• Journal of the Korean Statistical Society
• /
• v.36 no.1
• /
• pp.175-182
• /
• 2007

The support vector machine has been successful in many applications because of its flexibility and high accuracy. However, when a training data set is large or imbalanced, the support vector machine may suffer from significant computational problem or loss of accuracy in predicting minority classes. We propose a modified version of the support vector machine using the K-means clustering that exploits the information in class labels during the clustering process. For large data sets, our method can save the computation time by reducing the number of data points without significant loss of accuracy. Moreover, our method can deal with imbalanced data sets effectively by alleviating the influence of dominant class.

• Journal of the Korea Society of Computer and Information
• /
• v.21 no.2
• /
• pp.71-77
• /
• 2016

There has been unknown polynomial time algorithm for cellular formation problem (CFP) that is one of the NP-hard problem. Therefore metaheuristic method has been applied this problem to obtain approximated solution. This paper shows the existence of polynomial-time heuristic algorithm in CFP. The proposed algorithm performs coarse-grained and fine-grained cell formation process. In coarse-grained cell formation process, the cell can be formed in accordance with machine frequently used that is the number of other products use same machine with special product. As a result, the machine can be assigned to most used cell. In fine-grained process, the product and machine are moved into other cell that has a improved grouping efficiency. For 35 experimental data, this heuristic algorithm performs better grouping efficiency for 12 data than best known of meta-heuristic methods.

• International journal of advanced smart convergence
• /
• v.10 no.4
• /
• pp.165-171
• /
• 2021

The current study investigated cyber learners' use and perceptions of online machine translation (MT) tools. The results show that learners use several MT tools frequently and extensively for various second language learning (L2) purposes according to their needs. The learners' overall perceptions of using MT for English learning were generally positive. The learners reported several advantages of machine translation: ease of use, helpful feedback, effective revision, and facilitation of self-directed learning. At the same time, a considerable number of learners were aware of MT's drawbacks, such as awkward sentences, inaccurate grammar, and inappropriate words, and thus held a negative or skeptical view on the quality and accuracy of MT. These findings have important pedagogical implications for using MT in the context of a cyber university. For successful integration of MT in English classes, teachers need to provide appropriate guidelines and training that will help learners use MT effectively.

• Nuclear Engineering and Technology
• /
• v.53 no.12
• /
• pp.4060-4066
• /
• 2021

Flow-accelerated corrosion (FAC) of carbon steel piping is a significant problem in nuclear power plants. The basic process of FAC is currently understood relatively well; however, the accuracy of prediction models of the wall-thinning rate under an FAC environment is not reliable. Herein, we propose a methodology to construct pipe wall-thinning rate prediction models using artificial neural networks and a convolutional neural network, which is confined to a straight pipe without geometric changes. Furthermore, a methodology to generate training data is proposed to efficiently train the neural network for the development of a machine learning-based FAC prediction model. Consequently, it is concluded that machine learning can be used to construct pipe wall thinning rate prediction models and optimize the number of training datasets for training the machine learning algorithm. The proposed methodology can be applied to efficiently generate a large dataset from an FAC test to develop a wall thinning rate prediction model for a real situation.

• Journal of IKEEE
• /
• v.22 no.4
• /
• pp.948-952
• /
• 2018

Haze removal is an object of scientific desire due to its various practical applications. Existing algorithms are founded upon histogram equalization, contrast maximization, or the growing trend of applying machine learning in image processing. Since machine learning-based algorithms solve problems based on the data, they usually perform better than those based on traditional image processing/computer vision techniques. However, to achieve such a high performance, one of the requisites is a large and reliable training database, which seems to be unattainable owing to the complexity of real hazy and haze-free images acquisition. As a result, researchers are currently using the synthetic database, obtained by introducing the synthetic haze drawn from the standard uniform distribution into the clear images. In this paper, we propose the enhanced equidistribution, improving upon our previous study on equidistribution, and use it to make a new database for training machine learning-based haze removal algorithms. A large number of experiments verify the effectiveness of our proposed methodology.

• Journal of information and communication convergence engineering
• /
• v.21 no.3
• /
• pp.225-232
• /
• 2023

Owing to advancements in intelligent transportation systems (ITS) and artificial-intelligence technologies, various machine-learning models can be employed to simulate and predict the number of traffic accidents under different weather conditions. Furthermore, we can analyze the relationship between weather and traffic accidents, allowing us to assess whether the current weather conditions are suitable for travel, which can significantly reduce the risk of traffic accidents. In this study, we analyzed 30000 traffic flow data points collected by traffic cameras at nearby intersections in Washington, D.C., USA from October 2012 to May 2017, using Pearson's heat map. We then predicted, analyzed, and compared the performance of the correlation between continuous features by applying several machine-learning algorithms commonly used in ITS, including random forest, decision tree, gradient-boosting regression, and support vector regression. The experimental results indicated that the gradient-boosting regression machine-learning model had the best performance.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.7
• /
• pp.113-123
• /
• 2020

Obstructive sleep apnea (OSA) among sleep disorders is one of relatively common diseases. Patients can be checked for the disease through sleep polysomnography. However, as far as he diagnosis of OSA using polysomnography (PSG) is concerned, many practical problems such as an increasing number of patients, expensive testing cost, discomfort during examination, and the limited number of people for testing have been pointed out. Accordingly, for the purpose of substituting PSG researchers have been actively conducting studies on OSA diagnosis based on machine learning using bio signals. In this regard, we review a rich body of existing OSA diagnosis studies applying machine learning techniques based on bio-signal data. As a result, this paper presents a novel taxonomy of the reviewed studies and provides their comprehensive comparative analysis results. Also, we reveal various limitations of the studies using the bio signals and suggest several improvements about utilization of the used machine learning methods. Finally, this paper presents future research topics related to the application of machine learning techniques using bio signals.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_2
• /
• pp.771-783
• /
• 2023

This study aims to compare the performance of each machine learning model for preparing a grid-based disaster risk map related to flooding in Jung-gu, Ulsan, for Typhoon Chaba which occurred in 2016. Dynamic data such as rainfall and river height, and static data such as building, population, and land cover data were used to conduct a risk analysis of flooding disasters. The data were constructed as 10 m-sized grid data based on the national point number, and a sample dataset was constructed using the risk value calculated for each grid as a dependent variable and the value of five influencing factors as an independent variable. The total number of sample datasets is 15,910, and the training, verification, and test datasets are randomly extracted at a 6:2:2 ratio to build a machine-learning model. Machine learning used random forest (RF), support vector machine (SVM), and k-nearest neighbor (KNN) techniques, and prediction accuracy by the model was found to be excellent in the order of SVM (91.05%), RF (83.08%), and KNN (76.52%). As a result of deriving the priority of influencing factors through the RF model, it was confirmed that rainfall and river water levels greatly influenced the risk.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.48 no.3
• /
• pp.187-194
• /
• 2012

This study surveyed the fishing efficiency for Japanese common squid based on improvements made to LED fishing lamps by utilizing the training ship of Gangwon Provincial College. The training ship, Haesong-ho (24 tons), was equipped with seventy-two 150W electric power LED fishing lamps (10.8kW), and their fishing efficiency and fuel consumption level were surveyed for a total of ten times during the period between June 15, 2009 and July 27, 2009. In addition, the training ship was equipped with seventy-one 300W electric power LED fishing lamps (21.3kW) and their fishing efficiency and fuel consumption level were surveyed for a total of five times during the period between January 17, 2010 and August 4, 2010. In order to compare the fishing efficiency of LED fishing lamps, the catch of another fishing vessel equipped with Metal Halide (MH) fishing lamps of 120kW for the same period and at the same fishing grounds. The fuel consumption levels during the fishing operation of Haesong-ho was about 1,047.7 liters which was approximately 19.9% of the total fuel consumption level of 5,262.6 liters, and the fuel consumption level per operation hour was on average 9.2 liters. The number of Japanese common squid caught by the LED fishing lamp-equipped fishing vessel ranged from 12 to 1,640 squid for each fishing trial and the average quantity was 652. The number of Japanese common squid caught by the MH fishing lamp-equipped 10 fishing vessels ranged from 40 to 4,800 squid and the average quantity was 2,055. The fishing of Japanese common squid was done by the use of hand lines and an automatic jigging machine. The number of Japanese common squid caught per hand line and a single roller of the automatic jigging machine was in the proportion of 50.8% to 49.2% with respect to the LED fishing lamp-equipped fishing vessel. However, the number of Japanese common squid caught per hand line and a single roller of the automatic jigging machine was in the proportion of 86.4% to 13.6% with respect to the MH fishing lamp-equipped fishing vessel where most of the catch was made by hand lines. On the other hand, in comparing the number of Japanese common squid caught per automatic jigging machine, the number of squid caught by the LED fishing lamp-equipped fishing vessel was about the same or greater than the number of squid caught by the MH fishing lamp-equipped fishing vessel.

• Korean Journal of Artificial Intelligence
• /
• v.5 no.1
• /
• pp.29-37
• /
• 2017

The range of problems that can be handled by the activation of big data and the development of hardware has been rapidly expanded and machine learning such as deep learning has become a very versatile technology. In this paper, mnist data set is used as experimental data, and the Cross Entropy function is used as a loss model for evaluating the efficiency of machine learning, and the value of the loss function in the steepest descent method is We applied the Gradient Descent Optimize algorithm to minimize and updated weight and bias via backpropagation. In this way we analyze optimal reliability value corresponding to the number of exercises and optimal reliability value without overfitting. And comparing the overfitting time according to the number of data changes based on the number of training times, when the training frequency was 1110 times, we obtained the result of 92%, which is the optimal reliability value without overfitting.