• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2011.04a
• /
• pp.224-229
• /
• 2011

In this paper, in order to improve safety of hybrid electric vehicle a fault detection algorithm is introduced. The proposed algorithm uses SVDD techniques. Two methods for learning a lot of data are used in this technique. One method is to learn the data incrementally. Another method is to remove the data that does not affect the next learning. Using lines connecting support vectors selection of removing data is made. Using this method, lot of computation time and storage can be saved while learning many data. A battery data of commercial hybrid electrical vehicle is used in this study. In the study fault boundary via SVDD is described and relevant algorithm for virtual fault data is verified. It takes some time to generate fault boundary, nevertheless once the boundary is given, fault diagnosis can be conducted in real time basis.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.2
• /
• pp.101-112
• /
• 1995

In order to construct a feature map-based phoneme classification system for speech recognition, two procedures are usually required. One is clustering and the other is labeling. In this paper, we present a phoneme classification system based on the Kohonen's Self-Organizing Feature Map (SOFM) for clusterer and labeler. It is known that the SOFM performs self-organizing process by which optimal local topographical mapping of the signal space and yields a reasonably high accuracy in recognition tasks. Consequently, SOFM can effectively be applied to the recognition of phonemes. Besides to improve the performance of the phoneme classification system, we propose the learning algorithm combined with the classical K-mans clustering algorithm in fine-tuning stage. In order to evaluate the performance of the proposed phoneme classification algorithm, we first use totaly 43 phonemes which construct six intra-class feature maps for six different phoneme classes. From the speaker-dependent phoneme classification tests using these six feature maps, we obtain recognition rate of $87.2\%$ and confirm that the proposed algorithm is an efficient method for improvement of recognition performance and convergence speed.

• YAKHAK HOEJI
• /
• v.58 no.2
• /
• pp.141-149
• /
• 2014

Background: Oriental lifestyle for treating diseases has been developed and well-accepted for a long time among Koreans. Sasang Constitution theory, originated from Korean traditional medicine, suggests that medication treatment should be differentiated by each patient's body classification (So-yang [SY], So-eum [SE], Tae-yang [TY], and Tae-eum [TE]), in contrary to western medicine's theory that medication should be applied equally by disease indication without such classification. However, the pharmacotherapeutic outcomes of these theories have not been compared to date. In this study, we aimed to compare the two theories by evaluating blood pressure (BP), which is lowered as a therapeutic outcome, among hypertensive patients taking angiotensin II receptor blockers (ARBs) or calcium channel blockers (CCBs), two most commonly used antihypertensive classes in Korea. Methods: From April 2006 to June 2012, we retrospectively collected data on hypertensive patients with Sasang Constitution classification at Kyunghee University Hospital at Gangdong, one of the East-West collaborative medical centers in Korea. We collected information on age, gender, underlying diseases, antihypertensive drugs (ARB, CCB, ARB+CCB), and BP by reviewing the electronic medical records. We excluded patients with missing blood pressure at baseline or follow-up, or those who had a change in their antihypertensive drug class during follow-up. Results: We selected a total of 573 patients (SY: 165, SE: 158, TY: 0, TE: 250). Baseline BPs were on average 139.0/82.0 mmHg for SY, 137.8/78.5 mmHg for SE, and 138.7/79.2 mmHg for TE. In all three groups, CCBs were the most prescribed, followed by combination therapy with ARB+CCB, then ARBs. BP reduction after 1 month of initial medication was significantly different among the drug classes, but not in Sasang constitutional classification (ARB [SY: -12.4/-4.7, SE: -12.3/-2.5, TE: -8.6/-1.8], CCB [SY: -12.3/-5.4, SE: -13.0/-2.3, TE: -10.8/-6.0], ARB+CCB [SY: -15.6/-6.7, SE: -18.4/-8.1, TE: -20.2/-6.7], drug [$P{\leq}0.05$/P>0.05], constitutional type [P>0.05/P>0.05]). Conclusion: We observed significant differences in reduction of blood pressure by classes of drugs (ARB+CCB>CCB>ARB) but not by Sasang constitutional classification. Therefore, current approach of antihypertensive pharmacotherapy assisted by Western medicine is appropriate for treatment of hypertension. However, further larger scale or prospective studies are required in order to confirm these results.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.27 no.2
• /
• pp.71-86
• /
• 2022

The water quality index (WQI) has been widely used to evaluate marine water quality. The WQI in Korea is categorized into five classes by marine environmental standards. But, the WQI calculation on huge datasets is a very complex and time-consuming process. In this regard, the current study proposed machine learning (ML) based models to predict WQI class by using water quality datasets. Sihwa Lake, one of specially-managed coastal zone, was selected as a modeling site. In this study, adaptive boosting (AdaBoost) and tree-based pipeline optimization (TPOT) algorithms were used to train models and each model performance was evaluated by metrics (accuracy, precision, F1, and Log loss) on classification. Before training, the feature importance and sensitivity analysis were conducted to find out the best input combination for each algorithm. The results proved that the bottom dissolved oxygen (DO_Bot) was the most important variable affecting model performance. Conversely, surface dissolved inorganic nitrogen (DIN_Sur) and dissolved inorganic phosphorus (DIP_Sur) had weaker effects on the prediction of WQI class. In addition, the performance varied over features including stations, seasons, and WQI classes by comparing spatio-temporal and class sensitivities of each best model. In conclusion, the modeling results showed that the TPOT algorithm has better performance rather than the AdaBoost algorithm without considering feature selection. Moreover, the WQI class for unknown water quality datasets could be surely predicted using the TPOT model trained with satisfactory training datasets.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.17 no.4
• /
• pp.101-111
• /
• 2014

Spatial events are predictable using data mining classification algorithms. Decision trees have been used as one of representative classification algorithms. And they were normally used in the classification tasks that have label class values. However since using rule ranking methods, spatial prediction have been applied in the spatial prediction problems. This paper compared rule ranking methods for the spatial prediction application using a decision tree. For the comparison experiment, C4.5 decision tree algorithm, and rule ranking methods such as Laplace, M-estimate and m-branch were implemented. As a spatial prediction case study, landslide which is one of representative spatial event occurs in the natural environment was applied. Among the rule ranking methods, in the results of accuracy evaluation, m-branch showed the better accuracy than other methods. However in case of m-brach and M-estimate required additional time-consuming procedure for searching optimal parameter values. Thus according to the application areas, the methods can be selectively used. The spatial prediction using a decision tree can be used not only for spatial predictions, but also for causal analysis in the specific event occurrence location.

• The Korean Journal of Pesticide Science
• /
• v.9 no.2
• /
• pp.140-145
• /
• 2005

Quantitative Structure Activity Relationship (QSAR) assumes the relatedness between physical property and biological activity. However, activity data measured at single concentration such as percent activity have not been used extensively for modeling purpose. This probably comes from the fact that these values are qualitative instead of quantitative. To utilize percent activity data for molecular modeling, we classified the whole data into two classes. One class represents the active while the other signifies the inactive. The percent activity data of ${\beta}$-Ketoacetoanilides measured for TLB (Tomato Late Blight) were investigated. CoMFA (Comparative Molecular Field Analysis) was used as a discriminant function. Using CoMFA provides 3D (three dimensional) information, which is crucial for chemical insight. It can also serve as a predictive model. The resultant model classified the given data correctly (98%). When LOO (leave-one-out) crossvalidation procedure was applied, the classification accuracy was 69%. Therefore two class approximation of percent activity data with CoMFA can be utilized to understand the relationship between chemical structure and biological activity and design subsequent chemical analogs.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.18 no.5
• /
• pp.718-723
• /
• 2008

Bearing is one of the important mechanical elements used in various industrial equipments. Most of failures occurred during the equipment operation result from bearing defects and breakages. Therefore, monitoring of bearings is essential in preventing equipment breakdowns and reducing unexpected loss. The purpose of this paper is to present an online monitoring method to predict bearing states using vibration signals. Bearing vibrations, which are collected as a form of profile signal, are first analyzed by a discrete wavelet transform. Next, some statistical features are obtained from the resultant wavelet coefficients. In order to select significant ones among them, analysis of variance (ANOVA) is employed in this paper. Statistical features screened in this way are used as input variables to support vector machine (SVM). An hierarchical SVM tree is proposed for dealing with multi-class problems. The result of numerical experiments shows that the proposed SVM tree has a competent performance for classifying bearing fault states.

• Journal of KIISE:Software and Applications
• /
• v.37 no.2
• /
• pp.155-160
• /
• 2010

This paper proposes a method for Korean comparative sentence classification which is a part of comparison mining. Comparison mining, one area of text mining, analyzes comparative relations from the enormous amount of text documents. Three-step process is needed for comparison mining - 1) identifying comparative sentences in the text documents, 2) classifying those sentences into several classes, 3) analyzing comparative relations per each comparative class. This paper aims at the second task. In this paper, we use transformation-based learning (TBL) technique which is a well-known learning method in the natural language processing. In our experiment, we classify comparative sentences into seven classes using TBL and achieve an accuracy of 80.01%.

• Journal of Computing Science and Engineering
• /
• v.5 no.3
• /
• pp.183-196
• /
• 2011

There has recently been a surge of interest in relational database mining that aims to discover useful patterns across multiple interlinked database relations. It is crucial for a learning algorithm to explore the multiple inter-connected relations so that important attributes are not excluded when mining such relational repositories. However, from a data privacy perspective, it becomes difficult to identify all possible relationships between attributes from the different relations, considering a complex database schema. That is, seemingly harmless attributes may be linked to confidential information, leading to data leaks when building a model. Thus, we are at risk of disclosing unwanted knowledge when publishing the results of a data mining exercise. For instance, consider a financial database classification task to determine whether a loan is considered high risk. Suppose that we are aware that the database contains another confidential attribute, such as income level, that should not be divulged. One may thus choose to eliminate, or distort, the income level from the database to prevent potential privacy leakage. However, even after distortion, a learning model against the modified database may accurately determine the income level values. It follows that the database is still unsafe and may be compromised. This paper demonstrates this potential for privacy leakage in multi-relational classification and illustrates how such potential leaks may be detected. We propose a method to generate a ranked list of subschemas that maintains the predictive performance on the class attribute, while limiting the disclosure risk, and predictive accuracy, of confidential attributes. We illustrate and demonstrate the effectiveness of our method against a financial database and an insurance database.

• Journal of KIISE:Software and Applications
• /
• v.34 no.1
• /
• pp.11-17
• /
• 2007

One of the popular methods used for pattern classification is the MBR (Memory-Based Reasoning) algorithm. Since it simply computes distances between a test pattern and training patterns or hyperplanes stored in memory, and then assigns the class of the nearest training pattern, it cannot explain how the classification result is obtained. In order to overcome this problem, we propose an incremental teaming algorithm based on RPA (Recursive Partition Averaging) to extract IF-THEN rules that describe regularities inherent in training patterns. But rules generated by RPA eventually show an overfitting phenomenon, because they depend too strongly on the details of given training patterns. Also RPA produces more number of rules than necessary, due to over-partitioning of the pattern space. Consequently, we present the IREA (Incremental Rule Extraction Algorithm) that overcomes overfitting problem by removing useless conditions from rules and reduces the number of rules at the same time. We verify the performance of proposed algorithm using benchmark data sets from UCI Machine Learning Repository.