Centroid-based Clustering Algorithms Research Articles

Regional differences of chronic rhinosinusitis endotypes based on tissue inflammatory and remodeling biomarkers

Objective: To analyze the characteristics of patients with chronic rhinosinusitis (CRS) in the South China region based on pathological tissue biomarkers for regional comparison. Methods: The study population consisted of CRS in-patients in the First Affiliated Hospital of Sun Yat-sen University from October 2019 to June 2022. Among all the 181 cases, 123 of them were male and 58 were female, with an average age of 40. Retrospectively collected clinical data included demographic information, preoperative symptom scores, preoperative endoscopic images, preoperative paranasal sinus computed tomography scanning images, and inflammatory serological features. In addition, 52 variables of pathological tissue biomarkers including cytokines, chemokines and remodeling factors were collected for analysis. Cluster analysis was performed on the integrated data of training set through centroid-based clustering algorithm, and the inflammatory characteristics, post-operation control status, and airway diseases comorbidity of each endotype were analyzed. R project (version 4.2.2) was used in statistical analysis. Results: Cluster analysis divided 181 patients with CRS into 4 endotypes. Cluster 1 (n=101, 55.80%) showed a locally low inflammatory status. Cluster 2 (n=23, 12.71%) showed a mixed type of inflammation with predominantly neutrophilic inflammation and tissue remodeling. Cluster 3 (n=11, 6.08%) was characterized by type Ⅱ inflammation without tissue remodeling. Cluster 4 (n=46, 25.41%) was mainly characterized by type Ⅱ inflammation with tissue remodeling, showing higher comorbidity rate of asthma and allergic rhinitis. This cluster presented more severe symptoms, significant olfactory dysfunction, extensive overall inflammation based on objective examination results, a notable increase in total eosinophil count and proportion in peripheral blood, and the highest uncontrolled rate observed one year post-surgery. In comparison to other regions, the endotype classification of CRS in Southern China was characterized by a predominant pattern of locally low inflammatory status, a moderate level of type Ⅱ inflammation with tissue remodeling, and a lesser presence of neutrophilic inflammation. Conclusion: CRS distribution in Southern China is mainly characterized by low inflammatory endotype and type Ⅱ inflammation with tissue remodeling. The latter shows more severe clinical manifestations and higher uncontrol rate after surgery.

Morphological Accuracy Data Clustering: A Novel Algorithm for Enhanced Cluster Analysis

In today’s data-driven world, we are constantly exposed to a vast amount of information. This information is stored in various information systems and is used for analysis and management purposes. One important approach to handle these data is through the process of clustering or categorization. Clustering algorithms are powerful tools used in data analysis and machine learning to group similar data points together based on their inherent characteristics. These algorithms aim to identify patterns and structures within a dataset, allowing for the discovery of hidden relationships and insights. By partitioning data into distinct clusters, clustering algorithms enable efficient data exploration, classification, and anomaly detection. In this study, we propose a novel centroid-based clustering algorithm, namely, the morphological accuracy clustering algorithm (MAC algorithm). The proposed algorithm uses a morphological accuracy measure to define the centroid of the cluster. The empirical results demonstrate that the proposed algorithm achieves a stable clustering outcome in fewer iterations compared to several existing centroid-based clustering algorithms. Additionally, the clusters generated by these existing algorithms are highly susceptible to the initial centroid selection made by the user.

Fusion of Centroid-Based Clustering With Graph Clustering: An Expectation-Maximization-Based Hybrid Clustering.

This article extends the expectation-maximization (EM) formulation for the Gaussian mixture model (GMM) with a novel weighted dissimilarity loss. This extension results in the fusion of two different clustering methods, namely, centroid-based clustering and graph clustering in the same framework in order to leverage their advantages. The fusion of centroid-based clustering and graph clustering results in a simple "soft" asynchronous hybrid clustering method. The proposed algorithm may start as a pure centroid-based clustering algorithm (e.g., k -means), and as the time evolves, it may eventually and gradually turn into a pure graph clustering algorithm [e.g., basic greedy asynchronous distributed interference avoidance (GADIA) (Babadi and Tarokh, 2010)] as the algorithm converges and vice versa. The "hard" version of the proposed hybrid algorithm includes the standard Hopfield neural networks (and, thus, Bruck's Ln algorithm by (Bruck, 1990) and the Ising model in statistical mechanics), Babadi and Tarokh's basic GADIA in 2010, and the standard k -means (Steinhaus, 1956), (MacQueen, 1967) [i.e., the Lloyd algorithm (Lloyd, 1957, 1982)] as its special cases. We call the "hard version" of the proposed clustering as "hybrid-nongreedy asynchronous clustering (H-NAC)." We apply the H-NAC to various clustering problems using well-known benchmark datasets. The computer simulations confirm the superior performance of the H-NAC compared to the k -means clustering, k -GADIA, spectral clustering, and a very recent clustering algorithm structured graph learning (SGL) by Kang et al. (2021), which represents one of the state-of-the-art clustering algorithms.

Centroid Based Clustering Approach for Extractive Text Summarization

Abstract: Extractive text summarization is the process of identifying the most important information from a large text and presenting it in a condensed form. One popular approach to this problem is the use of centroid-based clustering algorithms, which group together similar sentences based on their content and then select representative sentences from each cluster to form a summary. In this research, we present a centroid-based clustering algorithm for email summarization that combines the use of word embeddings with a clustering algorithm. We compare our algorithm to existing summarization techniques. Our results show that our approach stands close to existing methods in terms of summary quality, while also being computationally efficient. Overall, our work demonstrates the potential of centroid-based clustering algorithms for extractive text summarization and suggests avenues for further research in this area.

Centroid Based Celestial Clustering Algorithm: A Novel Unsupervised Learning Method for Haemogram Data Clustering

Accuracy of clustering is the most important parameter as far as automated disease identification is concerned. There have always been attempts to automate the process of disease prediction from haemogram data. However, there are several components in blood test results and very often we find that a variety of combinations of these component results are to be used to detect a disease. This makes identification of diseases really hard and necessitates the use of data analysis techniques. As new diseases are arising from time to time, a useful method for prediction is unsupervised learning and the corresponding data analysis technique is clustering. An easy, efficient and centroid based clustering algorithm that has been in practice widely is k-means. Its simplicity and efficiency make it a natural choice for most of the clustering applications. However, k-means is largely dependent on the selection of initial cluster centers and a bad choice can make it fall to local optima, thereby sacrificing accuracy. Besides, it is non-deterministic in nature. This paper proposes a novel, nature inspired, clustering method, named Centroid Based Celestial Clustering, which overcomes the above issues. Our method is deterministic and converges to global optima on spherical datasets. We experimentally evaluate our algorithm for speed of execution and cluster quality against well-known clustering algorithms using statistical evaluation metrics like silhouette width, adjusted rand index and Dunn index. We use the method to predict diseases identifiable from blood tests and our experiments show that the accuracy of prediction is very promising.

An Efficient Method to Accurately Cluster Large Number of High Dimensional Facial Images

Accurately clustering large, high dimensional datasets is a challenging problem in unsupervised learning. K-means is considered to be a fast, widely used and accurate centroid based data partitioning algorithm for spherical datasets. However, its non-determinism and heavy dependence on the selection of initial cluster centers along with vulnerability to noise make it a poor candidate for clustering large datasets with high dimensionality. To overcome these, we develop a novel, nature inspired, centroid based clustering algorithm, inspired from the principles of particle physics. Our method ensures that the convergence to local optima and non-deterministic outputs are avoided. We experiment the method on large datasets of human face images. Besides, our method addresses the problem of outliers and presence of not well-separated data in these datasets. We use a deep learning model for extracting facial features into a vector of 128 dimensions. We validate the quality and accuracy of our methods using different statistical parameters like <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">f-measure, accuracy, error rate, average in group proportion and normalized cluster size rand index</i> . These evaluations show that our method exhibits better accuracy and quality in clustering large face image datasets, in comparison with other existing mechanisms. The strength of our algorithms is more visible as the size of the dataset grows.

Maxmin Data Range Heuristic-Based Initial Centroid Method of Partitional Clustering for Big Data Mining

The centroid-based clustering algorithm depends on the number of clusters, initial centroid, distance measures, and statistical approach of central tendencies. The initial centroid initialization algorithm defines convergence speed, computing efficiency, execution time, scalability, memory utilization, and performance issues for big data clustering. Nowadays various researchers have proposed the cluster initialization techniques, where some initialization techniques reduce the number of iterations with the lowest cluster quality, and some initialization techniques increase the cluster quality with high iterations. For these reasons, this study proposed the initial centroid initialization based Maxmin Data Range Heuristic (MDRH) method for K-Means (KM) clustering that reduces the execution times, iterations, and improves quality for big data clustering. The proposed MDRH method has compared against the classical KM and KM++ algorithms with four real datasets. The MDRH method has achieved better effectiveness and efficiency over RS, DB, CH, SC, IS, and CT quantitative measurements.

Forming coordination group for coordinated traffic congestion management schemes

It has been well known that properly coordinating travelers’ route choices will help mitigate urban traffic congestion. Accordingly, coordinated routing mechanisms (CRMs), functioned as coordinated network congestion management schemes, have been developed in recent years. They often employ game theory models to strategically coordinate the route decisions of a group of travelers en route with the aim to reduce network traffic congestion. However, existing studies do not provide the solution about how to form the traveler groups, such as who and how many travelers the CRMs should involve. This issue may jeopardize the efficiency and scalability of the CRMs in practical applications. To make up this gap, this study seeks to develop the clustering aided network modeling approaches to form the coordination groups (CGs), which have the proper members and sizes for supporting the merits of the congestion management schemes. Specifically, we define coordination potential as an index to indicate the potential benefit for coordinating travelers’ route decisions on network congestion reduction. Then we develop rigorous approaches to understand the mathematical features and further quantify the coordination potential between two travelers, and among multiple travelers. Built upon that, we form the CGs through a well-designed adaptive centroid-based clustering algorithm (ACCA). It secures a local optimal clustering solution, balancing the intra-cluster and inter-cluster CP, so that we can ensure a small system performance loss as we implement a CRM on each CG. The efficiency of the CG formation approach is evaluated by the CG-CRM scheme, which partitions the travelers into multiple CGs and independently coordinates the travelers’ route choices in each CG by a CRM. The numerical experiments built upon both Sioux Falls and Hardee city networks confirm the efficiency and effectiveness of the CG formation approach. Mainly, the CG-CRM outperforms the independent best response routing mechanism in system performance. As compared to the CRM working on a single group involving all travelers, the CG-CRM requires significantly less computation load with a minor compromise on the system performance. This merit becomes more apparent under high penetration and congested traffic condition. Therefore, we claim that the CG formation approach will significantly help the implementation of the coordinated congestion mitigation schemes in practice.

Cs-means: Determining optimal number of clusters based on a level-of-similarity

This paper proposes a centroid-based clustering algorithm which is capable of clustering data-points with n-features, without having to specify the number of clusters to be formed. The core logic behind the algorithm is a similarity measure, which collectively decides whether to assign an incoming data-point to a pre-existing cluster, or create a new cluster and assign the data-point to it. The proposed clustering algorithm is application-specific and is applicable when the need is to perform clustering analysis of a stream of data-points, where the similarity measure between an incoming data-point and the cluster to which the data-point is to be associated with, is greater than the predefined Level-of-Similarity.

Clustering by Hybrid K-Harmonic Means and Biogeography Based Optimization Algorithm for Medical Data

In today trendy world hybrid based optimized data clustering is unique and imperative clustering tool in the area of data mining, which is dynamic research of actual creation problems. The oldest and furthermost commonly used popular clustering technique is the K-means(KM) algorithm, which is very complex and for the initialization of the cluster centroid and it will easily go for premature converge. This initialization problem of K-means can be evaded by built in boost function of K-Harmonic Means, which is centroid based clustering algorithm and also unresponsive for collection of initial partition clustering , but it can easily go for pre-matured conjunction in local optima. To avoid this convergence problem, this proposed algorithm uses Boosting K-harmonic means(KHM) algorithm with BBO to produce more precise, robust, better clustering solution in few number of iterations, evade conning in local optima and simply convergence to relate with Harmonic Means, BBO algorithms. Biogeography based algorithm works with the concept of emigration and immigration of inhabitants from one location to another location, Which has high computation cost. For avoiding this high computation cost in this hybrid optimization technique Biogeography-Based Optimization (BBO) is integrated with K-Harmonic means algorithm to produce optimum and effective clustering solution with faster convergence. BBO is universal optimization methods to solve utmost of the optimization problem, which is an production based generation of evolutionary algorithm (EA)that augments a function by stochastically and repeatedly improving the clustering solution of quality, or fitness function. The experimental results of this paper shown as the projected method is very resourceful and faster to afford better clustering solution in less number of repetitions for medical data.

Improving a Centroid-Based Clustering by Using Suitable Centroids from Another Clustering

Fast centroid-based clustering algorithms such as k-means usually converge to a local optimum. In this work, we propose a method for constructing a better clustering from two such suboptimal clustering solutions based on the fact that each suboptimal clustering has benefits regarding to including some of the correct clusters. We develop the new method COTCLUS to find two centroids from one clustering and replace them by two centroids from the other clustering so that the maximum decrease in the mean square error of the first clustering is achieved. After modifying centroids, k-means algorithm with few iterations is applied for fine-tuning. In an iterative algorithm, the resulting clustering is further improved using a new given clustering. The proposed method can find optimal clustering in a very small number of iterations for datasets with well-separated clusters. We demonstrate by experiments that the proposed method outperforms the selected competitive methods.

A cluster-based optimization approach to support the participation of an aggregator of a larger number of prosumers in the day-ahead energy market

Optimizing the participation of a large number of prosumers in the electricity markets is a challenging problem, especially for portfolios with thousands or millions of flexible resources. To address this problem, this paper proposes a cluster-based optimization approach to support an aggregator in the definition of demand and supply bids for the day-ahead energy market. This approach consists of two steps. In the first step, the aggregated flexibility of the entire portfolio is computed by a centroid-based clustering algorithm. In the second step, the supply and demand bids are defined by an optimization model that can assume the form of a deterministic or a two-stage stochastic problem. A case study of 10,000 prosumers from the Iberian market is used to evaluate and compare the performance of the bidding optimization models with and without pre-clustering. The numerical results show that the optimized bidding strategies outperform an inflexible strategy by more than 20% of cost savings. The centroid-based clustering algorithm reduces effectively the execution times of the bidding optimization problems, without affecting the quality of the energy bids.

Applying Machine Learning Algorithms to Segment High-Cost Patient Populations.

Efforts to improve the value of care for high-cost patients may benefit from care management strategies targeted at clinically distinct subgroups of patients. To evaluate the performance of three different machine learning algorithms for identifying subgroups of high-cost patients. We applied three different clustering algorithms-connectivity-based clustering using agglomerative hierarchical clustering, centroid-based clustering with the k-medoids algorithm, and density-based clustering with the OPTICS algorithm-to a clinical and administrative dataset. We then examined the extent to which each algorithm identified subgroups of patients that were (1) clinically distinct and (2) associated with meaningful differences in relevant utilization metrics. Patients enrolled in a national Medicare Advantage plan, categorized in the top decile of spending (n = 6154). Post hoc discriminative models comparing the importance of variables for distinguishing observations in one cluster from the rest. Variance in utilization and spending measures. Connectivity-based, centroid-based, and density-based clustering identified eight, five, and ten subgroups of high-cost patients, respectively. Post hoc discriminative models indicated that density-based clustering subgroups were the most clinically distinct. The variance of utilization and spending measures was the greatest among the subgroups identified through density-based clustering. Machine learning algorithms can be used to segment a high-cost patient population into subgroups of patients that are clinically distinct and associated with meaningful differences in utilization and spending measures. For these purposes, density-based clustering with the OPTICS algorithm outperformed connectivity-based and centroid-based clustering algorithms.

Clustering big IoT data by metaheuristic optimized mini-batch and parallel partition-based DGC in Hadoop

Clustering algorithms are an important branch of data mining family which has been applied widely in IoT applications such as finding similar sensing patterns, detecting outliers, and segmenting large behavioral groups in real-time. Traditional full batchk-means for clustering IoT big data is confronted by large scaled storage and high computational complexity problems. In order to overcome the latency inherited from full batchk-means, two big data processing methods were often used: the first method is to use small batches as the input data to multiple computers for reducing the computation efforts. However, depending on the sensed data which may be heterogeneously fused from different sources in an IoT network, the size of each mini batch may vary in each iteration of clustering process. When these input data are subject to clustering their centers would shift drastically, which affects the final clustering results. The second method is parallel computing, it decreases the runtime while the overall computational effort remains the same. Furthermore, some centroid based clustering algorithm such ask-means converges easily into local optima. In light of this, in this paper, a new partitioned clustering method that is optimized by metaheuristic is proposed for IoT big data environment. The method has three main activities: Firstly, a sample of the dataset is partitioned into mini batches. It is followed by adjusting the centroids of the mini batches of data. The third step is collating the mini batches to form clusters, so the quality of the clusters would be maximized. How the positions of the centroids could be optimally attuned at the mini batches are governed by a metaheuristic called Dynamic Group Optimization. The data are processed in parallel in Hadoop. Extensive experiments are conducted to investigate the performance. The results show that our proposed method is a promising tool for clustering fused IoT data efficiently.

Learning Automata Clustering

Clustering of data points has been a profound research avenue in the history of machine learning algorithms. Using learning automata which are autonomous decision making entities, in this paper, the learning automata clustering algorithm is proposed. In learning automata clustering, each data point is affiliated with a learning automaton where the learning automaton determines the cluster membership of that data point. The cluster rectification is done through a reinforcement signal for each learning automaton which is fabricated from the Euclidean distance of that data point and the mean value of its designated cluster. Finally, the learning automata clustering is compared with four centroid-based clustering algorithms, K-means, K-means++, K-medians, and K-medoids and results demonstrate the high clustering accuracy and comparable Silhouette coefficient of the proposed method.

A fast and accurate algorithm for unsupervised clustering around centroids

A centroid-based clustering algorithm is proposed that works in a totally unsupervised fashion and is significantly faster and more accurate than existing algorithms. The algorithm, named CLUBS+ (for CLustering Using Binary Splitting), achieves these results by combining features of hierarchical and partition-based algorithms. Thus, CLUBS+ consists of two major phases, i.e., a divisive phase and an agglomerative phase, each followed by a refinement phase. Each major phase consists of successive steps in which the samples are repartitioned using a criterion based on least quadratic distance. This criterion possesses unique analytical properties that are elucidated in the paper and exploited by the algorithm to achieve a very fast computation. The paper presents the results of the extensive experiments performed: these confirm that the new algorithm is fast, impervious to noise, and produces results of better quality than other algorithms, such as BOOL, BIRCH, and k-means++, even when the analyst can determine the correct number of clusters—a very difficult task from which users are spared by CLUBS+.

Clustering Heterogeneous Data with k-Means by Mutual Information-Based Unsupervised Feature Transformation

Traditional centroid-based clustering algorithms for heterogeneous data with numerical and non-numerical features result in different levels of inaccurate clustering. This is because the Hamming distance used for dissimilarity measurement of non-numerical values does not provide optimal distances between different values, and problems arise from attempts to combine the Euclidean distance and Hamming distance. In this study, the mutual information (MI)-based unsupervised feature transformation (UFT), which can transform non-numerical features into numerical features without information loss, was utilized with the conventional k-means algorithm for heterogeneous data clustering. For the original non-numerical features, UFT can provide numerical values which preserve the structure of the original non-numerical features and have the property of continuous values at the same time. Experiments and analysis of real-world datasets showed that, the integrated UFT-k-means clustering algorithm outperformed others for heterogeneous data with both numerical and non-numerical features.

Advanced Mobile Security System Operated by Bioelectrical Sensor

This article describes a novel type of an advanced mobile security system, a myokinetic password, using surface electromyographic (EMG) signals. We proposed an advanced security system that can increase password strength without modifying password length or increasing alphabet character variation. We present an implementation of an intuitive password system that combines muscle activation to enhance password complexity. The muscle activation levels from EMG signals are used to split each individual code composition into multiple states, thereby increasing possible password combinations. We proposed a centroid-based clustering algorithm to cluster a feature into groups according to the muscle activation. We also verified the possibility of expanding the technique to support up to three muscle activation levels in the experiment. By utilizing the proposed method, the number of possible input key combination is increased from 256 to 810,000 with three muscle levels.

Noise-enhanced clustering and competitive learning algorithms

Noise can provably speed up convergence in many centroid-based clustering algorithms. This includes the popular k-means clustering algorithm. The clustering noise benefit follows from the general noise benefit for the expectation–maximization algorithm because many clustering algorithms are special cases of the expectation–maximization algorithm. Simulations show that noise also speeds up convergence in stochastic unsupervised competitive learning, supervised competitive learning, and differential competitive learning.

The Heterogeneous P-Median Problem for Categorization Based Clustering

The p-median offers an alternative to centroid-based clustering algorithms for identifying unobserved categories. However, existing p-median formulations typically require data aggregation into a single proximity matrix, resulting in masked respondent heterogeneity. A proposed three-way formulation of the p-median problem explicitly considers heterogeneity by identifying groups of individual respondents that perceive similar category structures. Three proposed heuristics for the heterogeneous p-median (HPM) are developed and then illustrated in a consumer psychology context using a sample of undergraduate students who performed a sorting task of major U.S. retailers, as well as a through Monte Carlo analysis.