Article
Computer Science, Artificial Intelligence
Leonardo Ramos Emmendorfer, Anne Magaly de Paula Canuto
Summary: A novel linkage criterion for Hierarchical agglomerative clustering (HAC) is proposed and evaluated in this paper, named GAL. Empirical analysis shows that the results obtained by the proposed criterion surpass all existing reference methods in terms of performance.
APPLIED SOFT COMPUTING
(2021)
Article
Biochemical Research Methods
Mohammad Taheri-Ledari, Amirali Zandieh, Seyed Peyman Shariatpanahi, Changiz Eslahchi
Summary: The proposal of algorithmic approaches for protein domain decomposition remains a lively research area due to the inherent ambiguity of the problem. While previous efforts have focused on developing clustering algorithms, the enhanced measures of proximity between amino acids have been relatively unexplored. Competitive performance of diffusion kernels on protein graphs suggests promising potential for parsing proteins into domains and conducting structural analysis.
BMC BIOINFORMATICS
(2022)
Article
Physics, Multidisciplinary
Eric K. Tokuda, Cesar H. Comin, Luciano da F. Costa
Summary: Hierarchical agglomerative methods are effective and popular for clustering data, but have not been systematically compared regarding false positives when searching for clusters. A cluster model involving a higher density nucleus, transition, and outliers is used to quantify the relevance of obtained clusters and address false positive issues. Experiment results show many methods detecting two clusters in unimodal data, with single-linkage method being more resilient to false positives.
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS
(2022)
Article
Chemistry, Multidisciplinary
Na Xu, Robert B. Finkelman, Shifeng Dai, Chuanpeng Xu, Mengmeng Peng
Summary: Research findings suggest that the average linkage hierarchical clustering algorithm is more accurate and reliable for analyzing modes of element occurrence in coal from the Adaohai coal mine compared to other statistical methods. Analytical results on selenium, beryllium, and thallium indicate discrepancies with geochemical principles, such as substituting for phosphorus, being associated with rare earth elements, and occurring in iron sulfides, respectively.
Article
Engineering, Environmental
Jianhua Yan, Jianping Chen, Jiewei Zhan, Shengyuan Song, Yansong Zhang, Mingyu Zhao, Yongqiang Liu, Wanglai Xu
Summary: Identifying rock discontinuity sets is crucial for analyzing rock slope stability, and this paper introduces a new hierarchical agglomerative clustering method called MAGNES, which performs well in detecting discontinuity sets using the average linkage criterion.
BULLETIN OF ENGINEERING GEOLOGY AND THE ENVIRONMENT
(2022)
Article
Thermodynamics
Nestor Gonzalez-Cabrera, Jose Ortiz-Bejar, Alejandro Zamora-Mendez, Mario R. Arrieta Paternina
Summary: This study introduces an optimal clustering strategy to extract representative demand curves from hourly demand data, aiming to solve the transmission expansion planning problem. The proposed approach provides an efficient reduction of high dimensionality data for the TEP problem through the implementation of HACA algorithm. Results show that the method demonstrates high efficiency and superior functionality when implemented on the IEEE 118-node network, outperforming the K-means method.
Article
Computer Science, Information Systems
Md Monjur Ul Hasan, Reza Shahidi, Dennis K. Peters, Lesley James, Ray Gosine
Summary: Various approaches for data clustering, such as partitioning, hierarchical, and machine learning methods, have been discussed in the literature. However, most of these approaches require prior knowledge about the clusters and may not be robust enough for higher-dimensional data. In this study, a new clustering algorithm called Piecemeal Clustering is proposed, which successfully clusters data without prior knowledge of the number of clusters and works well with both low- and high-dimensional data. Experimental results on two real-world datasets show that the proposed algorithm outperforms seven other state-of-the-art algorithms.
Article
Computer Science, Artificial Intelligence
Xingcheng Ran, Yue Xi, Yonggang Lu, Xiangwen Wang, Zhenyu Lu
Summary: Data clustering is a widely used technique in various fields to divide objects into different clusters based on similarity measures. Hierarchical clustering methods generate consistent partitions of data at different levels, allowing analysis of complex data structures. This article comprehensively reviews various hierarchical clustering methods, including recent developments, and examines the role of similarity measures in the clustering process.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Energy & Fuels
Yifan Zhao, Peng Xue, Cheng Fan, Bojia Li, Nan Zhang, Tao Ma, Jingchao Xie, Jiaping Liu
Summary: This study proposes a novel reference spectrum analysis method to establish local reference spectra based on 147,374 spectra in Beijing, which can characterize solar resources throughout the year. The deep learning autoencoder is adopted to extract intrinsic features of each spectrum for clustering and 10 clusters are divided by hierarchical clustering. The new local reference spectra show shape variations compared to the standard spectrum AM1.5D, with 61% maximum differences among photovoltaic applications.
Article
Mathematics, Interdisciplinary Applications
Nathanael Randriamihamison, Nathalie Vialaneix, Pierre Neuvial
Summary: This article discusses the various extensions of hierarchical agglomerative clustering (HAC) with Ward's linkage, applicability conditions, and different versions of graphical representation as dendrograms. It also highlights the distinction between the consistency property and absence of crossover within the dendrogram. The study shows that the constrained version of HAC can sometimes provide more relevant results despite optimizing the objective criterion on a reduced set of solutions.
JOURNAL OF CLASSIFICATION
(2021)
Article
Computer Science, Information Systems
Nana Liu, Zeshui Xu, Xiao-Jun Zeng, Peijia Ren
Summary: This paper introduces a new method for clustering LOR information using the AHC algorithm, by extending existing distance measure methods and simplifying aggregation methods. A numerical case study is presented to illustrate the algorithm's usage, and discussions are made on the features of the algorithm.
INFORMATION SCIENCES
(2021)
Article
Engineering, Marine
Eun-Ji Kang, Hyeong-Tak Lee, Dae-Gun Kim, Kyoung-Kuk Yoon, Ik-Soon Cho
Summary: This study categorizes ship pilots into cautious, efficient, and hazardous types through quantitative analysis, which contributes to the safety management of pilots.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2022)
Article
Physics, Multidisciplinary
Scott Payne, Edgar Fuller, George Spirou, Cun-Quan Zhang
Summary: The Automatic Quasi-Clique Merger algorithm is a novel hierarchical clustering algorithm that can automatically return different numbers of clusters without relying on parameters. It is suitable for any dataset with a similarity measure and can adaptively unfold the agglomeration process based on the clusters in the dataset.
PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Kalpathy Jayanth Krishnan, Kishalay Mitra
Summary: This study proposes a modified Self Organizing Map algorithm for clustering time series data. By modifying the original steps of the algorithm and using specific initialization methods and similarity measures, this algorithm outperforms other popular clustering algorithms in terms of clustering performance and computation time.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Meng-Chen Li, Likarn Wang
Summary: The method of hierarchical clustering is used to locate intrusion-induced disturbances on sensing fibers of a dual Mach-Zehnder interferometer. The disturbance is located by finding the x coordinate of the centroid of the largest cluster on the Euclidean plane. Average linkage and complete linkage criteria are compared for clustering analysis to determine which provides better accuracy. Using differential signals in the clustering analysis reduces locating error.
IEEE SENSORS JOURNAL
(2023)