Article
Computer Science, Artificial Intelligence
Chengmao Wu, Xialu Zhang
Summary: This paper proposes an enhanced self-adaptive weighted possibilistic fuzzy clustering algorithm by introducing the principle of maximum entropy and a robust loss function. The algorithm outperforms existing possibilistic fuzzy clustering-related algorithms and effectively handles noisy data.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Multidisciplinary
Chengmao Wu, Xue Xiao
Summary: In this paper, a total-aware suppressed possibilistic c-means (TSPCM) clustering algorithm is proposed to address the consistency clustering issue in the possibilistic c-means (PCM) algorithm. Experimental results show that the proposed TSPCM and its robust algorithms are significantly superior to existing PCM-related algorithms.
Article
Computer Science, Theory & Methods
Violaine Antoine, Jose A. Guerrero, Gerardo Romero
Summary: Clustering is a data analysis method that groups objects based on similarity. The possibilistic fuzzy c-means (PFCM) is a widely used algorithm that generates a possibilistic partition, allowing for uncertainty and imprecision in noisy environments. Semi-supervised clustering has improved clustering performance by incorporating labeled patterns. In this study, we propose an extension of PFCM that combines labeled patterns with the possibilistic framework, using an adaptive distance measure for added flexibility. Experimental results demonstrate the effectiveness of our new semi-supervised possibilistic fuzzy c-means algorithm on various datasets.
FUZZY SETS AND SYSTEMS
(2022)
Article
Computer Science, Information Systems
Chengmao Wu, Dongxue Yu
Summary: This paper proposes a generalized possibilistic c-means clustering algorithm with double weighting exponents to address the issue of consistency clustering in PCM. The algorithm introduces double weighting exponents into the PCM algorithm and establishes a generalized possibilistic c-means clustering model. Two improved single-exponent possibilistic clustering algorithms, IPCM1 and IPCM2, are proposed and their local convergence is proven. Experimental results show that IPCM1 and IPCM2 outperform existing PCM-related algorithms in terms of clustering performance, robustness to noise, and sensitivity to the weighting exponent. The work of this paper is of great significance for promoting the development of the possibilistic c-means clustering theory.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Josephine Bernadette M. Benjamin, Miin-Shen Yang
Summary: With the rapid growth of social media, virtual communities, and networks, multiview data has become increasingly popular. However, applying single-view clustering methods to multiview data poses challenges. This article proposes novel weighted multiview possibilistic c-means (PCM) algorithms for clustering multiview data, which can identify the importance of each view and select relevant features within each view.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Miin-Shen Yang, Josephine B. M. Benjamin
Summary: This paper investigates the PCM clustering algorithm with Lasso and proposes two approaches, S-PCM1 and S-PCM2. Experimental results and comparisons demonstrate the effectiveness and usefulness of the proposed algorithms.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Zohre Moattar Husseini, Mohammad Hossein Fazel Zarandi, Abbas Ahmadi
Summary: Microarray technology allows for the measurement of expression levels for thousands of genes in different samples simultaneously. This paper proposes adaptive interval type2-possibilistic C-means and adaptive interval type2-possibilistic fuzzy C-means clustering methods to handle the characteristics and uncertainties of microarray data. The proposed algorithms demonstrate better clustering performance compared to well-known soft clustering methods.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Computer Science, Artificial Intelligence
Mohamed A. Mahfouz
Summary: This paper introduces a novel efficient soft clustering algorithm SPCM based on the possibilistic paradigm, which reduces the required exponentiation and division operations by limiting membership values to ordered discrete values. The algorithm assigns appropriate soft membership weights to objects by dividing distances into intervals at each iteration. Experimental results show that SPCM achieves an average of 35% reduction in runtime and 6% increase in performance metrics compared to PCM, with higher reductions for larger datasets. Additionally, SPCM demonstrates comparable performance with lower computational complexity compared to variants of related algorithms.
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Miin-Shen Yang, Josephine B. M. Benjamin
Summary: The study introduces a feature-weighted possibilistic c-means clustering algorithm that improves clustering performance by calculating feature weights to identify important features and eliminating irrelevant features to reduce feature dimension.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2021)
Article
Spectroscopy
Tan Yang, Wu Xiao-hong, Wu Bin, Shen Yan-jun, Liu Jin-mao
Summary: Infrared spectroscopy is used to identify the chemical composition of substances based on molecular vibration and quantum-jump theory. This paper proposes an improved clustering algorithm to process high-dimensional and overlapping spectral data. The algorithm shows higher classification accuracy and less sensitivity to parameters.
SPECTROSCOPY AND SPECTRAL ANALYSIS
(2022)
Article
Computer Science, Artificial Intelligence
Fanyu Bu, Chengsheng Hu, Qingchen Zhang, Changchuan Bai, Laurence T. Yang, Thar Baker
Summary: The article introduces an incremental high-order possibilistic c-means algorithm (IHoPCM) on a cloud-edge computing system for medical data coclustering of multiple hospitals in different locations. Each hospital employs a local edge computing system to learn feature tensors of medical data objects using a deep computation model, which are then uploaded to a cloud computing platform for clustering using the high-order possibilistic c-means algorithm.
IEEE TRANSACTIONS ON FUZZY SYSTEMS
(2021)
Article
Computer Science, Information Systems
Amir Aradnia, Maryam Amir Haeri, Mohammad Mehdi Ebadzadeh
Summary: The K-means algorithm is commonly used for data clustering, but it is limited to linear separable clusters. Kernel K-means extends this method to capture nonlinear structures without storing large kernel matrices, improving clustering efficiency.
INFORMATION SCIENCES
(2022)
Article
Engineering, Electrical & Electronic
Tong Xiao, Yiliang Wan, Jianjun Chen, Wenzhong Shi, Jianxin Qin, Deping Li
Summary: An improved rough-fuzzy possibilistic c-means clustering algorithm combined with multiresolution scales information is proposed to reduce classification uncertainty.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Computer Science, Artificial Intelligence
Hossein Saberi, Reza Sharbati, Behzad Farzanegan
Summary: This paper proposes a new algorithm called IPFCM to cluster noisy data by moving cluster centers towards high-density regions and reducing the impact of noisy data on cluster analysis through a validity index. The results demonstrate the superiority of IPFCM over previous clustering algorithms in separating cluster centers and decreasing the impact of noise data.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Qingchen Zhang, Laurence T. Yang, Zhikui Chen, Peng Li
Summary: The study introduced a high-order PCM algorithm (HOPCM) for big data clustering, along with a distributed method based on MapReduce. Additionally, a privacy-preserving HOPCM algorithm (PPHOPCM) was devised for protecting private data in the cloud.
IEEE TRANSACTIONS ON BIG DATA
(2022)