Article
Computer Science, Artificial Intelligence
Mohammad Rezaei, Pasi Franti
Summary: We propose two new clustering algorithms, k-sets and k-swaps, for data with set objects. The algorithms calculate the mean of sets in a cluster and the distance between a set and the mean. The k-sets algorithm is derived from classical k-means principles and repeats assignment and update steps until convergence. We introduce the k-swaps variant as a wrapper around k-means to avoid local minima. Experimental results demonstrate that this algorithm provides more accurate clustering results compared to k-medoids and other competitive methods.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Dongqiang Yang, Xinran Yang, Hui Jia, Lixian Xu, Jin Guo
Summary: Uncertainty research is a critical problem in artificial intelligence. This study proposes an incomplete fuzzy linguistic formal context and discusses fuzzy linguistic approximate concepts to deal with missing linguistic-valued information and reduce information loss.
INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Zafaryab Rasool, Sunil Aryal, Mohamed Reda Bouadjenek, Richard Dazeley
Summary: Density Peak Clustering (DPC) is a popular clustering algorithm that uses pairwise similarity to detect arbitrary shaped clusters. However, it is not robust for datasets with different densities and is sensitive to scale changes in data representation. This paper proposes an effective data-dependent similarity measure called MP-Similarity, and integrates it into DPC to create MP-DPC. The experiments show that MP-DPC outperforms DPC with Euclidean distance and existing similarity measures, and is robust to changes in data scales.
PATTERN RECOGNITION
(2023)
Article
Mathematics
B. Srirekha, Shakeela Sathish, R. Narmada Devi, Miroslav Mahdal, Robert Cep, K. Elavarasan
Summary: This paper introduces an object ranking concept to define a consistency set and reduction of the attributes by structural features. An incomplete information system works on the three-way concepts using the SE-ISI Context. Granularity is emphasized with join (meet) irreducible sets using the object ranking concepts. A dual operator is defined based on the object ranking concepts and its properties and conditions are verified. This elaborates on the four kinds of reduction of the attributes.
Article
Green & Sustainable Science & Technology
Qiang Shang, Yang Yu, Tian Xie
Summary: In this paper, a hybrid new traffic state classification method based on unsupervised clustering is proposed. The method utilizes traffic data for clustering to achieve traffic state classification, and it shows superior performance compared to other methods.
Article
Computer Science, Artificial Intelligence
Jiaojiao Yang, Andong Qiu, Zhouwang Yang
Summary: As a widely used unsupervised learning method, clustering model plays an important role in exploring data structures. Spectral analysis is a useful technique for solving clustering problems. However, existing methods have limitations. In this paper, a new fuzzy clustering model is proposed, which reconstructs the Laplacian matrix using geometrically-nearest-neighbor similarity measurement and utilizes an algorithm based on the Alternating Direction Method of Multipliers (ADMM) for optimization.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Xia Ji, Lei Yang, Sheng Yao, Peng Zhao, Xuejun Li
Summary: With the development of data collection technologies, multi-view clustering has become a popular research topic. However, existing methods for incomplete multi-view clustering have limitations such as parameter adjustment, high computational complexity, and inability to handle cases with no paired samples. This paper proposes a Fast and General Incomplete Multi-view Adaptive Clustering method (FGPMAC) that overcomes these limitations and demonstrates superiority in experiments on real datasets.
COGNITIVE COMPUTATION
(2023)
Article
Computer Science, Artificial Intelligence
YunLong Mi, Wenqi Liu, Yong Shi, Jinhai Li
Summary: In human concept learning, semi-supervised learning combines labeled and unlabeled data, and this approach needs to be redesigned for new data input. This study proposes a novel method for dynamic semi-supervised learning using concept spaces and structures, which mimics human cognitive processes.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Jianlun Liu, Shaohua Teng, Lunke Fei, Wei Zhang, Xiaozhao Fang, Zhuxiu Zhang, Naiqi Wu
Summary: This paper proposes a novel Consensus Learning approach to Incomplete Multi-view Clustering (CLIMC), which effectively addresses the limitations of existing methods by exploiting complementary multi-view information and exploring cross-view relations among data points through a consensus similarity graph. Extensive experiments demonstrate the effectiveness of CLIMC over state-of-the-arts.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Artificial Intelligence
Zahra Sadat Sajjadi, Mahdi Esmaeili, Mostafa Ghobaei-Arani, Behrouz Minaei-Bidgoli
Summary: In recent years, researchers have shown increasing interest in using social network data to extract meaningful information, particularly in applications such as link prediction, community detection, and protein module identification. The combination of structural and attribute similarity has been a common approach in graph clustering solutions, but the limited use of node features in sparse social networks remains a challenge. This paper proposes a hybrid clustering approach for link prediction in heterogeneous information networks by considering the weight of direct edges and the correlation between adjacency vectors. The results show that this method effectively reduces entropy and execution time while maintaining cluster density.
KNOWLEDGE AND INFORMATION SYSTEMS
(2023)
Article
Computer Science, Information Systems
Wen-Jue He, Zheng Zhang, Yuhong Wei
Summary: This paper proposes a novel Scalable Incomplete Multi-view Clustering with Adaptive Data Completion (SIMC_ADC) method for large-scale IMC problems. The method combines representative anchor learning, similarity recovery, and adaptive instance completion to achieve promising instance-level restoration.
INFORMATION SCIENCES
(2023)
Article
Engineering, Industrial
Zequan Chen, Guofa Li, Jialong He, Zhaojun Yang, Jili Wang
Summary: A new parallel adaptive structure reliability analysis method called RBIK is proposed in this study, which incorporates a global convergence condition and optimal importance sampling function, and utilizes K-medoids clustering for candidate sample analysis to achieve parallel operation. RBIK distinguishes itself by focusing on rapidly satisfying the global convergence condition of the Kriging model, rather than individual candidate samples.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2022)
Article
Computer Science, Artificial Intelligence
Zheng Zhang, Wen-Jue He
Summary: The success of current multi-view clustering lies in the assumption of completeness on each view, but this assumption is rarely satisfied in real-world applications. Incomplete multi-view clustering (IMC) focuses on clustering partially observed instances, caused by data corruption or sensor failure. However, the current research faces obstacles including the reliance on pairwise similarity measurement, the challenge of recovering and enhancing high-order similarities, and the limitation to dual-view IMC. This paper proposes a Tensorized Topological Graph Learning (TTGL) approach for generalized incomplete multi-view clustering, which considers topological graph construction, missing feature completion, and high-order uncertain correlation enhancement. Experimental results on benchmark datasets show the effectiveness of TTGL compared to state-of-the-art IMC clustering algorithms. Source code for TTGL is available at: https://github.com/DarrenZZhang/TTGL.
INFORMATION FUSION
(2023)
Article
Computer Science, Artificial Intelligence
Krishna Kumar Sharma, Ayan Seal
Summary: Multi-view clustering is gaining more attention due to the presence of multiple views in real-world datasets, providing complementary and consensus information. An adaptive mixture similarity function based on geometric distance and S-divergence is introduced for uncertain data clustering, integrated with k-medoids to reduce the impact of outliers and noises. Extensive experimental results demonstrate the effectiveness and robustness of the proposed method against noise and outliers.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Mathematics, Applied
Li Chen, Shuisheng Zhou, Jiajun Ma, Mingliang Xu
Summary: Kernel-based clustering algorithms have the advantage of identifying and capturing nonlinear structures in datasets, but face challenges in handling large-scale datasets due to memory constraints. In this study, an incomplete Cholesky factorization method is proposed to generate low-rank approximations of kernel matrices, accelerating kernel clustering and saving memory space. Experimental results show that the proposed algorithm achieves similar performance to traditional kernel k-means clustering, but with the capability to handle large-scale datasets.
APPLIED MATHEMATICS AND COMPUTATION
(2021)
Article
Computer Science, Artificial Intelligence
Xiao Zhang, Changlin Mei, Degang Chen, Yanyan Yang, Jinhai Li
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2020)
Article
Computer Science, Artificial Intelligence
Zehua Jiang, Keyu Liu, Jingjing Song, Xibei Yang, Jinhai Li, Yuhua Qian
Summary: This paper proposes a method of crosswise computing reducts, which calculates multiple reducts by grouping and crosswise selecting, and designs an acceleration strategy. Experimental results show that the method can significantly reduce computation time, provide reducts with higher stability, and maintain classification performance.
APPLIED SOFT COMPUTING
(2021)
Article
Automation & Control Systems
Yong Shi, Yunlong Mi, Jinhai Li, Wenqi Liu
Summary: Concept-cognitive learning (CCL) is a new field that focuses on incremental concept learning and dynamic knowledge processing. The existing research aims to address the lack of generalization ability in CCL systems, and proposes a new CCLM model to naturally integrate new data for enhanced concept learning flexibility.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Bo Yang, Jinhai Li
Summary: The study shows that the relationship between the self-questioning dynamical evolutionary game model and Ising model is independent of network structure, and the dividing lines obtained are suitable for arbitrary networks. Nodes with large degree exhibit higher stability and robustness compared to those with small degree.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2021)
Article
Automation & Control Systems
Yunlong Mi, Yong Shi, Jinhai Li, Wenqi Liu, Mengyu Yan
Summary: This article introduces a fuzzy-based concept learning model (FCLM) to address the limitations of standard concept learning algorithms in handling continuous data and ignoring object information. By utilizing concept hierarchical relations in concept lattices, the proposed model achieves state-of-the-art classification performance and demonstrates effectiveness in concept discovery.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Yingsheng Chen, Jinhai Li, Jinjin Li, Rongde Lin, Dongxiao Chen
Summary: This paper explores the issue of optimal scale selection in multi-scale decision information systems, emphasizing the dynamic changes and increasing amount of information in big data. It further investigates the change laws of optimal scale when adding an object, developing sufficient and necessary conditions for updating the optimal scale, making the theoretical study more complete.
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
YunLong Mi, Wenqi Liu, Yong Shi, Jinhai Li
Summary: In human concept learning, semi-supervised learning combines labeled and unlabeled data, and this approach needs to be redesigned for new data input. This study proposes a novel method for dynamic semi-supervised learning using concept spaces and structures, which mimics human cognitive processes.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Min Fan, Shan Luo, Jinhai Li
Summary: Knowledge discovery combined with network structure is an emerging field. This paper proposes a network formal context of three-way decision (NFC3WD) and studies network weaken-concepts and their corresponding sub-networks. Furthermore, the concept logic of the network and the properties of its operators are put forward, and rule extraction algorithms are designed. Moreover, these algorithms are applied to COVID-19 diagnosis examples, showing the superiority of the proposed method.
APPLIED INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Jiaojiao Niu, Degang Chen, Jinhai Li, Hui Wang
Summary: This article focuses on the learning of granular rules and proposes a novel fuzzy rule-based classification model. It improves the readability and efficiency through granular reducts and demonstrates its effectiveness with numerical experiments.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Huilai Zhi, Jinhai Li, Yinan Li
Summary: This study proposes a mathematical framework for multilevel conflict analysis from an outsider's perspective. By utilizing fuzzy formal concept analysis, the uncertainties of two sides can be managed simultaneously. Experimental results show that managing uncertainties can lead to more conflict resolutions, and increasing analysis levels helps achieve more economical conflict resolutions.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Yingsheng Chen, Jinhai Li, Jinjin Li, Dongxiao Chen, Rongde Lin
Summary: The multi-scale decision information system is a typical granular computing model that requires consideration of uncertainty and optimal scale selection. This study investigates the updating law of the local optimal scale under the condition of dynamic object increase and proposes a method to define and update the local optimal scale based on the uncertainty of decision classes. Experimental results demonstrate the correctness and effectiveness of the proposed method in calculating the local optimal scale.
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
(2023)
Article
Computer Science, Artificial Intelligence
Jinhai Li, Ye Feng
Summary: The problem of optimal scale selection for multi-scale decision information systems is an important issue in granular computing research, particularly in the context of dynamic data updates. Scholars have focused on determining the changes in the optimal scale for dynamic data, especially with regards to newly added objects. Existing studies have only addressed the conditions for the optimal scale becoming smaller. Therefore, it is necessary to explore the conditions for the optimal scale remaining unchanged or becoming larger. This paper uses three-way decision theory to study this problem and provides a solution to finding the changing laws of the optimal scale for object updating in multi-scale decision information systems.
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
(2023)
Article
Computer Science, Artificial Intelligence
Xiao Zhang, Changlin Mei, Jinhai Li, Yanyan Yang, Ting Qian
Summary: Data reduction is a crucial technique for preprocessing data, aiming to select the most representative information to reduce the original data. The development of data reduction techniques for large-scale or huge data has gained significant attention. This article investigates the simultaneous selection of feature and instance in data reduction using fuzzy rough sets, presenting an integrated approach called BSFRS. Experimental results demonstrate the effectiveness of BSFRS in data reduction.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Keyi Guo, Jinhai Li, Xiao Zhang
Summary: The concept-cognitive learning (CCL) process is a specific step in simulating human brain learning. Different CCL models result in different concept learning outcomes. The existing CCL model based on granule approximations lacks accuracy and cannot guarantee consistency. We propose a new CCL method with hybrid lattice structure to improve learning accuracy and develop algorithms for various clue scenarios. Additionally, we introduce a non-logical associative CCL method to handle cases with unreasonable learning outcomes. Experimental results demonstrate the effectiveness of these methods.
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
(2023)
Article
Computer Science, Artificial Intelligence
Zhiming Liu, Jinhai Li, Xiao Zhang, Xi-Zhao Wang
Summary: This article proposes a novel concept-cognitive learning method called SI2CCLM, which addresses the dependency on attribute order issue in existing methods by adopting a stochastic strategy independent of attribute order. A classification algorithm based on SI2CCLM is developed, and the analysis of the algorithm's parameters and convergence is conducted.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
