• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.10a
• /
• pp.248-249
• /
• 2015

Clustering is one of the most important unsupervised learning methods that clusters data into homogeneous groups. However, cluster centers tend leaning to high density clusters because clustering is based on the distances between data points and cluster centers. In this paper, a modified clustering method forcing cluster centers to be apart by introducing a center-scattering term in the Fuzzy C-Means objective function is introduced. The proposed method converges more to real centers with small number of iterations compared to the original one. All the strengths can be verified with experimental results.

• Journal of the Ergonomics Society of Korea
• /
• v.19 no.3
• /
• pp.23-37
• /
• 2000

Usability is one of the most important factors that affect customers' decision to purchase a product. Several studies have been conducted to model the relationship between the product design variables and the product usability. Since there could be hundreds of design variables to be considered in the model, a variable screening method is required. Traditional variable screening methods are based on expert opinions (Expert screening) in most Kansei engineering studies. Suggested in this study are statistical methods for screening important design variables by using the principal component regression(PCR), cluster analysis, and partial least squares(PLS) method. Product variables with high effect (PCR screening and PLS screening) or representative variables (Cluster screening) can be used to model the usability. Proposed variable screening methods are used to model the usability for 36 audio/visual products. The three analysis methods (PCR, Cluster, and PLS) show better model performance than the Expert screening in terms of $R^2$, the number of variables in the model, and PRESS. It is expected that these methods can be used for screening the product design variables efficiently.

• Journal of Korean Society for Atmospheric Environment
• /
• v.32 no.5
• /
• pp.513-525
• /
• 2016

To propose an effective ensemble methods in predicting $PM_{10}$ concentration, six experiments were designed by different ensemble average methods (e.g., non-weighted, single weighted, and cluster weighted methods). The single weighted method was calculated the weighted value using both multiple regression analysis and singular value decomposition and the cluster weighted method was estimated the weighted value based on temperature, relative humidity, and wind component using multiple regression analysis. The effects of ensemble average methods were significantly better in weighted average than non-weight. The results of ensemble experiments using weighted average methods were distinguished according to methods calculating the weighted value. The single weighted average method using multiple regression analysis showed the highest accuracy for hourly $PM_{10}$ concentration, and the cluster weighted average method based on relative humidity showed the highest accuracy for daily mean $PM_{10}$ concentration. However, the result of ensemble spread analysis showed better reliability in the single weighted average method than the cluster weighted average method based on relative humidity. Thus, the single weighted average method was the most effective method in this study case.

• Journal of Environmental Science International
• /
• v.25 no.6
• /
• pp.749-756
• /
• 2016

This study shows how the growth of the top part of plants cultivated using the single-row strawberry method, with 12 cm plant spacing, as well as that of plants cultivated through conventional planting, is characterized by the presence of many leaves in the first flower cluster harvest. The leaf area and crown diameter were the largest in the 12 cm spacing method. The hight top fresh weight (59.2 g) was detected wen the 12 cm spacing method was used followed by conventional planting and, 9 cm and 6 cm spacing method. The K and Ca contents in the first flower cluster were the highest when the 12 cm spacing method (2.0% and 2.1%, respectively) and conventional planting, (0.42% and 0.86%, respectively) were used, and these values were significantly higher than the K and Ca contents obtained using the other two methods. The N, P, Mg, Fe, and B contents show no significant differences across the planting methods. The sugar content of the first flower cluster fruits was the highest when the 12 cm spacing method was used, while the sugar content of the fourth flower cluster fruits was highest after conventional planting. Firmness was the highest in the first, third, and fourth flower clusters after conventional planting, while no significant differences were observed for the 6 cm, 9 cm, and 12 cm spacing methods. A yield of 25 g or above during November to December was observed to be the highest when the 12 cm spacing method was used, while a yield of 10-16 g was the highest when both the 9 cm and 12 cm spacing methods werw used. The yield of products in January-April was the highest when the 12 cm spacing and conventional planting methods were used, and total product yield was also the highest for these methods. A significant portion of non-marketable products (39 g) was obtained when the conventional planting method was used.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.939-940
• /
• 2008

In this paper, we propose a novel image clustering method based on weighted sub-sampling to reduce clustering time and the number of clusters for target detection and tracking. Our proposed method first obtain sub-sampling image with specific weights which is the number of target pixels in sampling region. After performing clustering procedure, the cluster center position is properly obtained using weights of target pixels in the cluster. Therefore, our proposed method can not only reduce clustering time, but also obtain proper cluster center.

• Korean Management Science Review
• /
• v.24 no.2
• /
• pp.57-72
• /
• 2007

Many Applications of scientific research have coupled with functional data signal clustering techniques to discover novel characteristics that can be used for the diagnoses of several issues. In this article we present an interpretable multi-scale cluster hierarchy framework for clustering functional data using its multi-aspect frequency information. The suggested method focuses on how to effectively select transformed features/variables in unsupervised manner so that finally reduce the data dimension and achieve the multi-purposed clustering. Specially, we apply our suggested method to mutual fund returns and make superior-performing funds group based on different aspects such as global patterns, seasonal variations, levels of noise, and their combinations. To promise our method producing a quality cluster hierarchy, we give some empirical results under the simulation study and a set of real life data. This research will contribute to financial market analysis and flexibly fit to other research fields with clustering purposes.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.12 no.6
• /
• pp.135-142
• /
• 2004

The severe frontal crash tests are NCAP with belted occupant at 35mph and FMVSS 208 with unbelted occupant at 25mph, This paper describes the design process of occupant protection systems, airbag and seat belt, under the two tests. In this study, NCAP simulations are performed by Monte Carlo search method and cluster analysis. The Monte Carlo search method is a global optimization technique and requires execution of a series of deterministic analyses, The procedure is as follows. 1) Define the region of interest 2) Perform Monte Carlo simulation with uniform distribution 3) Transform output to obtain points grouped around the local minima 4) Perform cluster analysis to obtain groups that are close to each other 5) Define the several feasible design ranges. The several feasible designs are acquired and checked under FMVSS 208 simulation with unbelted occupant at 25mph.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.23 no.6
• /
• pp.415-423
• /
• 2018

This study uses a level selection algorithm with fixed sampling frequency for modular multilevel converter (MMC) systems. Theoretically, the proposed method increases the level infinitely while the sampling time remains the same. The proposed method called cluster stream buffer (CSB) consists of several clusters, wherein each cluster is composed of 32 submodules that depend on the level of the submodules in the MMC system. To increase the level of the MMC system, additional clusters are used, and the sampling time between clusters is determined from the sampling time between levels needed for utilizing the entire level from the MMC system. This method is crucial in the control of MMC-type HVDC systems because it improves scalability and precision.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.17 no.2
• /
• pp.25-33
• /
• 1992

Manufacturing cell formation is one of the most important problems faced in designing cellular manufacturing systems. The purpose of this study is to design effective manufacturing cell systems by developing a method which forms machines/parts into optimal machine cells/part families. The 0-1 data matrix structure is used to form a basis for manufacturing cell formation. In this paper, we propose a CE method to reorder the 0-1 data matrix for manufacturing cell formation. The resulting solutions are shown to demonstrate the effectiveness of the CE method.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39B no.11
• /
• pp.763-770
• /
• 2014

In traditional routing protocols including LEACH for wireless sensor networks, nodes suffer from unbalanced energy consumption because the nodes require large transmission energy as the distance to the sink node increase. Multi-hop based routing protocols have been studied to address this problem. In existing protocols, each cluster head usually chooses the closest head as a relay node. We propose LEACH-CHT, in which cluster heads choose the path with least energy consumption to send data to the sink node. In our research, each hop, a cluster head selects the least cost path to the sink node. This method solves the looping problem efficiently as well as make it possible that a cluster head excludes other cluster heads placed farther than its location from the path, without additional energy consumption. By balancing the energy consumption among the nodes, our proposed scheme outperforms existing multi-hop schemes by up to 36% in terms of average network lifetime.