Knowledge-Based Systems最新文献_第7页

Multi-task ordinal regression with task weight discovery 多任务顺序回归与任务权重发现

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-15 DOI: 10.1016/j.knosys.2024.112616

Yanshan Xiao , Mengyue Zeng , Bo Liu , Liang Zhao , Xiangjun Kong , Zhifeng Hao

Ordinal regression (OR) deals with the classification problems that the classes are ranked in order. At present, most OR approaches are designed for individual tasks, the research on multi-task OR is limited. These multi-task OR approaches assume that different tasks have the same relatedness and contribute equally to the overall model. However, in practice, different tasks may have distinct relatedness to the overall model. If they are treated equally, the performance of the overall model may be restricted. In this paper, we propose a novel multi-task OR approach with task weight discovery (MORTD). We assign each task a weight that indicates its relatedness to the overall model. Based on the task weights, a maximum margin multi-task OR model is constructed. Then, we adopt a heuristic framework to construct the multi-task OR classifier and update the task weights alternately. In this framework, the dual coordinate descent method is adapted to train the multi-task OR classifier efficiently. In real-world OR applications, the relatedness of multiple tasks may not be exactly the same. The contribution of MORTD is that it can discover the weights of tasks to yield a more precise classification model. Substantial experiments on real-life OR datasets illustrate that compared to the existing multi-task OR methods, MORTD is able to deliver higher classification accuracy and meanwhile needs less training time.

序数回归（Ordinal Regression，OR）处理的是按顺序排列类别的分类问题。目前，大多数正序回归方法都是针对单个任务设计的，对多任务正序回归的研究还很有限。这些多任务回归方法假定不同任务具有相同的相关性，对整体模型的贡献相同。然而，在实践中，不同的任务可能与整体模型有不同的相关性。如果对它们一视同仁，整体模型的性能可能会受到限制。在本文中，我们提出了一种带有任务权重发现（MORTD）的新型多任务 OR 方法。我们为每个任务分配一个权重，表明其与整体模型的相关性。根据任务权重，我们构建了一个最大裕度多任务 OR 模型。然后，我们采用启发式框架来构建多任务 OR 分类器，并交替更新任务权重。在这个框架中，双坐标下降法被用来高效地训练多任务 OR 分类器。在现实世界的 OR 应用中，多个任务的相关性可能并不完全相同。MORTD 的贡献在于它能发现任务的权重，从而产生更精确的分类模型。在现实生活中的 OR 数据集上进行的大量实验表明，与现有的多任务 OR 方法相比，MORTD 能够提供更高的分类精度，同时所需的训练时间也更短。

{"title":"Multi-task ordinal regression with task weight discovery","authors":"Yanshan Xiao , Mengyue Zeng , Bo Liu , Liang Zhao , Xiangjun Kong , Zhifeng Hao","doi":"10.1016/j.knosys.2024.112616","DOIUrl":"10.1016/j.knosys.2024.112616","url":null,"abstract":"<div><div>Ordinal regression (OR) deals with the classification problems that the classes are ranked in order. At present, most OR approaches are designed for individual tasks, the research on multi-task OR is limited. These multi-task OR approaches assume that different tasks have the same relatedness and contribute equally to the overall model. However, in practice, different tasks may have distinct relatedness to the overall model. If they are treated equally, the performance of the overall model may be restricted. In this paper, we propose a novel multi-task OR approach with task weight discovery (MORTD). We assign each task a weight that indicates its relatedness to the overall model. Based on the task weights, a maximum margin multi-task OR model is constructed. Then, we adopt a heuristic framework to construct the multi-task OR classifier and update the task weights alternately. In this framework, the dual coordinate descent method is adapted to train the multi-task OR classifier efficiently. In real-world OR applications, the relatedness of multiple tasks may not be exactly the same. The contribution of MORTD is that it can discover the weights of tasks to yield a more precise classification model. Substantial experiments on real-life OR datasets illustrate that compared to the existing multi-task OR methods, MORTD is able to deliver higher classification accuracy and meanwhile needs less training time.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112616"},"PeriodicalIF":7.2,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529374","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

Improving quaternion neural networks with quaternionic activation functions 用四元激活函数改进四元神经网络

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-15 DOI: 10.1016/j.knosys.2024.112619

Johannes Pöppelbaum, Andreas Schwung

In this paper, we propose novel quaternion activation functions where we modify either the quaternion magnitude or the phase, as an alternative to the commonly used split activation functions. We define criteria that are relevant for quaternion activation functions, and subsequently we propose our novel activation functions based on this analysis. Instead of applying a known activation function like the ReLU or Tanh on the quaternion elements separately, these activation functions consider the quaternion properties and respect the quaternion space

H

. In particular, all quaternion components are utilized to calculate all output components, carrying out the benefit of the Hamilton product in e.g. the quaternion convolution to the activation functions. The proposed activation functions can be incorporated in arbitrary quaternion valued neural networks trained with gradient descent techniques. We further discuss the derivatives of the proposed activation functions where we observe beneficial properties for the activation functions affecting the phase. Specifically, they prove to be sensitive on basically the whole input range, thus improved gradient flow can be expected. We provide an elaborate experimental evaluation of our proposed quaternion activation functions including comparison with the split ReLU and split Tanh on two image classification tasks using the CIFAR-10 and SVHN dataset. There, especially the quaternion activation functions affecting the phase consistently prove to provide better performance.

在本文中，我们提出了新的四元数激活函数，通过修改四元数幅度或相位来替代常用的分割激活函数。我们定义了与四元数激活函数相关的标准，并在此基础上提出了我们的新型激活函数。这些激活函数没有将已知的激活函数（如 ReLU 或 Tanh）单独应用于四元数元素，而是考虑了四元数属性并尊重四元数空间 H。我们提出的激活函数可用于使用梯度下降技术训练的任意四元数神经网络。我们进一步讨论了所提出的激活函数的导数，观察到激活函数影响相位的有利特性。具体来说，它们基本上对整个输入范围都很敏感，因此有望改善梯度流。我们对所提出的四元数激活函数进行了详细的实验评估，包括在使用 CIFAR-10 和 SVHN 数据集进行的两项图像分类任务中与拆分 ReLU 和拆分 Tanh 的比较。结果表明，尤其是影响相位的四元数激活函数始终能提供更好的性能。

{"title":"Improving quaternion neural networks with quaternionic activation functions","authors":"Johannes Pöppelbaum, Andreas Schwung","doi":"10.1016/j.knosys.2024.112619","DOIUrl":"10.1016/j.knosys.2024.112619","url":null,"abstract":"<div><div>In this paper, we propose novel quaternion activation functions where we modify either the quaternion magnitude or the phase, as an alternative to the commonly used split activation functions. We define criteria that are relevant for quaternion activation functions, and subsequently we propose our novel activation functions based on this analysis. Instead of applying a known activation function like the ReLU or Tanh on the quaternion elements separately, these activation functions consider the quaternion properties and respect the quaternion space <span><math><mi>H</mi></math></span>. In particular, all quaternion components are utilized to calculate all output components, carrying out the benefit of the Hamilton product in e.g. the quaternion convolution to the activation functions. The proposed activation functions can be incorporated in arbitrary quaternion valued neural networks trained with gradient descent techniques. We further discuss the derivatives of the proposed activation functions where we observe beneficial properties for the activation functions affecting the phase. Specifically, they prove to be sensitive on basically the whole input range, thus improved gradient flow can be expected. We provide an elaborate experimental evaluation of our proposed quaternion activation functions including comparison with the split ReLU and split Tanh on two image classification tasks using the CIFAR-10 and SVHN dataset. There, especially the quaternion activation functions affecting the phase consistently prove to provide better performance.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112619"},"PeriodicalIF":7.2,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529488","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}

引用次数: 0

INCOMPLETE multi-view clustering based on low-rank adaptive graph learning 基于低秩自适应图学习的 INCOMPLETE 多视角聚类

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-15 DOI: 10.1016/j.knosys.2024.112562

Jingyu Zhu , Minghua Wan , Guowei Yang , Zhangjing Yang

The challenge of acquiring complete data has led to substantial progress in incomplete multi-view clustering (IMVC) methods. Because graph structures can be excellent representations of data structure relationships, exceptional performance in handling incomplete data is demonstrated by graph-based methods at present. However, these methods still have their limitations. Most incomplete multi-view algorithms primarily focus on local information, neglecting global information. Therefore, these methods cannot dynamically recover the structural relationships in incomplete data by harnessing potential information from multiple perspectives and overall structural information. In response to the aforementioned concerns, we introduced an IMVC based on low-rank adaptive graph learning (IMVC-LAGL). This method initially constructs an affinity matrix based on the inter-view adjacency relationships. It also utilizes tensor low-rank constraints and consensus representation learning to explore higher-order correlations among different views. Subsequently, it adaptively reconstructs the incomplete graph structure to ultimately obtain a complete affinity relationship. It leads to excellent clustering results by integrating relevant information within views, overall structural information and potential information from multiple perspectives. We conducted experiments comparing our algorithm with eight incomplete multi-view algorithms using five different evaluation metrics. The results show that our algorithm achieves the best clustering results across eight datasets with varying missing rates. Particularly in the BBCSport dataset and YaleB dataset, the clustering accuracy of our algorithm is improved by 19.83 % and 16.41 %, respectively, compared with the second-best algorithm, under a 50 % missing rate.

获取完整数据的挑战促使不完整多视图聚类（IMVC）方法取得了重大进展。由于图结构可以很好地表示数据结构关系，目前基于图的方法在处理不完整数据方面表现出了卓越的性能。然而，这些方法仍有其局限性。大多数不完整多视图算法主要关注局部信息，而忽略了全局信息。因此，这些方法无法通过利用多视角的潜在信息和整体结构信息来动态恢复不完整数据中的结构关系。针对上述问题，我们引入了基于低秩自适应图学习的 IMVC（IMVC-LAGL）。这种方法首先根据视图间的邻接关系构建一个亲和矩阵。它还利用张量低阶约束和共识表示学习来探索不同视图之间的高阶相关性。随后，它自适应地重建不完整的图结构，最终获得完整的亲缘关系。通过整合视图内的相关信息、整体结构信息和来自多个视角的潜在信息，该算法能带来出色的聚类结果。我们使用五种不同的评价指标对我们的算法和八种不完整多视图算法进行了实验比较。结果表明，我们的算法在具有不同缺失率的八个数据集上取得了最佳聚类结果。特别是在 BBCSport 数据集和 YaleB 数据集中，在缺失率为 50% 的情况下，我们算法的聚类准确率比第二好的算法分别提高了 19.83% 和 16.41%。

{"title":"INCOMPLETE multi-view clustering based on low-rank adaptive graph learning","authors":"Jingyu Zhu , Minghua Wan , Guowei Yang , Zhangjing Yang","doi":"10.1016/j.knosys.2024.112562","DOIUrl":"10.1016/j.knosys.2024.112562","url":null,"abstract":"<div><div>The challenge of acquiring complete data has led to substantial progress in incomplete multi-view clustering (IMVC) methods. Because graph structures can be excellent representations of data structure relationships, exceptional performance in handling incomplete data is demonstrated by graph-based methods at present. However, these methods still have their limitations. Most incomplete multi-view algorithms primarily focus on local information, neglecting global information. Therefore, these methods cannot dynamically recover the structural relationships in incomplete data by harnessing potential information from multiple perspectives and overall structural information. In response to the aforementioned concerns, we introduced an IMVC based on low-rank adaptive graph learning (IMVC-LAGL). This method initially constructs an affinity matrix based on the inter-view adjacency relationships. It also utilizes tensor low-rank constraints and consensus representation learning to explore higher-order correlations among different views. Subsequently, it adaptively reconstructs the incomplete graph structure to ultimately obtain a complete affinity relationship. It leads to excellent clustering results by integrating relevant information within views, overall structural information and potential information from multiple perspectives. We conducted experiments comparing our algorithm with eight incomplete multi-view algorithms using five different evaluation metrics. The results show that our algorithm achieves the best clustering results across eight datasets with varying missing rates. Particularly in the BBCSport dataset and YaleB dataset, the clustering accuracy of our algorithm is improved by 19.83 % and 16.41 %, respectively, compared with the second-best algorithm, under a 50 % missing rate.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112562"},"PeriodicalIF":7.2,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529401","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

Similarity-driven adversarial testing of neural networks 神经网络的相似性驱动对抗测试

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-15 DOI: 10.1016/j.knosys.2024.112621

Katarzyna Filus, Joanna Domańska

Although Convolutional Neural Networks (CNNs) are among the most important algorithms of computer vision and the artificial intelligence-based systems, they are vulnerable to adversarial attacks. Such attacks can cause dangerous consequences in real-life deployments. Consequently, testing of the artificial intelligence-based systems from their perspective is crucial to reliably support human prediction and decision-making through computation techniques under varying conditions. While proposing new effective attacks is important for neural network testing, it is also crucial to design effective strategies that can be used to choose target labels for these attacks. That is why, in this paper we propose a novel similarity-driven adversarial testing methodology for target label choosing. Our motivation is that CNNs, similarly to humans, tend to make mistakes mostly among categories they perceive similar. Thus, the effort to make models predict a particular class is not equal for all classes. Motivated by this, we propose to use the most and least similar labels to the ground truth according to different similarity measures to choose the target label for an adversarial attack. They can be treated as best- and worst-case scenarios in practical and transparent testing methodologies. As similarity is one of the key components of human cognition and categorization, the approach presents a shift towards a more human-centered security testing of deep neural networks. The obtained numerical results show the superiority of the proposed methods to the existing strategies in the targeted and the non-targeted testing setups.

虽然卷积神经网络（CNN）是计算机视觉和人工智能系统中最重要的算法之一，但它们很容易受到恶意攻击。这种攻击会在实际部署中造成危险后果。因此，从它们的角度对基于人工智能的系统进行测试，对于在不同条件下通过计算技术可靠地支持人类预测和决策至关重要。虽然提出新的有效攻击对于神经网络测试很重要，但设计有效的策略来选择这些攻击的目标标签也很关键。因此，我们在本文中提出了一种用于选择目标标签的新型相似性驱动对抗测试方法。我们的动机是，CNN 与人类类似，往往会在它们认为相似的类别中犯大部分错误。因此，让模型预测特定类别的努力并不等同于预测所有类别。受此启发，我们建议根据不同的相似性度量，使用与地面实况最相似和最不相似的标签来选择对抗性攻击的目标标签。在实用、透明的测试方法中，它们可被视为最佳和最差情况。由于相似性是人类认知和分类的关键要素之一，该方法提出了一种以人为本的深度神经网络安全测试方法。所获得的数值结果表明，在目标和非目标测试设置中，所提出的方法优于现有的策略。

{"title":"Similarity-driven adversarial testing of neural networks","authors":"Katarzyna Filus, Joanna Domańska","doi":"10.1016/j.knosys.2024.112621","DOIUrl":"10.1016/j.knosys.2024.112621","url":null,"abstract":"<div><div>Although Convolutional Neural Networks (CNNs) are among the most important algorithms of computer vision and the artificial intelligence-based systems, they are vulnerable to adversarial attacks. Such attacks can cause dangerous consequences in real-life deployments. Consequently, testing of the artificial intelligence-based systems from their perspective is crucial to reliably support human prediction and decision-making through computation techniques under varying conditions. While proposing new effective attacks is important for neural network testing, it is also crucial to design effective strategies that can be used to choose target labels for these attacks. That is why, in this paper we propose a novel similarity-driven adversarial testing methodology for target label choosing. Our motivation is that CNNs, similarly to humans, tend to make mistakes mostly among categories they perceive similar. Thus, the effort to make models predict a particular class is not equal for all classes. Motivated by this, we propose to use the most and least similar labels to the ground truth according to different similarity measures to choose the target label for an adversarial attack. They can be treated as best- and worst-case scenarios in practical and transparent testing methodologies. As similarity is one of the key components of human cognition and categorization, the approach presents a shift towards a more human-centered security testing of deep neural networks. The obtained numerical results show the superiority of the proposed methods to the existing strategies in the targeted and the non-targeted testing setups.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112621"},"PeriodicalIF":7.2,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142553808","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}

引用次数: 0

A concept fringe-based concept-cognitive learning method in skill context 基于概念边缘的技能背景下的概念认知学习方法

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-15 DOI: 10.1016/j.knosys.2024.112618

Hai-Long Yang , Yin-Feng Zhou , Jin-Jin Li , Weiping Ding

Concept-cognitive learning has achieved remarkable results in simulating the learning of concepts. However, the existing concept-cognitive learning models mainly focus on how knowledge is acquired, but ignore the fact that knowledge transfer and knowledge forgetting may occur during the process of learning skills and solving items. This limits the application of concept-cognitive learning in predicting knowledge states and assessing competence states in skill contexts. To overcome this limitation, this paper provides a new concept-cognitive learning method for property-oriented concepts and object-oriented concepts in skill context. Corresponding to the conjunctive model and the disjunctive model, the inner and outer fringes of property-oriented concept and object-oriented concept are first defined, respectively. In this way, items or skills that are easily forgotten and those that are in the zone of proximal development can be found under both models. Furthermore, the Jaccard similarity coefficient is used to diversify the learning outcomes by finding items and skills that are most likely to occur knowledge forgetting or knowledge transfer. Thus, based on the fringes of concepts, the algorithms to learn property-oriented concepts and object-oriented concepts are provided, respectively. Finally, the case study on a real world example and the experimental evaluation on six data sets from UCI demonstrate that the proposed method is of practical significance and effective in terms of running time.

概念认知学习在模拟概念学习方面取得了显著成效。然而，现有的概念-认知学习模型主要关注知识是如何获得的，却忽视了在学习技能和解决题目的过程中可能发生的知识迁移和知识遗忘。这就限制了概念认知学习在预测知识状态和评估技能情境中的能力状态方面的应用。为了克服这一局限，本文针对技能情境中的面向属性概念和面向对象概念提出了一种新的概念认知学习方法。与连接模型和非连接模型相对应，首先分别定义了面向属性概念和面向对象概念的内缘和外缘。这样，容易遗忘的项目或技能和处于最近发展区的项目或技能就可以在这两种模式下找到。此外，通过 Jaccard 相似系数，找到最有可能发生知识遗忘或知识迁移的项目和技能，从而实现学习成果的多样化。因此，基于概念的边缘，分别提供了学习面向属性概念和面向对象概念的算法。最后，对一个真实世界的例子进行了案例研究，并对来自 UCI 的六个数据集进行了实验评估，证明所提出的方法在运行时间方面具有实际意义和有效性。

{"title":"A concept fringe-based concept-cognitive learning method in skill context","authors":"Hai-Long Yang , Yin-Feng Zhou , Jin-Jin Li , Weiping Ding","doi":"10.1016/j.knosys.2024.112618","DOIUrl":"10.1016/j.knosys.2024.112618","url":null,"abstract":"<div><div>Concept-cognitive learning has achieved remarkable results in simulating the learning of concepts. However, the existing concept-cognitive learning models mainly focus on how knowledge is acquired, but ignore the fact that knowledge transfer and knowledge forgetting may occur during the process of learning skills and solving items. This limits the application of concept-cognitive learning in predicting knowledge states and assessing competence states in skill contexts. To overcome this limitation, this paper provides a new concept-cognitive learning method for property-oriented concepts and object-oriented concepts in skill context. Corresponding to the conjunctive model and the disjunctive model, the inner and outer fringes of property-oriented concept and object-oriented concept are first defined, respectively. In this way, items or skills that are easily forgotten and those that are in the zone of proximal development can be found under both models. Furthermore, the Jaccard similarity coefficient is used to diversify the learning outcomes by finding items and skills that are most likely to occur knowledge forgetting or knowledge transfer. Thus, based on the fringes of concepts, the algorithms to learn property-oriented concepts and object-oriented concepts are provided, respectively. Finally, the case study on a real world example and the experimental evaluation on six data sets from UCI demonstrate that the proposed method is of practical significance and effective in terms of running time.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112618"},"PeriodicalIF":7.2,"publicationDate":"2024-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529405","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

Trust number: Trust-based modeling for handling decision-making problems 信任数字基于信任的决策问题处理模型

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-13 DOI: 10.1016/j.knosys.2024.112631

Saeid Jafarzadeh Ghoushchi , Abbas Mardani , Luis Martínez

Fuzzy sets play an effective role in dealing with the uncertainty and ambiguity of input data in real-world decision-making problems. Nevertheless, the effectiveness of fuzzy sets becomes unreliable and even more uncertain when the input data come from untrustworthy sources. Therefore, a new measurement could be considered based on the data's degree of trust to reduce the deviation of unreliable information in fuzzy decision-making problems. The main aim of this study is to introduce a new information modeling called trust numbers (T-numbers), which models variations and deviations associated with triangular fuzzy numbers and their application to decision-making. In addition, it introduces new operations on T-numbers to develop a decision model based on this theory. The performance of this model was analyzed through its implementation in two case studies and by comparing the fuzzy technique for order of Preference by similarity to the ideal solution (F-TOPSIS) and its T-number extension(T-TOPSIS). Results indicate that T-numbers can be applied to classical fuzzy numbers when the available information is uncertain and a degree of distrust exists.

在现实世界的决策问题中，模糊集在处理输入数据的不确定性和模糊性方面发挥着有效的作用。然而，当输入数据来源不可信时，模糊集的有效性就会变得不可靠，甚至更加不确定。因此，可以考虑采用一种基于数据可信度的新测量方法，以减少模糊决策问题中不可靠信息的偏差。本研究的主要目的是引入一种新的信息模型，即信任数（T-numbers），它可以模拟与三角模糊数相关的变化和偏差，并将其应用于决策。此外，它还引入了 T 数的新运算，以开发基于该理论的决策模型。通过在两个案例研究中实施该模型，并通过比较与理想解决方案相似性排序的模糊技术（F-TOPSIS）及其 T 数扩展（T-TOPSIS），分析了该模型的性能。结果表明，当可用信息不确定且存在一定程度的不信任时，T 数可应用于经典模糊数。

{"title":"Trust number: Trust-based modeling for handling decision-making problems","authors":"Saeid Jafarzadeh Ghoushchi , Abbas Mardani , Luis Martínez","doi":"10.1016/j.knosys.2024.112631","DOIUrl":"10.1016/j.knosys.2024.112631","url":null,"abstract":"<div><div>Fuzzy sets play an effective role in dealing with the uncertainty and ambiguity of input data in real-world decision-making problems. Nevertheless, the effectiveness of fuzzy sets becomes unreliable and even more uncertain when the input data come from untrustworthy sources. Therefore, a new measurement could be considered based on the data's degree of trust to reduce the deviation of unreliable information in fuzzy decision-making problems. The main aim of this study is to introduce a new information modeling called trust numbers (T-numbers), which models variations and deviations associated with triangular fuzzy numbers and their application to decision-making. In addition, it introduces new operations on T-numbers to develop a decision model based on this theory. The performance of this model was analyzed through its implementation in two case studies and by comparing the fuzzy technique for order of Preference by similarity to the ideal solution (F-TOPSIS) and its T-number extension(T-TOPSIS). Results indicate that T-numbers can be applied to classical fuzzy numbers when the available information is uncertain and a degree of distrust exists.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112631"},"PeriodicalIF":7.2,"publicationDate":"2024-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529486","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}

引用次数: 0

CBRec: A causal way balancing multidimensional attraction effect in POI recommendations CBRec：以因果关系平衡 POI 推荐中的多维吸引效应

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112607

Bo Liu, Jun Zeng, Junhao Wen, Min Gao, Wei Zhou

In the next Point-of-Interest recommendation, sparse and uneven location data generate biases, resulting in homogeneous recommendation outcomes that fail to reflect user preferences. Although there are many related unbiased studies, they still exhibit limitations. They lack a unified debiasing paradigm and typically employ different methods to address various biases, resulting in complex and incompatible debiasing models. Additionally, they often overlook the potential advantages of biases, thus harming the quality of location features. To address these challenges, we propose a unified debiasing paradigm by intervening in location attraction to balance the positive and negative effects of bias. By analyzing the structural causal graph, we identify attraction as a feature influenced by bias. By comparing observational results affected by attraction with counterfactual results unaffected by it, we derive a unified debiasing paradigm that eliminates the effects of bias. Additionally, through feature fusion, we embed multidimensional attraction into user features, leveraging the advantages of bias to preserve the quality of location features. Finally, experimental results on five real-world datasets demonstrate that our proposed model outperforms recent sequential recommendation models.

在下一个兴趣点推荐中，稀疏且不均匀的位置数据会产生偏差，导致推荐结果千篇一律，无法反映用户偏好。虽然有很多相关的无偏见研究，但它们仍然存在局限性。它们缺乏统一的去偏差范式，通常采用不同的方法来解决各种偏差，导致去偏差模型复杂且不兼容。此外，它们往往忽视了偏差的潜在优势，从而损害了位置特征的质量。为了应对这些挑战，我们提出了一种统一的去除法范式，通过干预位置吸引力来平衡偏差的正负效应。通过分析结构因果图，我们确定吸引力是受偏差影响的一个特征。通过比较受吸引力影响的观察结果和不受吸引力影响的反事实结果，我们得出了一种统一的去偏差范式，可以消除偏差的影响。此外，通过特征融合，我们将多维吸引力嵌入到用户特征中，利用偏差的优势来保持位置特征的质量。最后，在五个真实世界数据集上的实验结果表明，我们提出的模型优于最近的顺序推荐模型。

{"title":"CBRec: A causal way balancing multidimensional attraction effect in POI recommendations","authors":"Bo Liu, Jun Zeng, Junhao Wen, Min Gao, Wei Zhou","doi":"10.1016/j.knosys.2024.112607","DOIUrl":"10.1016/j.knosys.2024.112607","url":null,"abstract":"<div><div>In the next Point-of-Interest recommendation, sparse and uneven location data generate biases, resulting in homogeneous recommendation outcomes that fail to reflect user preferences. Although there are many related unbiased studies, they still exhibit limitations. They lack a unified debiasing paradigm and typically employ different methods to address various biases, resulting in complex and incompatible debiasing models. Additionally, they often overlook the potential advantages of biases, thus harming the quality of location features. To address these challenges, we propose a unified debiasing paradigm by intervening in location attraction to balance the positive and negative effects of bias. By analyzing the structural causal graph, we identify attraction as a feature influenced by bias. By comparing observational results affected by attraction with counterfactual results unaffected by it, we derive a unified debiasing paradigm that eliminates the effects of bias. Additionally, through feature fusion, we embed multidimensional attraction into user features, leveraging the advantages of bias to preserve the quality of location features. Finally, experimental results on five real-world datasets demonstrate that our proposed model outperforms recent sequential recommendation models.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112607"},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142446717","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

Local density based on weighted K-nearest neighbors for density peaks clustering 基于加权 K 近邻的局部密度，用于密度峰聚类

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112609

Sifan Ding , Min Li , Tianyi Huang , William Zhu

Density peaks clustering (DPC), a traditional density-based clustering algorithm, has received considerable attention in recent years. DPC identifies clusters by designating density peaks, defined by local density, as cluster centers. However, DPC and its variants often struggle to identify high-density peaks, particularly in datasets with arbitrarily complex shapes. To address this issue, we propose a novel local density measure based on weighted K-nearest neighbors (KNN). First, we construct a new similarity measure, termed the constrained kernel rank-order distance, to determine the KNNs of each point. Next, we develop the concept of weighted KNNs by assigning a weight to each point, representing the probability of it becoming a KNN to other points. Subsequently, we redefine the local density based on the weighted KNN. Finally, we integrate this new local density measure into the DPC framework. Experiments demonstrate that the proposed algorithm outperforms existing DPC algorithms in terms of effectiveness. The source code can be downloaded from https://github.com/Gedanke/dpcCode.

密度峰聚类（DPC）是一种传统的基于密度的聚类算法，近年来受到广泛关注。DPC 通过指定由局部密度定义的密度峰作为聚类中心来识别聚类。然而，DPC 及其变体往往难以识别高密度峰，尤其是在具有任意复杂形状的数据集中。为了解决这个问题，我们提出了一种基于加权 K 近邻（KNN）的新型局部密度测量方法。首先，我们构建了一种新的相似性度量，称为约束核秩距离，用于确定每个点的 KNN。接下来，我们为每个点分配一个权重，代表该点成为其他点的 KNN 的概率，从而发展了加权 KNN 的概念。随后，我们根据加权 KNN 重新定义本地密度。最后，我们将这种新的局部密度测量方法整合到 DPC 框架中。实验证明，所提出的算法在有效性方面优于现有的 DPC 算法。源代码可从 https://github.com/Gedanke/dpcCode 下载。

{"title":"Local density based on weighted K-nearest neighbors for density peaks clustering","authors":"Sifan Ding , Min Li , Tianyi Huang , William Zhu","doi":"10.1016/j.knosys.2024.112609","DOIUrl":"10.1016/j.knosys.2024.112609","url":null,"abstract":"<div><div>Density peaks clustering (DPC), a traditional density-based clustering algorithm, has received considerable attention in recent years. DPC identifies clusters by designating density peaks, defined by local density, as cluster centers. However, DPC and its variants often struggle to identify high-density peaks, particularly in datasets with arbitrarily complex shapes. To address this issue, we propose a novel local density measure based on weighted K-nearest neighbors (KNN). First, we construct a new similarity measure, termed the constrained kernel rank-order distance, to determine the KNNs of each point. Next, we develop the concept of weighted KNNs by assigning a weight to each point, representing the probability of it becoming a KNN to other points. Subsequently, we redefine the local density based on the weighted KNN. Finally, we integrate this new local density measure into the DPC framework. Experiments demonstrate that the proposed algorithm outperforms existing DPC algorithms in terms of effectiveness. The source code can be downloaded from <span><span>https://github.com/Gedanke/dpcCode</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112609"},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142529376","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

GA-SmaAt-GNet: Generative adversarial small attention GNet for extreme precipitation nowcasting GA-SmaAt-GNet：用于极端降水预报的生成式对抗小注意 GNet

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112612

Eloy Reulen, Jie Shi, Siamak Mehrkanoon

In recent years, data-driven modeling approaches have gained significant attention across various meteorological applications, particularly in weather forecasting. However, these methods often face challenges in handling extreme weather conditions. In response, we present the GA-SmaAt-GNet model, a novel generative adversarial framework for extreme precipitation nowcasting. This model features a unique SmaAt-GNet generator, an extension of the successful SmaAt-UNet architecture, capable of integrating precipitation masks (binarized precipitation maps) to enhance predictive accuracy. Additionally, GA-SmaAt-GNet incorporates an attention-augmented discriminator inspired by the Pix2Pix architecture. This innovative framework paves the way for generative precipitation nowcasting using multiple data sources. We evaluate the performance of SmaAt-GNet and GA-SmaAt-GNet using real-life precipitation data from The Netherlands, revealing notable improvements in overall performance and for extreme precipitation events compared to other models. Specifically, our proposed architecture demonstrates its main performance gain in summer and autumn, when precipitation intensity is typically at its peak. Furthermore, we conduct uncertainty analysis on the GA-SmaAt-GNet model and the precipitation dataset, providing insights into its predictive capabilities. Finally, we employ Grad-CAM to offer visual explanations of our model’s predictions, generating activation heatmaps that highlight areas of input activation throughout the network.

近年来，数据驱动建模方法在各种气象应用中，特别是在天气预报中，受到了极大的关注。然而，这些方法在处理极端天气条件时往往面临挑战。为此，我们提出了 GA-SmaAt-GNet 模型，这是一个用于极端降水预报的新型生成对抗框架。该模型具有独特的 SmaAt-GNet 生成器，是成功的 SmaAt-UNet 架构的扩展，能够整合降水掩码（二值化降水图）以提高预测精度。此外，GA-SmaAt-GNet 还采用了受 Pix2Pix 架构启发的注意力增强判别器。这一创新框架为使用多种数据源进行降水预报铺平了道路。我们使用荷兰的真实降水数据评估了 SmaAt-GNet 和 GA-SmaAt-GNet 的性能，结果显示，与其他模型相比，它们在整体性能和极端降水事件方面都有显著提高。具体来说，我们提出的架构在降水强度通常达到峰值的夏季和秋季表现出了主要的性能优势。此外，我们还对 GA-SmaAt-GNet 模型和降水数据集进行了不确定性分析，从而深入了解其预测能力。最后，我们利用 Grad-CAM 对模型的预测结果进行可视化解释，生成激活热图，突出显示整个网络的输入激活区域。

{"title":"GA-SmaAt-GNet: Generative adversarial small attention GNet for extreme precipitation nowcasting","authors":"Eloy Reulen, Jie Shi, Siamak Mehrkanoon","doi":"10.1016/j.knosys.2024.112612","DOIUrl":"10.1016/j.knosys.2024.112612","url":null,"abstract":"<div><div>In recent years, data-driven modeling approaches have gained significant attention across various meteorological applications, particularly in weather forecasting. However, these methods often face challenges in handling extreme weather conditions. In response, we present the GA-SmaAt-GNet model, a novel generative adversarial framework for extreme precipitation nowcasting. This model features a unique SmaAt-GNet generator, an extension of the successful SmaAt-UNet architecture, capable of integrating precipitation masks (binarized precipitation maps) to enhance predictive accuracy. Additionally, GA-SmaAt-GNet incorporates an attention-augmented discriminator inspired by the Pix2Pix architecture. This innovative framework paves the way for generative precipitation nowcasting using multiple data sources. We evaluate the performance of SmaAt-GNet and GA-SmaAt-GNet using real-life precipitation data from The Netherlands, revealing notable improvements in overall performance and for extreme precipitation events compared to other models. Specifically, our proposed architecture demonstrates its main performance gain in summer and autumn, when precipitation intensity is typically at its peak. Furthermore, we conduct uncertainty analysis on the GA-SmaAt-GNet model and the precipitation dataset, providing insights into its predictive capabilities. Finally, we employ Grad-CAM to offer visual explanations of our model’s predictions, generating activation heatmaps that highlight areas of input activation throughout the network.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112612"},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142446715","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}

引用次数: 0

Utilizing Bayesian generalization network for reliable fault diagnosis of machinery with limited data 利用贝叶斯泛化网络对数据有限的机械进行可靠的故障诊断

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Knowledge-Based Systems

Pub Date : 2024-10-11 DOI: 10.1016/j.knosys.2024.112628

Minjie Feng , Haidong Shao , Minghui Shao , Yiming Xiao , Jie Wang , Bin Liu

To address the issues of overfitting, domain generalization challenges, and lack of credibility brought by limited data samples in mechanical fault diagnosis in practical engineering, this paper proposes a reliable Bayesian generalization network (BGNet). A Bayesian convolutional layer is constructed based on variational inference, treating all parameters in the convolutional layer as random variables. This approach makes a single model function similar to an ensemble of an infinite number of models, and thus enhancing the model's capability of overfitting resistance and domain generalization. The parameters of the variational distribution are updated to approximate the posterior distribution by local reparametrization and Monte Carlo sampling to optimize the evidence lower bound (ELBO) loss. Confidence information is extracted from the model results and, uncertainty estimation and decomposition schemes are designed to provide interpretability. The proposed method is applied to analyze the experimental data of bearing and gearbox faults. The results show that in a multi-source domain scenario with limited samples, the proposed method demonstrates high diagnostic accuracy, effectively describes the relationship between domain variability and uncertainty, and significantly outperforms several benchmark and state-of-the-art models.

为了解决实际工程中机械故障诊断所面临的过拟合、领域泛化挑战以及有限数据样本所带来的可信度不足等问题，本文提出了一种可靠的贝叶斯泛化网络（BGNet）。贝叶斯卷积层的构建基于变异推理，将卷积层中的所有参数都视为随机变量。这种方法使单一模型的功能类似于无限多个模型的集合，从而增强了模型的抗过拟合能力和领域泛化能力。通过局部重参数化和蒙特卡洛采样更新变分分布的参数以近似后验分布，从而优化证据下限（ELBO）损失。从模型结果中提取置信度信息，并设计不确定性估计和分解方案，以提供可解释性。提出的方法被用于分析轴承和齿轮箱故障的实验数据。结果表明，在样本有限的多源域情况下，所提出的方法具有很高的诊断准确性，能有效地描述域变异性和不确定性之间的关系，并明显优于几个基准模型和最先进的模型。

{"title":"Utilizing Bayesian generalization network for reliable fault diagnosis of machinery with limited data","authors":"Minjie Feng , Haidong Shao , Minghui Shao , Yiming Xiao , Jie Wang , Bin Liu","doi":"10.1016/j.knosys.2024.112628","DOIUrl":"10.1016/j.knosys.2024.112628","url":null,"abstract":"<div><div>To address the issues of overfitting, domain generalization challenges, and lack of credibility brought by limited data samples in mechanical fault diagnosis in practical engineering, this paper proposes a reliable Bayesian generalization network (BGNet). A Bayesian convolutional layer is constructed based on variational inference, treating all parameters in the convolutional layer as random variables. This approach makes a single model function similar to an ensemble of an infinite number of models, and thus enhancing the model's capability of overfitting resistance and domain generalization. The parameters of the variational distribution are updated to approximate the posterior distribution by local reparametrization and Monte Carlo sampling to optimize the evidence lower bound (ELBO) loss. Confidence information is extracted from the model results and, uncertainty estimation and decomposition schemes are designed to provide interpretability. The proposed method is applied to analyze the experimental data of bearing and gearbox faults. The results show that in a multi-source domain scenario with limited samples, the proposed method demonstrates high diagnostic accuracy, effectively describes the relationship between domain variability and uncertainty, and significantly outperforms several benchmark and state-of-the-art models.</div></div>","PeriodicalId":49939,"journal":{"name":"Knowledge-Based Systems","volume":"305 ","pages":"Article 112628"},"PeriodicalIF":7.2,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142442339","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