• Journal of Korea Multimedia Society
• /
• v.23 no.1
• /
• pp.1-7
• /
• 2020

Clustering methods essentially take a two-step approach; extracting feature vectors for dimensionality reduction and then employing clustering algorithm on the extracted feature vectors. However, for clustering images, the traditional clustering methods such as stacked auto-encoder based k-means are not effective since they tend to ignore the local information. In this paper, we propose a method first to effectively reduce data dimensionality using convolutional auto-encoder to capture and reflect the local information and then to accurately cluster similar data samples by using a hierarchical clustering approach. The experimental results confirm that the clustering results are improved by using the proposed model in terms of clustering accuracy and normalized mutual information.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.2
• /
• pp.92-98
• /
• 2017

Data clustering determines a group of patterns using similarity measure in a dataset and is one of the most important and difficult technique in data mining. Clustering can be formally considered as a particular kind of NP-hard grouping problem. K-means algorithm which is popular and efficient, is sensitive for initialization and has the possibility to be stuck in local optimum because of hill climbing clustering method. This method is also not computationally feasible in practice, especially for large datasets and large number of clusters. Therefore, we need a robust and efficient clustering algorithm to find the global optimum (not local optimum) especially when much data is collected from many IoT (Internet of Things) devices in these days. The objective of this paper is to propose new Hybrid Simulated Annealing (HSA) which is combined simulated annealing with K-means for non-hierarchical clustering of big data. Simulated annealing (SA) is useful for diversified search in large search space and K-means is useful for converged search in predetermined search space. Our proposed method can balance the intensification and diversification to find the global optimal solution in big data clustering. The performance of HSA is validated using Iris, Wine, Glass, and Vowel UCI machine learning repository datasets comparing to previous studies by experiment and analysis. Our proposed KSAK (K-means+SA+K-means) and SAK (SA+K-means) are better than KSA(K-means+SA), SA, and K-means in our simulations. Our method has significantly improved accuracy and efficiency to find the global optimal data clustering solution for complex, real time, and costly data mining process.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.40 no.4
• /
• pp.203-210
• /
• 2017

Data clustering is one of the most difficult and challenging problems and can be formally considered as a particular kind of NP-hard grouping problems. The K-means algorithm is one of the most popular and widely used clustering method because it is easy to implement and very efficient. However, it has high possibility to trap in local optimum and high variation of solutions with different initials for the large data set. Therefore, we need study efficient computational intelligence method to find the global optimal solution in data clustering problem within limited computational time. The objective of this paper is to propose a combined artificial bee colony (CABC) with K-means for initialization and finalization to find optimal solution that is effective on data clustering optimization problem. The artificial bee colony (ABC) is an algorithm motivated by the intelligent behavior exhibited by honeybees when searching for food. The performance of ABC is better than or similar to other population-based algorithms with the added advantage of employing fewer control parameters. Our proposed CABC method is able to provide near optimal solution within reasonable time to balance the converged and diversified searches. In this paper, the experiment and analysis of clustering problems demonstrate that CABC is a competitive approach comparing to previous partitioning approaches in satisfactory results with respect to solution quality. We validate the performance of CABC using Iris, Wine, Glass, Vowel, and Cloud UCI machine learning repository datasets comparing to previous studies by experiment and analysis. Our proposed KABCK (K-means+ABC+K-means) is better than ABCK (ABC+K-means), KABC (K-means+ABC), ABC, and K-means in our simulations.

• Communications for Statistical Applications and Methods
• /
• v.12 no.2
• /
• pp.497-508
• /
• 2005

This paper aims at studying on K-means Clustering focusing on initialization which affect the clustering results in K-means cluster analysis. The four different methods(the MA method, the KA method, the Max-Min method and the Space Partition method) were compared and the clustering result shows that there were some differences among these methods, especially that the MA method sometimes leads to incorrect clustering due to the inappropriate initialization depending on the types of data and the Max-Min method is shown to be more effective than other methods especially when the data size is large.

• Journal of the Korean Data and Information Science Society
• /
• v.21 no.1
• /
• pp.121-128
• /
• 2010

K-means clustering is an iterative algorithm in which items are moved among sets of clusters until the desired set is reached. K-means clustering has been widely used in many applications, such as market research, pattern analysis or recognition, image processing, etc. It can identify dense and sparse regions among data attributes or object attributes. But k-means algorithm requires many hours to get k clusters that we want, because it is more primitive, explorative. In this paper we propose a new method of k-means clustering using a center of gravity for grid-based sample. It is more fast than any traditional clustering method and maintains its accuracy.

• The Korean Journal of Applied Statistics
• /
• v.17 no.1
• /
• pp.135-144
• /
• 2004

We propose a reproducibility (validity) assessment procedure of K-means cluster analysis by randomly partitioning the data set into three parts, of which two subsets are used for developing clustering rules and one subset for testing consistency of clustering rules. Also, as an alternative to Rand index and corrected Rand index, we propose an entropy-based consistency measure between two clustering rules, and apply it to determination of the number of clusters in K-means clustering.

• Educational Technology International
• /
• v.18 no.2
• /
• pp.159-191
• /
• 2017

Academic analytics guides university decision-makers to assign limited resources more effectively. Especially, diverse academic courses clustered by the usage patterns and levels on Learning Management System(LMS) help understanding instructors' pedagogical approach and the integration level of technologies. Further, the clustering results can contribute deciding proper range and levels of financial and technical supports. However, in spite of diverse analytic methodologies, clustering analysis methods often provide different results. The purpose of this study is to present implications by using three different clustering analysis including Gaussian Mixture Model, K-Means clustering, and Hierarchical clustering. As a case, we have clustered academic courses based on the usage levels and patterns of LMS in higher education using those three clustering techniques. In this study, 2,639 courses opened during 2013 fall semester in a large private university located in South Korea were analyzed with 13 observation variables that represent the characteristics of academic courses. The results of analysis show that the strengths and weakness of each clustering analysis and suggest that academic leaders and university staff should look into the usage levels and patterns of LMS with more elaborated view and take an integrated approach with different analytic methods for their strategic decision on development of LMS.

• Communications for Statistical Applications and Methods
• /
• v.9 no.2
• /
• pp.505-514
• /
• 2002

Multiple correspondence analysis graphically shows the correspondent relationship among categories in multi-way contingency tables. It is well known that the proportions of the principal inertias as part of the total inertia is low in multiple correspondence analysis. Moreover, although this problem can be overcome by using the Benzecri formula, it is not enough to show clear correspondent relationship among categories (Greenacre and Blasius, 1994, Chapter 10). In addition, they show that Andrews' plot is useful in providing the correspondent relationship among categories. However, this method also does not give some concise interpretation among categories when the number of categories is large. Therefore, in this study, we will easily interpret the multiple correspondence analysis by applying the K-means clustering analysis.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.4
• /
• pp.703-708
• /
• 2022

K-means clustering is a representative clustering technique. However, there is a limitation in not being able to integrate the performance evaluation scale and the method of determining the minimum number of clusters. In this paper, a method for numerically determining the minimum number of clusters is introduced. The explained variance is presented as an integrated measure. We propose that the k-means clustering method should be performed in the subspace of the PCA in order to simultaneously satisfy the minimum number of clusters and the threshold of the explained variance. It aims to present an explanation in principle why principal component analysis and k-means clustering are sequentially performed in pattern recognition and machine learning.

• Journal of information and communication convergence engineering
• /
• v.16 no.4
• /
• pp.258-263
• /
• 2018

Color Doppler sonography is a useful tool for examining blood flow and related indices. However, it should be done by well-trained operator, that is, operator subjectivity exists. In this paper, we propose an automatic blood flow area extraction method from brachial artery that would be an essential building block of computer aided color Doppler analyzer. Specifically, our concern is to examine hypertension suspicious (prehypertension) patients who might develop their symptoms to established hypertension in the future. The proposed method uses fuzzy C-means clustering as quantization engine with careful seeding of the number of clusters from histogram analysis. The experiment verifies that the proposed method is feasible in that the successful extraction rates are 96% (successful in 48 out of 50 test cases) and demonstrated better performance than K-means based method in specificity and sensitivity analysis but the proposed method should be further refined as the retrospective analysis pointed out.