Expert Systems with Applications最新文献

{"title":"Goal-driven long-term marine vessel trajectory prediction with a memory-enhanced network","authors":"Xiliang Zhang ,&nbsp;Jin Liu ,&nbsp;Chengcheng Chen ,&nbsp;Lai Wei ,&nbsp;Zhongdai Wu ,&nbsp;Wenjuan Dai","doi":"10.1016/j.eswa.2024.125715","DOIUrl":"10.1016/j.eswa.2024.125715","url":null,"abstract":"<div><div>Enhancing the precision of marine vessel trajectory prediction (VTP) is crucial for collision avoidance, intelligent navigation, and crisis alert in maritime safety. Most RNN-based methods typically face memory weakening issues during long-sequence propagation, leading to the discarding of some key features and significant predictive error accumulation over extended time intervals. Moreover, they struggle to forecast those complex trajectories involving abnormal maneuvers such as sudden acceleration or deceleration, sharp turns, or U-turns, resulting in poor generalization capabilities. To address these pivotal challenges, this paper proposes a novel Memory-Enhanced Network (MENet) for VTP, catering to intricate sailing intention modeling with long-term motion pattern perception. Specifically, we design an embeddable memory-enhanced block (MEB) that adaptively aggregates memory vectors across multiple temporal scales to assist in better prediction without disrupting the original backbone structure. Also, a goal-driven vessel trajectory decoder (GD-VTD) is developed to facilitate reliable model inferences by combining vessel type and destination variables as guidance information. Furthermore, we reconstruct the traditional loss function based on relative distance metrics, incorporating predicted headings into the optimization process to generate consistent trajectories that comply with realistic vessel dynamics. Ultimately, MENet could learn diverse sailing intentions by assembling the above parts to predict long-term marine vessel trajectories. Extensive experimental results on Automatic Identification System (AIS) datasets from three coastal regions in the US demonstrate that our model exhibits superior accuracy and robustness compared to other baselines. Specifically, on the Everglades Port (EP) dataset, our method reduces MAE, RMSE, and MAPE errors by 7.25%, 7.82%, and 7.62%, respectively, compared to the existing best results during this experiment. This is another piece of evidence for the effectiveness of goal-driven trajectory prediction in real-world maritime settings.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125715"},"PeriodicalIF":7.5,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662279","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":"LRMM: Low rank multi-scale multi-modal fusion for person re-identification based on RGB-NI-TI","authors":"Di Wu ,&nbsp;Zhihui Liu ,&nbsp;Zihan Chen ,&nbsp;Shenglong Gan ,&nbsp;Kaiwen Tan ,&nbsp;Qin Wan ,&nbsp;Yaonan Wang","doi":"10.1016/j.eswa.2024.125716","DOIUrl":"10.1016/j.eswa.2024.125716","url":null,"abstract":"<div><div>Person Re-identification is a crucial task in video surveillance, aiming to match person images from non-overlapping camera views. Recent methods introduce the Near-Infrared (NI) modality to alleviate the limitations of traditional single visible light modality under low-light conditions, while they overlook the importance of modality-related information. To incorporate more additional complementary information to assist traditional person re-identification tasks, in this paper, a novel RGB-NI-TI multi-modal person re-identification approach is proposed. First, we design a multi-scale multi-modal interaction module to facilitate cross-modal information fusion across multiple scales. Secondly, we propose a low-rank multi-modal fusion module that leverages the feature and weight parallel decomposition and then employs low-rank modality-specific factors for multimodal fusion. It aims to make the model more efficient in fusing multiple modal features while reducing complexity. Finally, we propose a multiple modalities prototype loss to supervise the network jointly with the cross-entropy loss, enforcing the network to learn modality-specific information by improving the intra-class cross-modality similarity and expanding the inter-class difference. The experimental results on benchmark multi-modal Re-ID datasets (RGBNT201, RGBNT100, MSVR310) and constructed person Re-ID datasets (multimodal version Market1501, PRW) validate the effectiveness of the proposed approach compared with the state-of-the-art methods.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125716"},"PeriodicalIF":7.5,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662114","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":"To disclose or to conceal? Comparison of different disclosure policies in queues with loss-averse customers","authors":"Jian Cao ,&nbsp;Yongjiang Guo","doi":"10.1016/j.eswa.2024.125635","DOIUrl":"10.1016/j.eswa.2024.125635","url":null,"abstract":"<div><div>In many service industries, information disclosure about the product can alleviate customers’ loss aversion induced by uncertain product valuation. In this paper, we consider a single-server queueing system in which the manager who privately learns the valuation information discloses the valuation information strategically to loss-averse customers. We investigate the impact of the customers’ loss aversion on the system’s equilibrium arrival rate and the manager’s optimal disclosure policy. We find that loss aversion restrains customers from joining the queue. Surprisingly, we find that there is no one disclosure policy that always prevails over other disclosure policies. Specifically, the full disclosure policy is optimal only when the valuation is large and the degree of loss aversion is moderate. The full non-disclosure policy is optimal when the degree of loss aversion is too large or too small, or the valuation is small. The threshold disclosure policy is optimal when the valuation and the degree of loss aversion are moderate. Furthermore, under the threshold disclosure policy, the increasing degree of loss aversion makes managers be more reluctant to disclose the valuation.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125635"},"PeriodicalIF":7.5,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662187","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":"AFDFusion: An adaptive frequency decoupling fusion network for multi-modality image","authors":"Chengchao Wang ,&nbsp;Zhengpeng Zhao ,&nbsp;Qiuxia Yang ,&nbsp;Rencan Nie ,&nbsp;Jinde Cao ,&nbsp;Yuanyuan Pu","doi":"10.1016/j.eswa.2024.125694","DOIUrl":"10.1016/j.eswa.2024.125694","url":null,"abstract":"<div><div>The multi-modality image fusion goal is to create a single image that provides a comprehensive scene description and conforms to visual perception by integrating complementary information about the merits of the different modalities, <em>e.g</em>., salient intensities of infrared images and detail textures of visible images. Although some works explore decoupled representations of multi-modality images, they struggle with complex nonlinear relationships, fine modal decoupling, and noise handling. To cope with this issue, we propose an adaptive frequency decoupling module to perceive the associative invariant and inherent specific among cross-modality by dynamically adjusting the learnable low frequency weight of the kernel. Specifically, we utilize a contrastive learning loss for restricting the solution space of feature decoupling to learn representations of both the invariant and specific in the multi-modality images. The underlying idea is that: in decoupling, low frequency features, which are similar in the representation space, should be pulled closer to each other, signifying the associative invariant, while high frequencies are pushed farther away, also indicating the intrinsic specific. Additionally, a multi-stage training manner is introduced into our framework to achieve decoupling and fusion. Stage I, <em>MixEncoder</em> and <em>MixDecoder</em> with the same architecture but different parameters are trained to perform decoupling and reconstruction supervised by the contrastive self-supervised mechanism. Stage II, two feature fusion modules are added to integrate the invariant and specific features and output the fused image. Extensive experiments demonstrated the proposed method superiority over the state-of-the-art methods in both qualitative and quantitative evaluation on two multi-modal image fusion tasks.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125694"},"PeriodicalIF":7.5,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662280","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 method based on hybrid cross-multiscale spectral-spatial transformer network for hyperspectral and multispectral image fusion","authors":"Yingxia Chen ,&nbsp;Mingming Wei ,&nbsp;Yan Chen","doi":"10.1016/j.eswa.2024.125742","DOIUrl":"10.1016/j.eswa.2024.125742","url":null,"abstract":"<div><div>Convolutional neural networks (CNNs) have made a significant contribution to hyperspectral image (HSI) generation. However, capturing long-range dependencies can be challenging with CNNs due to the limitations of their local receptive fields, which can lead to distortions in fused images. Transformers excel at capturing long-range dependencies but have limited capacity for handling fine details. Additionally, prior<!--> <!-->work has often overlooked the extraction of global features during the image preprocessing stage, resulting in the potential loss of fine details. To address these issues, we propose a hybrid cross-multiscale spectral-spatial Transformer (HCMSST) that combines the advantages of CNNs in feature extraction and Transformers in capturing long-range dependencies. To fully extract and retain local and global information in the shallow feature extraction phase, the network incorporates<!--> <!-->CNNs with a staggered cascade-dense residual block (SCDRB). This block employs staggered residuals to establish direct connections both<!--> <!-->within and between branches and integrates attention modules to enhance the response to important features. This approach facilitates unrestricted information exchange and fosters deeper feature representations. To address the limitations<!--> <!-->of Transformer in processing fine details, we introduce multiscale spatial-spectral coding-decoding structures to obtain comprehensive spatial-spectral features, which are utilized to capture the long-range dependencies via the cross-multiscale spectral-spatial Transformer (CMSST). Further, the CMSST incorporates a cross-level dual-stream feature interaction strategy that integrates spatial and spectral features from different levels and then feeds the fused features back to their corresponding branches for information interaction. Experimental results indicate that the proposed HCMSST achieves superior performance compared to many state-of-the-art (SOTA) methods. Specifically, HCMSST reduces the ERGAS metric by 3.05% compared to the SOTA methods on the CAVE dataset, while on the Harvard dataset, it achieves a 2.69% reduction in ERGAS compared to the SOTA results.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125742"},"PeriodicalIF":7.5,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662184","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}

{"title":"Negative sampling strategy based on multi-hop neighbors for graph representation learning","authors":"Kaiyu Zhang,&nbsp;Guoming Sang,&nbsp;Junkai Cheng,&nbsp;Zhi Liu,&nbsp;Yijia Zhang","doi":"10.1016/j.eswa.2024.125688","DOIUrl":"10.1016/j.eswa.2024.125688","url":null,"abstract":"<div><div>Contrastive learning (CL) has recently achieved significant success in the field of recommendation system. However, current studies mainly focus on obtaining high-quality positive samples and focus less on selecting negative samples. In existing recommendation system based on graph contrastive learning, most methods select negative samples by randomly selecting samples that have not interacted with the target node. Although random negative sampling is easy to implement and has wide applicability, it may lead to problems such as unbalanced data distribution and selection of false negative samples, which can degrade model performance. To address the above issues, we propose a novel negative sampling strategy called the Multi-hop Neighbors Negative Sampling method, named NSHN. Specifically, we select the information of 3-hop neighbors of each node as candidate negative samples. In addition, to reduce the impact of false negative noise on negative samples, we propose an adaptive denoising training strategy that adaptively prunes noise interactions during training. Experimental results demonstrate that our method performs well on four datasets and outperforms graph contrastive learning methods that use random negative sampling. The source code is available at: <span><span>https://github.com/zhangkaiyu-zky/NSHN</span><svg><path></path></svg></span></div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125688"},"PeriodicalIF":7.5,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662121","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":"TSIDS: Spatial–temporal fusion gating Multilayer Perceptron for network intrusion detection","authors":"Jie Fu ,&nbsp;Lina Wang ,&nbsp;Jianpeng Ke ,&nbsp;Kang Yang ,&nbsp;Rongwei Yu","doi":"10.1016/j.eswa.2024.125687","DOIUrl":"10.1016/j.eswa.2024.125687","url":null,"abstract":"<div><div>Due to the heterogeneous and dynamic nature of networks, modeling spatiotemporal correlations has become a trend. Although spatiotemporal-based network intrusion detection systems (NIDSs) enhance the performance of intrusion classification, they still suffer from inadequacies in the multi-classification of intrusions and model generalization ability. First, the static attack topologies of network traffic always ignore some important information; Second, the interaction between spatial and temporal dimensions is rarely considered. To mitigate these issues, this paper proposes TSIDS, a spatiotemporal analysis-based approach that extracts the interaction of network behaviors for intrusion detection. TSIDS combines the spatial analysis module to extract spatial information between different events, and the temporal analysis module to learn the temporal dependencies from historical traffic data. To model spatial correlations of temporal features, we propose a feature fusion module based on our customized gating Multilayer Perceptron (cgMLP). The experimental results on four datasets show that our work is effective in intrusion detection, especially multi-classification, and outperforms other baseline methods.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125687"},"PeriodicalIF":7.5,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662125","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":"CoSEF-DBP: Convolution scope expanding fusion network for identifying DNA-binding proteins through bilingual representations","authors":"Hua Zhang ,&nbsp;Xiaoqi Yang ,&nbsp;Pengliang Chen ,&nbsp;Cheng Yang ,&nbsp;Bi Chen ,&nbsp;Bo Jiang ,&nbsp;Guogen Shan","doi":"10.1016/j.eswa.2024.125763","DOIUrl":"10.1016/j.eswa.2024.125763","url":null,"abstract":"<div><div>Precisely recognizing DNA-binding proteins (DBPs) from sequences is crucial for a profound comprehension of the mechanisms governing protein-DNA interactions in various cellular processes. However, traditional in-silico methods for DBP identification encounter several challenges, such as time-consuming evolutionary modeling based on multiple sequence alignments, and intricate feature engineering associated with machine or deep learning approaches. In this paper, we introduce a novel end-to-end predictor for identifying DNA-binding proteins without intricate feature engineering, which innovatively enriches the semantics of amino acid sequences through the fusion of bilingual representations derived from distinct language models. We further design a convolution scope expanding (CoSE) module to widen the receptive fields of convolution kernels, thereby forming protein-level CoSE representation sequences. These representations are subsequently integrated via BiLSTM in conjunction with a simplified capsule network, enhancing the hierarchical feature extraction capability. Extensive experiments confirm that our model surpasses existing baselines across diverse benchmark datasets, notably achieving at least a 5.1% improvement in MCC value on the UniSwiss dataset.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125763"},"PeriodicalIF":7.5,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662189","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 cost-minimized two-stage three-way dynamic consensus mechanism for social network-large scale group decision-making: Utilizing K-nearest neighbors for incomplete fuzzy preference relations","authors":"Jiaxin Zhan,&nbsp;Mingjie Cai","doi":"10.1016/j.eswa.2024.125705","DOIUrl":"10.1016/j.eswa.2024.125705","url":null,"abstract":"<div><div>In the era of big data, large scale group decision-making (LSGDM) with social networks (SNs) (namely, SN-LSGDM) has become a hot topic in the field of decision science. Faced with the explosive growth of information, decision-makers (DMs) face immense challenges in processing and integrating vast amounts of data, often finding it difficult to fully comprehend all the information, leading to potentially incomplete expressions of their fuzzy preference relations (FPRs). This limitation in information processing not only affects the quality of decision-making but also increases the difficulty and cost of reaching a consensus. To overcome these challenges and enhance the efficiency and accuracy of decision-making, this paper designs a consensus model that minimizes adjustment costs in light of a dynamic trust network. Firstly, we introduce a measurement method based on <span><math><mi>K</mi></math></span>-nearest neighbor (KNN) information, which comprehensively considers the trust level of DMs and the similarity of preference relations, effectively filling in missing preference information and improving the completeness and accuracy of decision-making. In addition, an improved <span><math><mi>k</mi></math></span>-means clustering algorithm is adopted, which takes into account the mutual influences between DMs and the cost of unit adjustment. On this basis, a two-stage minimum adjustment cost consensus reaching mechanism based on three-way decision (TWD) is proposed, using comprehensive adjustment priority as the criterion for division, to achieve feedback adjustment at the individual and subgroup levels, ensuring the coordination and consistency of the decision-making plan. At the same time, an optimization model is introduced to achieve cost minimization. Through detailed case studies and comparative analysis, the feasibility and superiority of this method in practical applications have been demonstrated.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125705"},"PeriodicalIF":7.5,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662207","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":"Exploring cluster-dependent isomorphism in multi-objective evolutionary optimization","authors":"Wei Zheng ,&nbsp;Jianyong Sun","doi":"10.1016/j.eswa.2024.125684","DOIUrl":"10.1016/j.eswa.2024.125684","url":null,"abstract":"<div><div>In this paper, a Two-Round learning-based Algorithm for Continuous box-constrained multi-objective Evolutionary optimization (TRACE) under the decomposition framework is proposed, in which the isomorphism relationship between the clustered Pareto Front and Pareto solution set is explored and a new time-varying adaptive crossover operator is developed. The learning process involves two stages. In the first stage, the <span><math><mi>K</mi></math></span>-means is applied to cluster the population of objective vectors. By exploring the property of cluster-dependent isomorphism between the objective space and the decision space, a parent individual for each individual is selected from the corresponding clusters in the decision space. The time-varying adaptive crossover operator is then used together with the classical polynomial mutation operator to generate a new solution based on the selected parent individuals. As part of the environmental selection process, the <span><math><mi>K</mi></math></span>-means is applied again to the combination of parent and offspring individuals in the objective space to assist in the selection of suitable solutions for each decomposed subspace. TRACE is compared with 11 state-of-the-art multi-objective evolutionary algorithms on totally 43 difficult problems with different characteristics. Furthermore, TRACE is compared with three promising multi-objective evolutionary algorithms for community detection in attribute networks. Extensive experiments show that TRACE significantly outperforms the compared algorithms in most instances.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"263 ","pages":"Article 125684"},"PeriodicalIF":7.5,"publicationDate":"2024-11-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142662123","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}