range method to identify sub-time series with a high probability of being an OUTO. To validate the performance of OUTO-Miner, 13 competitive algorithms and 17 datasets are selected. The results demonstrate that OUTO-Miner outperforms all competitive algorithms in terms of runtime, memory consumption, and outlier detection. All algorithms can be downloaded from <span><span>https://github.com/wuc567/Pattern-Mining/tree/master/OUTO-Miner</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122497"},"PeriodicalIF":8.1,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144634417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Achieving fair medical image segmentation in foundation models with adversarial visual prompt tuning","authors":"Yuqi Li ,&nbsp;Yanli Li ,&nbsp;Kai Zhang ,&nbsp;Fuyan Zhang ,&nbsp;Chuanguang Yang ,&nbsp;Zhongliang Guo ,&nbsp;Weiping Ding ,&nbsp;Tingwen Huang","doi":"10.1016/j.ins.2025.122501","DOIUrl":"10.1016/j.ins.2025.122501","url":null,"abstract":"<div><div>Recent advances in deep learning have significantly enhanced medical image analysis capabilities. Medical image segmentation, a critical application in this domain, enables precise delineation of anatomical structures and pathological regions, substantially supporting clinical decision-making. However, current segmentation methods primarily optimize for overall performance without considering disparities across demographic groups, raising important fairness concerns. To address this gap, we propose <strong><u>Adv</u></strong>ersarial <strong><u>V</u></strong>isual <strong><u>P</u></strong>rompt <strong><u>T</u></strong>uning (<strong>AdvVPT</strong>), a parameter-efficient approach that enhances fairness in foundation models for medical image segmentation. AdvVPT introduces trainable visual prompts within the image encoder while keeping the backbone frozen, requiring only 0.812M additional parameters. These prompts are optimized through adversarial training to absorb demographic-specific biased information from image embeddings, achieved by maximizing prediction errors for sensitive attributes and increasing embedding distances between visual prompts and image features. Experimental evaluation on the Harvard-FairSeg dataset demonstrates that AdvVPT achieves state-of-the-art fairness performance across multiple demographic attributes. For racial fairness, AdvVPT achieves an ES-Dice score of 0.8996 and an ES-IoU score of 0.8222 on optic cup segmentation, substantially outperforming existing methods. For gender fairness using the SAT backbone, AdvVPT achieves an ES-Dice of 0.9297 and ES-IoU of 0.8614, demonstrating both superior performance and improved balance between male and female subgroups.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122501"},"PeriodicalIF":8.1,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144653825","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Learning to mine all minimal evidences for unverified claims","authors":"Yanan Liu ,&nbsp;Hai Wan ,&nbsp;Jianfeng Du ,&nbsp;Yao Wang ,&nbsp;Kunxun Qi ,&nbsp;Weilin Luo","doi":"10.1016/j.ins.2025.122505","DOIUrl":"10.1016/j.ins.2025.122505","url":null,"abstract":"<div><div>Mining evidences is crucial in checking unverified claims. Most existing methods usually find a single evidence expressed by a set of sentences to verify a given claim. However, treating a set of sentences as unique evidence is insufficient or even misleading, <em>e.g.</em>, when it involves both supporting and refuting information. Besides, gathering evidence from different perspectives helps us better analyze and understand the claim. In this article, we suggest mining <em>all</em> irreducible evidences for supporting or refuting a claim, where we treat a minimal set of sentences in the given text corpus for either the support or the refutation as an irreducible point of view and call it a <em>minimal evidence</em>. We develop a neural-symbolic framework to mine all minimal evidences. It exploits a logical algorithm to compute all minimal evidences one by one, where every minimal evidence is computed through two neural models <em>scorer</em> and <em>reasoner</em> learnt from the annotated minimal evidences. Experimental results demonstrate that our framework is effective in finding multiple minimal evidences for both textual and structural claims. Furthermore, we investigate several implementation combinations for scorers and reasoners so as to seek the best practice on our framework.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122505"},"PeriodicalIF":8.1,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144686339","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploration of rough approximation operators with supervised justifiable granularity principle","authors":"Lei-Jun Li ,&nbsp;Mei-Zheng Li ,&nbsp;Ju-Sheng Mi","doi":"10.1016/j.ins.2025.122504","DOIUrl":"10.1016/j.ins.2025.122504","url":null,"abstract":"<div><div>Rough set theory is a representative granular computing model that has received widespread attention. Rough approximation operators (RAOs) serve as foundational components in rough set models, leveraging information granules to approximate abstract concepts. The justifiable granularity principle (JGP) is one of the fundamentals in granular computing, and has achieved great success in the design and evaluation of information granules. Within this context, this study investigates RAOs based on the JGP in classification learning. First, the limitations of two types of popular RAOs, namely probabilistic and fuzzy RAOs, are analyzed from a classification learning perspective. It is concluded that different samples lack discrimination w.r.t. the decision classes in these RAOs. Subsequently, the supervised JGP (SJGP) is proposed. The relative coverage and relative specificity of information granules are formulated w.r.t. the decision classes. These are integrated into existing RAOs to address the challenges. Finally, a new type of reduct is introduced, and a unified framework for heuristic algorithms is also developed correspondingly. The proposed RAOs are applied to attribute reduction. Experimental results demonstrate the reasonableness and superiority of integrating SJGP into RAOs.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"720 ","pages":"Article 122504"},"PeriodicalIF":8.1,"publicationDate":"2025-07-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144653826","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Lightweight real-time discriminative Siamese deep coupling framework for robust aerial tracking","authors":"Hanlin Huang ,&nbsp;Guixi Liu ,&nbsp;Ruke Xiong ,&nbsp;Yinghao Li ,&nbsp;Qian Lu ,&nbsp;Zhiyu Wu","doi":"10.1016/j.ins.2025.122510","DOIUrl":"10.1016/j.ins.2025.122510","url":null,"abstract":"<div><div>Recently, transformer-based Unmanned Aerial Vehicle (UAV) trackers have achieved notable success. However, the computationally intensive transformer model limits these trackers to static templates and shallow backbone networks, hampering their discriminative power and localization precision. Here, we propose a novel discriminative Siamese deep-coupling framework. This framework constructs a lightweight fine-grid anchor-free Siamese tracker with high spatial resolution specifically tailored for UAV scenarios, and complements its discriminative power with a targeted online discriminator. To achieve this, an efficient distractor detector is developed via knowledge transfer, enabling targeted detection of distractors that disturb the Siamese tracker. These distractors are utilized as training samples to construct a targeted online discriminator, which is deeply coupled with the Siamese tracker to enhance its discriminative power and specifically suppress hard distractors that hinder tracking performance. Additionally, a leading principal submatrix cluster sample space model and a scene-aware dynamic update strategy are developed to purify online samples and dynamically schedule the online discriminator update, significantly reducing the computational cost of the online discriminator optimization and boosting the tracker’s real-time performance. Finally, extensive experiments on eight UAV tracking benchmarks demonstrate that our tracker surpasses state-of-the-art transformer-based UAV trackers while achieving 70 FPS on CPU.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"719 ","pages":"Article 122510"},"PeriodicalIF":8.1,"publicationDate":"2025-07-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144613992","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}