• Korean Journal of Remote Sensing
• /
• v.34 no.1
• /
• pp.151-166
• /
• 2018

This paper compares the four selections of performance used in the application of genetic algorithms (GAs) to automatically optimize multispectral pixel cluster for unsupervised classification from KOMPSAT-3 data, since the selection among three main types of operators including crossover and mutation is the driving force to determine the overall operations in the clustering GAs. Experimental results demonstrate that the tournament selection obtains a better performance than the other selections, especially for both the number of generation and the convergence rate. However, it is computationally more expensive than the elitism selection with the slowest convergence rate in the comparison, which has less probability of getting optimum cluster centers than the other selections. Both the ranked-based selection and the proportional roulette wheel selection show similar performance in the average Euclidean distance using the pixel clustering, even the ranked-based is computationally much more expensive than the proportional roulette. With respect to finding global optimum, the tournament selection has higher potential to reach the global optimum prior to the ranked-based selection which spends a lot of computational time in fitness smoothing. The tournament selection-based clustering GA is used to successfully classify the KOMPSAT-3 multispectral data achieving the sufficient the matic accuracy assessment (namely, the achieved Kappa coefficient value of 0.923).

• ETRI Journal
• /
• v.45 no.3
• /
• pp.448-461
• /
• 2023

This article performs a detailed data scrutiny on a chronic kidney disease (CKD) dataset to select efficient instances and relevant features. Data relevancy is investigated using feature extraction, hybrid outlier detection, and handling of missing values. Data instances that do not influence the target are removed using data envelopment analysis to enable reduction of rows. Column reduction is achieved by ranking the attributes through feature selection methodologies, namely, extra-trees classifier, recursive feature elimination, chi-squared test, analysis of variance, and mutual information. These methodologies are ranked via Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) using weight optimization to identify the optimal features for model building from the CKD dataset to facilitate better prediction while diagnosing the severity of the disease. An efficient hybrid ensemble and novel similarity-based classifiers are built using the pruned dataset, and the results are thereafter compared with random forest, AdaBoost, naive Bayes, k-nearest neighbors, and support vector machines. The hybrid ensemble classifier yields a better prediction accuracy of 98.31% for the features selected by extra tree classifier (ETC), which is ranked as the best by TOPSIS.

• Journal of information and communication convergence engineering
• /
• v.21 no.1
• /
• pp.82-89
• /
• 2023

Various machine-learning models may yield high predictive power for massive time series for time series prediction. However, these models are prone to instability in terms of computational cost because of the high dimensionality of the feature space and nonoptimized hyperparameter settings. Considering the potential risk that model training with a high-dimensional feature set can be time-consuming, we evaluate a feature-importance-based feature selection method to derive a tradeoff between predictive power and computational cost for time series prediction. We used two machine learning techniques for performance evaluation to generate prediction models from a retail sales dataset. First, we ranked the features using impurity- and Local Interpretable Model-agnostic Explanations (LIME) -based feature importance measures in the prediction models. Then, the recursive feature elimination method was applied to eliminate unimportant features sequentially. Consequently, we obtained a subset of features that could lead to reduced model training time while preserving acceptable model performance.

• Korea Trade Review
• /
• v.44 no.1
• /
• pp.269-283
• /
• 2019

Due to environmental issues such as global warming, the importance of renewable energy is growing. Solar Power System is one of the most growing eco-friendly energy industries in the world, but Korea's solar energy industry faces fierce competition due to the trade regulations and changes in energy related laws in the major markets such as the U.S., EU and China. Therefore, Korea needs to diversify its export markets towards emerging markets. This paper analyzed 162 countries in the world and developed a model to measure how promising the countries are. GSMI(Grid connected Solar Market Index) and OSMI(Off-grid Solar Market Index) are invented based on the models. By using the developed model and the data of 162 countries over the 15-year period from 2000 to 2014, the foreign markets are ranked for searching the export market. According to the analysis, China, Japan, U.S, India and Taiwan ranked first to fifth in GSMI and OSMI ranking, which were followed by China, India, Bangladesh, Philippines and Afghanistan. The model developed through this research is expected to provide a more reasonable and scientific approach to the advancement of the Korean solar energy industry into overseas markets.

• Korean Journal of Ecology and Environment
• /
• v.56 no.4
• /
• pp.393-405
• /
• 2023

This study was conducted to select indicators for assessing national biodiversity. For this purpose, 140 biodiversity-related indicators were identified as a result of inventorying biodiversity-related indicators used in Korea and abroad, and when these indicators were applied to the pressure, status, and response indicator system, it was found that status indicators accounted for the largest number of indicators, with 29 pressure, 59 status, and 44 response. We also categorized the status indicators into genes, species, habitat, function, and quality, and found that species and habitat indicators accounted for the majority. Pressure indicators were categorized into direct exploitation, pollution, alien species, climate change, and habitat change. As a result, it was found that direct exploitation and pollution accounted for most of the pressure indicators. In addition, this study used internationally used indicator selection criteria to establish criteria for selecting domestic biodiversity assessment indicators. Using this list of indicators and indicator selection criteria, we evaluated the prioritization of domestically applicable biodiversity indicators through relevant expert consultations. 1) Vegetation class, 2) Land cover indicators, and 3) Change of protected area ranked highly. In fact, these indicators have been used in many studies due to the availability of assessable data. However, most of the highly scored indicators are based on ecosystem area, and further consideration of ecosystem functions and components(species) is needed.

• The Journal of the Korea Contents Association
• /
• v.20 no.11
• /
• pp.38-48
• /
• 2020

The domestic AI speaker market is growing into a full-fledged early audience market beyond the innovative consumer market with 3 million domestic supply units at the end of 2018, but the reality is that for various reasons, we are not satisfied with the use. There are many previous papers on AI Speaker, but the majority of research so far tends to be biased towards the acceptance of the device's own performance. Many changes are being made, such as OTT providers trying to secure the market through collaboration with AI speaker providers. This study tried to identify the priorities for content services, which can be another major selection factor for AI speakers, excluding the factors of unsatisfactory technology. First, this study identified the priorities among AI speaker selection factors using AHP (Analytic Hierarchy Process), based on the AI speaker selection factors derived through literature research. The most important hierarchical factor are Concierge Service, Education Service, and Entertainment Service order in AI speaker selection, and the primary content among the individual factors was the one that ranked weather/temperature/fine dust (11.6%) and child caring content was in the second place (10.8%), and then music service was in the third place (9.8%). The three top priorities were derived from the items in the top tier 1, 2 and 3 priorities. Of the total 15 individual services, 6 sub-layers of Concierge Service (weather/temperature/fine dust, news, voice schedule notification) and Education Service (foreign language, toddler, reading books) were in the top 8, and two of the Entertainment Service Music service and movie service ranked third and sixth.

• Journal of Information Processing Systems
• /
• v.14 no.5
• /
• pp.1167-1175
• /
• 2018

Microarray data plays an essential role in diagnosing and detecting cancer. Microarray analysis allows the examination of levels of gene expression in specific cell samples, where thousands of genes can be analyzed simultaneously. However, microarray data have very little sample data and high data dimensionality. Therefore, to classify microarray data, a dimensional reduction process is required. Dimensional reduction can eliminate redundancy of data; thus, features used in classification are features that only have a high correlation with their class. There are two types of dimensional reduction, namely feature selection and feature extraction. In this paper, we used k-means algorithm as the clustering approach for feature selection. The proposed approach can be used to categorize features that have the same characteristics in one cluster, so that redundancy in microarray data is removed. The result of clustering is ranked using the Relief algorithm such that the best scoring element for each cluster is obtained. All best elements of each cluster are selected and used as features in the classification process. Next, the Random Forest algorithm is used. Based on the simulation, the accuracy of the proposed approach for each dataset, namely Colon, Lung Cancer, and Prostate Tumor, achieved 85.87%, 98.9%, and 89% accuracy, respectively. The accuracy of the proposed approach is therefore higher than the approach using Random Forest without clustering.

• Proceedings of the Korea Society for Energy Engineering kosee Conference
• /
• 1998.10a
• /
• pp.3-8
• /
• 1998

The present work proposes the potential fuzzy framework, based on fuzzy set theory, for supporting decision-making problems, especially, selection problems of a best design in the area of nuclear energy system. The framework proposed is composed of the hierarchical structure module, the assignment module, the fuzzification module, and the defuzzification module. In the structure module, the relationship among decision objectives, decision criteria, decision sub-criteria, and decision alternatives is hierarchically structured. In the assignment module, linguistic or rank scoring approach can be used to assign subjective and/or vague values to the decision analyst's judgment on decision variables. In the fuzzification module, fuzzy numbers are assigned to these values of decision variables. Using fuzzy arithmetic operations, for each alternative, fuzzy preference index as a fuzzy synthesis measure is obtained. In the defuzzification module, using one of methods ranking fuzzy numbers, these indices are defuzzified to overall utility values as a cardinality measure determining final scores. According these values, alternatives of interest are ranked and an optimal alternative is chosen. To illustrate the applicability of the framework proposed to selection problem, as a case example, the best option choice of four design options under five decision criteria for primary containment wall thickening around large penetrations in an advanced nuclear energy system is studied.

• The KIPS Transactions:PartD
• /
• v.14D no.3 s.113
• /
• pp.347-354
• /
• 2007

It is not easy for the customers to search the valuable information on the goods among countless items available in the Internet. In order to save time and efforts in searching the goods the customers want, it is very important for a recommender system to have a capability to predict accurately customers' preferences. In this paper we present a refined neighbor selection algorithm for clustering based collaborative filtering in recommender systems. The algorithm exploits a graph approach and searches more efficiently for set of influential customers with respect to a given customer; it searches with concepts of weighted similarity and ranked clustering. The experimental results show that the recommender systems using the proposed method find the proper neighbors and give a good prediction quality.

• Journal of Environmental Health Sciences
• /
• v.45 no.2
• /
• pp.164-172
• /
• 2019

Objectives: This study's purpose is finding children's activity spaces that demand environmental safety management. Methods: The method of this study is analysing children's life patterns based on a questionnaire survey. Results: This study analyzed children's life patterns through a questionnaire survey. In total, 2,447 questionnaires were provided to analyze children's life patterns. The results of the questionnaire indicated a highly simple form because many children generally stayed in their home (66%) or nursery facility (2%). In the case of other facilities, playground was ranked first and amusement park was ranked second. In addition, kids cafe (including play facilities installed in shopping centers, etc.), library, and internet cafe were among the responses. Conclusions: The priority for new high-maintenance children's activity spaces are academy (rank 1), kids cafe (rank 2), indoor playground (rank 3).