Expert Systems with Applications最新文献_第8页

A Basic Sequential Variable Neighbourhood Descent and neighbourhood uses for the Beam Angle Optimisation problem

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-17 DOI: 10.1016/j.eswa.2025.126801

Maicholl Gutiérrez, Guillermo Cabrera-Guerrero, Carolina Lagos

Intensity Modulated Radiation Therapy (IMRT) is a treatment alternative for cancer treatment. The primary objective of IMRT is to eradicate cancer cells from the tumour site while minimising harm to the surrounding Organs at Risk (OAR). The first step toward achieving this goal is determining the optimal beam angle configuration (BAC) for the treatment plan. In this paper, we propose studying the Basic Sequential Variable Neighbourhood Descent (BVND) algorithm that explores the search space using two types of movements. The first movement (

N_{1}

) replaces a beam angle in the BAC by a

\pm 5 °

beam angle, while the second movement (

N_{2}

) replaces a beam angle in the BAC by a randomly chosen beam angle. Unlike traditional VNS algorithms, the BVND algorithm obtains a final solution that is locally optimal for the neighbourhood

N_{1}

and approximately locally optimal for neighbourhood

N_{2}

. This BVND algorithm ensures that the final solution is locally optimal for neighbourhood

N_{1}

and approximately locally optimal for neighbourhood

N_{2}

. Results show that

N_{2}

movement is more commonly used in early iterations, often giving better improvements, while

N_{1}

movement is more commonly used in later iterations, maintaining a stable improvement across iterations. We try our approach on a set of clinical prostate cases from a hospital in Chile. The BVND is a robust algorithm that can obtain high-quality treatment plans at the cost of more computational time compared to other VNS-based algorithms, although it is shown to perform quite well, too, under a limited number of function evaluations. The BVND proved to be a robust algorithm that can obtain high-quality treatment plans at the cost of more computational time than other VNS-based algorithms. Furthermore, it has also been shown to perform quite well under a limited number of function evaluations.

{"title":"A Basic Sequential Variable Neighbourhood Descent and neighbourhood uses for the Beam Angle Optimisation problem","authors":"Maicholl Gutiérrez, Guillermo Cabrera-Guerrero, Carolina Lagos","doi":"10.1016/j.eswa.2025.126801","DOIUrl":"10.1016/j.eswa.2025.126801","url":null,"abstract":"<div><div>Intensity Modulated Radiation Therapy (IMRT) is a treatment alternative for cancer treatment. The primary objective of IMRT is to eradicate cancer cells from the tumour site while minimising harm to the surrounding Organs at Risk (OAR). The first step toward achieving this goal is determining the optimal beam angle configuration (BAC) for the treatment plan. In this paper, we propose studying the Basic Sequential Variable Neighbourhood Descent (BVND) algorithm that explores the search space using two types of movements. The first movement (<span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span>) replaces a beam angle in the BAC by a <span><math><mrow><mo>±</mo><mn>5</mn><mo>°</mo></mrow></math></span> beam angle, while the second movement (<span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span>) replaces a beam angle in the BAC by a randomly chosen beam angle. Unlike traditional VNS algorithms, the BVND algorithm obtains a final solution that is locally optimal for the neighbourhood <span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span> and approximately locally optimal for neighbourhood <span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span>. This BVND algorithm ensures that the final solution is locally optimal for neighbourhood <span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span> and approximately locally optimal for neighbourhood <span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span>. Results show that <span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span> movement is more commonly used in early iterations, often giving better improvements, while <span><math><msub><mrow><mi>N</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span> movement is more commonly used in later iterations, maintaining a stable improvement across iterations. We try our approach on a set of clinical prostate cases from a hospital in Chile. The BVND is a robust algorithm that can obtain high-quality treatment plans at the cost of more computational time compared to other VNS-based algorithms, although it is shown to perform quite well, too, under a limited number of function evaluations. The BVND proved to be a robust algorithm that can obtain high-quality treatment plans at the cost of more computational time than other VNS-based algorithms. Furthermore, it has also been shown to perform quite well under a limited number of function evaluations.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126801"},"PeriodicalIF":7.5,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143438167","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Impact of cognitive dissonance on social hysteresis: Insights from the expressed and private opinions model

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-17 DOI: 10.1016/j.eswa.2025.126851

Barbara Kamińska, Katarzyna Sznajd-Weron

The growing interest in models of opinion dynamics that distinguish between private beliefs and publicly expressed opinions spans several academic disciplines, ranging from the social sciences to the hard sciences. These models have been developed to study decision-making processes and have been applied to a variety of social phenomena, such as pluralistic ignorance, the spiral of silence, and preference falsification. However, a significant gap exists in understanding social hysteresis – a concept essential for explaining the delayed societal responses to rapid changes in the modern world. This research addresses this gap by examining the impact of cognitive dissonance on social hysteresis using an updated model of expressed and private opinions (EPOs). We propose that, from a modeling perspective, reducing cognitive dissonance can be reframed as avoiding self-anticonformity, which enables us to draw on recent psychological experiments on strategic anticonformity. The model is analyzed both analytically and through Monte Carlo simulations. To promote accessibility and reproducibility, we have made the specialized NetLogo implementation of the model publicly available on GitHub. Additionally, we provide a concise review of existing EPOs models to contextualize our approach. By incorporating a cognitive dissonance mitigation mechanism into an agent-based

q

-voter-type model of EPOs, we demonstrate that this mechanism induces social hysteresis. Consequently, we argue that refraining from rationalizing publicly expressed opinions could mitigate social hysteresis and facilitate consensus, offering insights into potential strategies for managing societal responses to change.

{"title":"Impact of cognitive dissonance on social hysteresis: Insights from the expressed and private opinions model","authors":"Barbara Kamińska, Katarzyna Sznajd-Weron","doi":"10.1016/j.eswa.2025.126851","DOIUrl":"10.1016/j.eswa.2025.126851","url":null,"abstract":"<div><div>The growing interest in models of opinion dynamics that distinguish between private beliefs and publicly expressed opinions spans several academic disciplines, ranging from the social sciences to the hard sciences. These models have been developed to study decision-making processes and have been applied to a variety of social phenomena, such as pluralistic ignorance, the spiral of silence, and preference falsification. However, a significant gap exists in understanding social hysteresis – a concept essential for explaining the delayed societal responses to rapid changes in the modern world. This research addresses this gap by examining the impact of cognitive dissonance on social hysteresis using an updated model of expressed and private opinions (EPOs). We propose that, from a modeling perspective, reducing cognitive dissonance can be reframed as avoiding self-anticonformity, which enables us to draw on recent psychological experiments on strategic anticonformity. The model is analyzed both analytically and through Monte Carlo simulations. To promote accessibility and reproducibility, we have made the specialized NetLogo implementation of the model publicly available on <span><span>GitHub</span><svg><path></path></svg></span>. Additionally, we provide a concise review of existing EPOs models to contextualize our approach. By incorporating a cognitive dissonance mitigation mechanism into an agent-based <span><math><mi>q</mi></math></span>-voter-type model of EPOs, we demonstrate that this mechanism induces social hysteresis. Consequently, we argue that refraining from rationalizing publicly expressed opinions could mitigate social hysteresis and facilitate consensus, offering insights into potential strategies for managing societal responses to change.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126851"},"PeriodicalIF":7.5,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143438161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A higher-order neural cognitive diagnosis model with hierarchical attention networks

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-17 DOI: 10.1016/j.eswa.2025.126848

Tao Huang , Yuxia Chen , Jing Geng , Huali Yang , Shengze Hu

As fundamental abilities, higher-order abilities play an important role in representing a multidimensional synthesis of concepts, skills, and learning statuses. Cognitive diagnosis, a critical technology for assessing these abilities, aims to identify students’ specific skill statuses based on their response data. However, existing cognitive diagnostic models (CDMs) face significant challenges in accurately assessing these complex competencies, particularly in capturing the hierarchical structure of higher-order abilities—specifically, the progression from lower- to higher-order skills. To address this challenge, we propose a novel Higher-Order Neural Cognitive Diagnosis (HO-NCD) model, which leverages hierarchical attention mechanisms to assess higher-order abilities. The core of our model lies in its ability to model the transition from lower- to higher-order abilities, capturing both the intrinsic relationships between different levels of cognitive attributes. Specifically, a unified embedding layer encodes the characteristics of students, exercises, and concepts in a shared latent space, enabling a comprehensive representation. Subsequently, we model the transition from lower- to higher-order abilities through a hierarchical attention network, allowing the model to capture both direct and indirect relationships. Finally, a neural network is constructed to simulate the interactions among students, exercises, and concepts, thereby predicting future student performance. The effectiveness and interpretability of the HO-NCD model were evaluated using three benchmark datasets — Junyi, ASSIST2017, and PISA2015 — demonstrating its superior performance compared to existing models. The code is available at https://github.com/ccc-615/HO-NCD.

{"title":"A higher-order neural cognitive diagnosis model with hierarchical attention networks","authors":"Tao Huang , Yuxia Chen , Jing Geng , Huali Yang , Shengze Hu","doi":"10.1016/j.eswa.2025.126848","DOIUrl":"10.1016/j.eswa.2025.126848","url":null,"abstract":"<div><div>As fundamental abilities, higher-order abilities play an important role in representing a multidimensional synthesis of concepts, skills, and learning statuses. Cognitive diagnosis, a critical technology for assessing these abilities, aims to identify students’ specific skill statuses based on their response data. However, existing cognitive diagnostic models (CDMs) face significant challenges in accurately assessing these complex competencies, particularly in capturing the hierarchical structure of higher-order abilities—specifically, the progression from lower- to higher-order skills. To address this challenge, we propose a novel Higher-Order Neural Cognitive Diagnosis (HO-NCD) model, which leverages hierarchical attention mechanisms to assess higher-order abilities. The core of our model lies in its ability to model the transition from lower- to higher-order abilities, capturing both the intrinsic relationships between different levels of cognitive attributes. Specifically, a unified embedding layer encodes the characteristics of students, exercises, and concepts in a shared latent space, enabling a comprehensive representation. Subsequently, we model the transition from lower- to higher-order abilities through a hierarchical attention network, allowing the model to capture both direct and indirect relationships. Finally, a neural network is constructed to simulate the interactions among students, exercises, and concepts, thereby predicting future student performance. The effectiveness and interpretability of the HO-NCD model were evaluated using three benchmark datasets — Junyi, ASSIST2017, and PISA2015 — demonstrating its superior performance compared to existing models. The code is available at <span><span>https://github.com/ccc-615/HO-NCD</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126848"},"PeriodicalIF":7.5,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143464070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Time-based Knowledge-aware framework for Multi-Behavior Recommendation

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-17 DOI: 10.1016/j.eswa.2025.126840

Xiujuan Li , Nan Wang , Xin Liu , Jin Zeng , Jinbao Li

Multi-behavior recommendation alleviates the data sparsity problem in single-behavior recommendation by exploiting the multi-dimensional behavioral information of users to construct rich connections between users and items. However, existing multi-behavior recommendation methods tend to ignore the temporal information of user interactions, which makes it difficult to dynamically understand user preferences. In addition, introducing too much behavioral information may lead to negative migration problem in the model (i.e., the newly introduced behavioral information conflicts with the original information), which leads to model performance degradation. Based on the above background and challenges, we propose a Time-based Knowledge-aware Multi-Behavior Recommendation framework (TKMB). The framework combines multi-behavior and temporal information of users, and achieves comprehensive modeling of user preferences and item information through three main views: the local multi-behavior interaction view, the global multi-behavior interaction view and the knowledge-aware view. The first two separately design a local and global self-attention mechanism to distinguish the importance of different behaviors. And designs an adaptive time gating mechanism to dynamically capture users’ personalized preferences. The latter constructs high-order representations at the item level and proposes a graph reconstruction strategy and knowledge-aware contrastive learning to enhance the robustness of the model. Finally, a multi-view aggregation mechanism is introduced to aggregate multi-scale representations. The results of extensive experiments and ablation experiments on two real datasets further validate the effectiveness and superiority of TKMB.

{"title":"Time-based Knowledge-aware framework for Multi-Behavior Recommendation","authors":"Xiujuan Li , Nan Wang , Xin Liu , Jin Zeng , Jinbao Li","doi":"10.1016/j.eswa.2025.126840","DOIUrl":"10.1016/j.eswa.2025.126840","url":null,"abstract":"<div><div>Multi-behavior recommendation alleviates the data sparsity problem in single-behavior recommendation by exploiting the multi-dimensional behavioral information of users to construct rich connections between users and items. However, existing multi-behavior recommendation methods tend to ignore the temporal information of user interactions, which makes it difficult to dynamically understand user preferences. In addition, introducing too much behavioral information may lead to negative migration problem in the model (i.e., the newly introduced behavioral information conflicts with the original information), which leads to model performance degradation. Based on the above background and challenges, we propose a Time-based Knowledge-aware Multi-Behavior Recommendation framework (TKMB). The framework combines multi-behavior and temporal information of users, and achieves comprehensive modeling of user preferences and item information through three main views: the local multi-behavior interaction view, the global multi-behavior interaction view and the knowledge-aware view. The first two separately design a local and global self-attention mechanism to distinguish the importance of different behaviors. And designs an adaptive time gating mechanism to dynamically capture users’ personalized preferences. The latter constructs high-order representations at the item level and proposes a graph reconstruction strategy and knowledge-aware contrastive learning to enhance the robustness of the model. Finally, a multi-view aggregation mechanism is introduced to aggregate multi-scale representations. The results of extensive experiments and ablation experiments on two real datasets further validate the effectiveness and superiority of TKMB.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126840"},"PeriodicalIF":7.5,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143453661","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Insect identification by combining different neural networks

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-16 DOI: 10.1016/j.eswa.2025.126935

Loris Nanni , Nicola Maritan , Daniel Fusaro , Sheryl Brahnam , Francesco Boscolo Meneguolo , Maria Sgaravatto

Background

Traditional insect species classification relies on taxonomic experts examining unique physical characteristics of specimens, a time-consuming and error-prone process. Machine learning (ML) offers a promising alternative by identifying subtle morphological and genetic differences computationally. However, most existing approaches classify undescribed species as outliers, which limits their utility for biodiversity monitoring.

Objective

This study aims to develop an ML method capable of simultaneously classifying described species and grouping undescribed species by genus, thereby advancing the field of automated insect classification.

Method

We propose a novel ensemble approach combining neural networks (convolutional and attention-based) and Support Vector Machines (SVM), with both DNA barcoding and insect images as input data. To optimize the neural networks for diverse data types, we transform one-dimensional feature vectors into matrices using wavelet transforms. Additionally, a transformer-based architecture integrates DNA barcoding and image features for enhanced classification accuracy.

Experimental Results

Our method was evaluated on a comprehensive dataset containing paired insect images and DNA barcodes for 1,040 species across four insect orders. The results demonstrate superior performance compared to existing methods in classifying described species and grouping undescribed ones by genus.

Conclusion

The proposed approach represents a significant advancement in automated insect classification, addressing both described and undescribed species. This method has the potential to revolutionize global biodiversity monitoring. The MATLAB/PyTorch source code and dataset used are available at https://github.com/LorisNanni/Insect-identification.

{"title":"Insect identification by combining different neural networks","authors":"Loris Nanni , Nicola Maritan , Daniel Fusaro , Sheryl Brahnam , Francesco Boscolo Meneguolo , Maria Sgaravatto","doi":"10.1016/j.eswa.2025.126935","DOIUrl":"10.1016/j.eswa.2025.126935","url":null,"abstract":"<div><h3>Background</h3><div>Traditional insect species classification relies on taxonomic experts examining unique physical characteristics of specimens, a time-consuming and error-prone process. Machine learning (ML) offers a promising alternative by identifying subtle morphological and genetic differences computationally. However, most existing approaches classify undescribed species as outliers, which limits their utility for biodiversity monitoring.</div></div><div><h3>Objective</h3><div>This study aims to develop an ML method capable of simultaneously classifying described species and grouping undescribed species by genus, thereby advancing the field of automated insect classification.</div></div><div><h3>Method</h3><div>We propose a novel ensemble approach combining neural networks (convolutional and attention-based) and Support Vector Machines (SVM), with both DNA barcoding and insect images as input data. To optimize the neural networks for diverse data types, we transform one-dimensional feature vectors into matrices using wavelet transforms. Additionally, a transformer-based architecture integrates DNA barcoding and image features for enhanced classification accuracy.</div></div><div><h3>Experimental Results</h3><div>Our method was evaluated on a comprehensive dataset containing paired insect images and DNA barcodes for 1,040 species across four insect orders. The results demonstrate superior performance compared to existing methods in classifying described species and grouping undescribed ones by genus.</div></div><div><h3>Conclusion</h3><div>The proposed approach represents a significant advancement in automated insect classification, addressing both described and undescribed species. This method has the potential to revolutionize global biodiversity monitoring. The MATLAB/PyTorch source code and dataset used are available at <span><span>https://github.com/LorisNanni/Insect-identification</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126935"},"PeriodicalIF":7.5,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143428994","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimization of integrated accurate ride-tide planning and vessel scheduling in multi-functional ports with long channels

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-16 DOI: 10.1016/j.eswa.2025.126894

Xinyu Zhang , Wenqiang Guo , Zaili Yang , Jingyun Wang , Chengbo Wang

Driven by economic globalization and the reduction of transport costs, “large-scale vessels” and “multi-functional ports” have emerged as the new paradigm in marine transportation. Within this context, traditional ride-tide planning and common rule-based vessel scheduling methods sometimes become ineffective in optimizing navigational potential, leading to serious waiting problems for large-scale vessels and increasing port congestion. This paper aims to develop a new point-by-point ship ride-tide (PSRT) approach to address accurate ride-tide planning and vessel scheduling issues for a long channel in a multi-functional port. The method is developed based on coupling the vessel’s speed change and tidal level variation to determine accurate ride-tide planning for large-scale vessels. A mixed integer linear programming (MILP) model is presented, in which the vessel’s accurate tide ride, dynamic sailing speeds, and vessel scheduling priority are explicitly considered. Due to the computational inefficiency of the MILP model in large-scale scenarios, we decompose it into a master problem and several subproblems and develop an improved branch-and-price (B&P) algorithm with three enhanced methods to solve this model. Computational experiments for Huanghua Port show that the PSRT method extends available tidal time windows (ATTWs) for large-scale vessels by an average of 20% compared to the traditional single-point tide-ride approach for the same under keel clearance. Moreover, the proposed improved B&P algorithm significantly outperforms existing methods such as column generation, branch and bound, and an improved genetic algorithm as well as the port scheduling schemes adopted in reality. This study has effectively unlocked the navigational potential of long channels, making new contributions to enabling ports to accommodate and serve a greater number of large and ultra-large vessels.

{"title":"Optimization of integrated accurate ride-tide planning and vessel scheduling in multi-functional ports with long channels","authors":"Xinyu Zhang , Wenqiang Guo , Zaili Yang , Jingyun Wang , Chengbo Wang","doi":"10.1016/j.eswa.2025.126894","DOIUrl":"10.1016/j.eswa.2025.126894","url":null,"abstract":"<div><div>Driven by economic globalization and the reduction of transport costs, “large-scale vessels” and “multi-functional ports” have emerged as the new paradigm in marine transportation. Within this context, traditional ride-tide planning and common rule-based vessel scheduling methods sometimes become ineffective in optimizing navigational potential, leading to serious waiting problems for large-scale vessels and increasing port congestion. This paper aims to develop a new point-by-point ship ride-tide (PSRT) approach to address accurate ride-tide planning and vessel scheduling issues for a long channel in a multi-functional port. The method is developed based on coupling the vessel’s speed change and tidal level variation to determine accurate ride-tide planning for large-scale vessels. A mixed integer linear programming (MILP) model is presented, in which the vessel’s accurate tide ride, dynamic sailing speeds, and vessel scheduling priority are explicitly considered. Due to the computational inefficiency of the MILP model in large-scale scenarios, we decompose it into a master problem and several subproblems and develop an improved branch-and-price (B&P) algorithm with three enhanced methods to solve this model. Computational experiments for Huanghua Port show that the PSRT method extends available tidal time windows (ATTWs) for large-scale vessels by an average of 20% compared to the traditional single-point tide-ride approach for the same under keel clearance. Moreover, the proposed improved B&P algorithm significantly outperforms existing methods such as column generation, branch and bound, and an improved genetic algorithm as well as the port scheduling schemes adopted in reality. This study has effectively unlocked the navigational potential of long channels, making new contributions to enabling ports to accommodate and serve a greater number of large and ultra-large vessels.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126894"},"PeriodicalIF":7.5,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143446054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SD-MIL: Multiple instance learning with dual perception of scale and distance information fusion for whole slide image classification

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-16 DOI: 10.1016/j.eswa.2025.126831

Yining Xie , Zequn Liu , Jiajun Chen , Wei Zhang , Jing Zhao , Jiayi Ma

In computer-aided pathology diagnosis, multiple instance learning (MIL) has become a key method for addressing disease diagnosis problems in whole slide images (WSIs). However, current MIL models have limitations in capturing dependencies among instances and local contextual information. Additionally, the imbalance in the number of positive and negative instances affects MIL models’ ability to identify important instances. To address these issues, we propose a dual perception of scale and distance information fusion method (SD-MIL). SD-MIL consists of two parts: multi-scale window regional self-attention (MWRSA) and adaptive prototype distance-guided instance feature enhancement (PGFE). MWRSA utilizes three different-sized windows to compute regional multi-head self-attention (R-MSA) obtaining scale-aware instance features. This part explores instance long-range dependencies in local region and capture local contextual information at different scales. In the PGFE part, the distance parameter between instances and bag-level prototype is considered to assign different significance weights to instances resulting in distance-aware instance features, which guides model better focus on important instances. Then, learnable parameters optimize the fusion of scale-aware and distance-aware instance features, enhancing instance feature representation and ensuring the downstream aggregation model to generate high-quality bag features. Experimental results on three datasets show that SD-MIL outperforms state-of-the-art MIL methods. Meanwhile, SD-MIL consistently delivers performance improvements when the feature extraction network or downstream aggregation model is replaced.

{"title":"SD-MIL: Multiple instance learning with dual perception of scale and distance information fusion for whole slide image classification","authors":"Yining Xie , Zequn Liu , Jiajun Chen , Wei Zhang , Jing Zhao , Jiayi Ma","doi":"10.1016/j.eswa.2025.126831","DOIUrl":"10.1016/j.eswa.2025.126831","url":null,"abstract":"<div><div>In computer-aided pathology diagnosis, multiple instance learning (MIL) has become a key method for addressing disease diagnosis problems in whole slide images (WSIs). However, current MIL models have limitations in capturing dependencies among instances and local contextual information. Additionally, the imbalance in the number of positive and negative instances affects MIL models’ ability to identify important instances. To address these issues, we propose a dual perception of scale and distance information fusion method (SD-MIL). SD-MIL consists of two parts: multi-scale window regional self-attention (MWRSA) and adaptive prototype distance-guided instance feature enhancement (PGFE). MWRSA utilizes three different-sized windows to compute regional multi-head self-attention (R-MSA) obtaining scale-aware instance features. This part explores instance long-range dependencies in local region and capture local contextual information at different scales. In the PGFE part, the distance parameter between instances and bag-level prototype is considered to assign different significance weights to instances resulting in distance-aware instance features, which guides model better focus on important instances. Then, learnable parameters optimize the fusion of scale-aware and distance-aware instance features, enhancing instance feature representation and ensuring the downstream aggregation model to generate high-quality bag features. Experimental results on three datasets show that SD-MIL outperforms state-of-the-art MIL methods. Meanwhile, SD-MIL consistently delivers performance improvements when the feature extraction network or downstream aggregation model is replaced.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126831"},"PeriodicalIF":7.5,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143446055","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

High-Precision machining energy consumption prediction based on multi-sensor data fusion and Ns-Transformer network

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-16 DOI: 10.1016/j.eswa.2025.126903

Meihang Zhang , Hua Zhang , Wei Yan , Zhigang Jiang , Rui Tian

Achieving energy efficiency and cost-effectiveness in machining relies on accurate predictions of energy consumption. Despite the advancements in deep learning for predictive applications, precise energy modeling through multi-sensor data integration remains challenging, particularly due to the computational demands of large datasets. To address this, an Ns-Transformer-based strategy leveraging multi-sensor data fusion for high-precision energy consumption prediction is proposed. The methodology begins with data preprocessing, incorporating Lagrange interpolation, Butterworth filtering, principal component analysis, and correlation analysis to identify critical features. Key time-series features are then fused with energy consumption data to create an enriched feature space. The fused features are subjected to feature learning through a dual-layer Ns-Transformer network, followed by the application of linear regression to map the energy consumption state, thereby ensuring prediction accuracy. The framework employs distinct models for training and prediction, sharing parameters to reduce computational overhead. Experimental results demonstrate significant accuracy improvements, with mean squared error reductions exceeding 76% for carbon fiber and surpassing 83.2% for plastics, aluminum, and steel.

{"title":"High-Precision machining energy consumption prediction based on multi-sensor data fusion and Ns-Transformer network","authors":"Meihang Zhang , Hua Zhang , Wei Yan , Zhigang Jiang , Rui Tian","doi":"10.1016/j.eswa.2025.126903","DOIUrl":"10.1016/j.eswa.2025.126903","url":null,"abstract":"<div><div>Achieving energy efficiency and cost-effectiveness in machining relies on accurate predictions of energy consumption. Despite the advancements in deep learning for predictiveapplications, precise energy modeling through multi-sensor data integration remains challenging, particularly due to the computational demands of large datasets. To address this, an Ns-Transformer-based strategy leveraging multi-sensor data fusion for high-precision energy consumption prediction is proposed. The methodology begins with data preprocessing, incorporating Lagrange interpolation, Butterworth filtering, principal component analysis, and correlation analysis to identify critical features. Key time-series features are then fused with energy consumption data to create an enriched feature space. The fused features are subjected to feature learning through a dual-layer Ns-Transformer network, followed by the application of linear regression to map the energy consumption state, thereby ensuring prediction accuracy. The framework employs distinct models for training and prediction, sharing parameters to reduce computational overhead. Experimental results demonstrate significant accuracy improvements, with mean squared error reductions exceeding 76% for carbon fiber and surpassing 83.2% for plastics, aluminum, and steel.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126903"},"PeriodicalIF":7.5,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143453663","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

XLight: An interpretable multi-agent reinforcement learning approach for traffic signal control

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-16 DOI: 10.1016/j.eswa.2025.126938

Sibin Cai , Jie Fang , Mengyun Xu

Recently, deep reinforcement learning (DRL)-based traffic signal control (TSC) methods have garnered significant attention among researchers, achieving substantial progress. However, current research often focuses on performance improvement, neglecting interpretability. DRL-based TSC methods often face challenges in interpretability. This limitation poses significant obstacles to practical deployment, given the liability and regulatory constraints faced by governmental authorities responsible for traffic management and control. On the other hand, interpretable RL-based TSC methods offer greater flexibility to meet specific requirements. For instance, prioritizing the clearance of vehicles in a particular movement can be easily achieved by assigning higher weights to the state variables associated with that movement. To address this issue, we propose Xlight, an interpretable multi-agent reinforcement learning (MARL) approach for TSC, which enhances interpretability in three key aspects: (a) meticulously designing and selecting the state space, action space, and reward function. Especially, we propose an interpretable reward function for network-wide TSC and prove that maximizing this reward is equivalent to minimizing the average travel time (ATT) in the road network; (b) introducing more practical regulatable (i.e., interpretable) functions as TSC controllers; and (c) employing maximum entropy policy optimization, which simultaneously enhances interpretability and improves transferability. Next, to better align with practical applications of network-wide TSC, we propose several interpretable MARL-based methods. Among these, Multi-Agent Regulatable Soft Actor-Critic (MARSAC) not only possesses interpretability but also achieves superior performance. Finally, comprehensive experiments conducted across various TSC scenarios, including isolated intersection, synthetic network-wide intersections, and real-world network-wide intersections, demonstrate the effectiveness. For example, in terms of the ATT metric, our proposed method achieves improvements of 9.55%, 34.17%, 3.98%, and 42.93% compared to the Actuated Traffic Signal Control (ATSC) across a synthetic road network and 3 real-world road networks. Furthermore, in the synthetic network, our method demonstrates improvements of 4.04% and 3.21% in the Safety Score and Fuel Consumption metrics, respectively, when compared to the ATSC.

{"title":"XLight: An interpretable multi-agent reinforcement learning approach for traffic signal control","authors":"Sibin Cai , Jie Fang , Mengyun Xu","doi":"10.1016/j.eswa.2025.126938","DOIUrl":"10.1016/j.eswa.2025.126938","url":null,"abstract":"<div><div>Recently, deep reinforcement learning (DRL)-based traffic signal control (TSC) methods have garnered significant attention among researchers, achieving substantial progress. However, current research often focuses on performance improvement, neglecting interpretability. DRL-based TSC methods often face challenges in interpretability. This limitation poses significant obstacles to practical deployment, given the liability and regulatory constraints faced by governmental authorities responsible for traffic management and control. On the other hand, interpretable RL-based TSC methods offer greater flexibility to meet specific requirements. For instance, prioritizing the clearance of vehicles in a particular movement can be easily achieved by assigning higher weights to the state variables associated with that movement. To address this issue, we propose <strong><em>Xlight</em></strong>, an interpretable multi-agent reinforcement learning (MARL) approach for TSC, which enhances interpretability in three key aspects: (a) meticulously designing and selecting the state space, action space, and reward function. Especially, we propose an interpretable reward function for network-wide TSC and prove that maximizing this reward is equivalent to minimizing the average travel time (ATT) in the road network; (b) introducing more practical regulatable (i.e., interpretable) functions as TSC controllers; and (c) employing maximum entropy policy optimization, which simultaneously enhances interpretability and improves transferability. Next, to better align with practical applications of network-wide TSC, we propose several interpretable MARL-based methods. Among these, Multi-Agent Regulatable Soft Actor-Critic (MARSAC) not only possesses interpretability but also achieves superior performance. Finally, comprehensive experiments conducted across various TSC scenarios, including isolated intersection, synthetic network-wide intersections, and real-world network-wide intersections, demonstrate the effectiveness. For example, in terms of the ATT metric, our proposed method achieves improvements of 9.55%, 34.17%, 3.98%, and 42.93% compared to the Actuated Traffic Signal Control (ATSC) across a synthetic road network and 3 real-world road networks. Furthermore, in the synthetic network, our method demonstrates improvements of 4.04% and 3.21% in the Safety Score and Fuel Consumption metrics, respectively, when compared to the ATSC.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126938"},"PeriodicalIF":7.5,"publicationDate":"2025-02-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143453669","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Research on global trajectory planning for UAV based on the information interaction and aging mechanism Wolfpack algorithm

IF 7.5 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Expert Systems with Applications

Pub Date : 2025-02-15 DOI: 10.1016/j.eswa.2025.126867

Jinyu Zhang , Xin Ning , Shichao Ma , Rugang Tang

The planning of trajectories for multi-unmanned aerial vehicles (UAVs) has been a topic of intensive research in both military and civilian contexts. It is a crucial aspect of the overall intelligence capabilities of UAV formation systems. In order to enhance the capability of multi-UAVs autonomous trajectory planning and to facilitate attainment of optimal paths in mountainous environments, this paper proposes an information interaction and aging mechanism Wolfpack Algorithm (IIAM-WPA). Firstly, a mission environment model is established using digital elevation modelling technology to simulate the real mountainous environment. Secondly, a trajectory planning model is established by comprehensively considering the terrain, threats and formation security factors. Meanwhile, in order to comprehensively evaluate the planning results, a new composite objective function is proposed. The proposed IIAM-WPA method is finally employed to identify the optimal paths for multiple UAVs. The key improvements to the method are as follows: the initialization effect is enhanced by the Chebyshev chaotic mapping in initialization phase, thereby accelerating the convergence of the population. Furthermore, the aging mechanism of wolves is incorporated into the model to enhance the efficiency of wolf search. Meanwhile, communication between populations is augmented during the encirclement phase, which serves to enhance population diversity. Finally, a selective mutation mechanism is introduced to rescue the population from the local optimum trap. In order to ascertain the effectiveness of the proposed algorithm, the simulation results of UAV trajectory planning under different mission scenarios are presented and compared with various optimization techniques. The simulation results demonstrate that the maximum improvement rate of the proposed algorithm is 96.73% and 4.2% in single UAV and multi-UAV planning tasks, respectively. This further verifies the planning accuracy and efficiency of the IIAM-WPA method and effectively proves the effectiveness of the method in solving UAV trajectory planning problems.

{"title":"Research on global trajectory planning for UAV based on the information interaction and aging mechanism Wolfpack algorithm","authors":"Jinyu Zhang , Xin Ning , Shichao Ma , Rugang Tang","doi":"10.1016/j.eswa.2025.126867","DOIUrl":"10.1016/j.eswa.2025.126867","url":null,"abstract":"<div><div>The planning of trajectories for multi-unmanned aerial vehicles (UAVs) has been a topic of intensive research in both military and civilian contexts. It is a crucial aspect of the overall intelligence capabilities of UAV formation systems. In order to enhance the capability of multi-UAVs autonomous trajectory planning and to facilitate attainment of optimal paths in mountainous environments, this paper proposes an information interaction and aging mechanism Wolfpack Algorithm (IIAM-WPA). Firstly, a mission environment model is established using digital elevation modelling technology to simulate the real mountainous environment. Secondly, a trajectory planning model is established by comprehensively considering the terrain, threats and formation security factors. Meanwhile, in order to comprehensively evaluate the planning results, a new composite objective function is proposed. The proposed IIAM-WPA method is finally employed to identify the optimal paths for multiple UAVs. The key improvements to the method are as follows: the initialization effect is enhanced by the Chebyshev chaotic mapping in initialization phase, thereby accelerating the convergence of the population. Furthermore, the aging mechanism of wolves is incorporated into the model to enhance the efficiency of wolf search. Meanwhile, communication between populations is augmented during the encirclement phase, which serves to enhance population diversity. Finally, a selective mutation mechanism is introduced to rescue the population from the local optimum trap. In order to ascertain the effectiveness of the proposed algorithm, the simulation results of UAV trajectory planning under different mission scenarios are presented and compared with various optimization techniques. The simulation results demonstrate that the maximum improvement rate of the proposed algorithm is 96.73% and 4.2% in single UAV and multi-UAV planning tasks, respectively. This further verifies the planning accuracy and efficiency of the IIAM-WPA method and effectively proves the effectiveness of the method in solving UAV trajectory planning problems.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"273 ","pages":"Article 126867"},"PeriodicalIF":7.5,"publicationDate":"2025-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143438084","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0