• Journal of the Korean Chemical Society
• /
• v.53 no.6
• /
• pp.653-662
• /
• 2009

Developing effective tools for predicting absorption, distribution, metabolism, excretion properties and toxicity (ADME/T) of new chemical entities in the early stage of drug design is one of the most important tasks in drug discovery and development today. As one of these attempts, support vector machines (SVM) has recently been exploited for the prediction of ADME/T related properties. However, two problems in SVM modeling, i.e. feature selection and parameters setting, are still far from solved. The two problems have been shown to be crucial to the efficiency and accuracy of SVM classification. In particular, the feature selection and optimal SVM parameters setting influence each other, which indicates that they should be dealt with simultaneously. In this account, we present an integrated practical solution, in which genetic-based algorithm (GA) is used for feature selection and grid search (GS) method for parameters optimization. hERG ion-channel inhibitor classification models of ADME/T related properties has been built for assessing and testing the proposed GA-GS-SVM. We generated 6 different models that are 3 different single models and 3 different ensemble models using training set - 1891 compounds and validated with external test set - 175 compounds. We compared single model with ensemble model to solve data imbalance problems. It was able to improve accuracy of prediction to use ensemble model.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.6
• /
• pp.167-176
• /
• 2005

This paper describes how the state tying model based on the decision tree which is one of Acoustic models used for speech recognition optimizes the model by reducing the number of mixture Gaussians of the output probability distribution. The state tying modeling uses a finite set of questions which is possible to include the phonological knowledge and the likelihood based decision criteria. And the recognition rate can be improved by increasing the number of mixture Gaussians of the output probability distribution. In this paper, we'll reduce the number of mixture Gaussians at the highest point of recognition rate by clustering the Gaussians. Bhattacharyya and Euclidean method will be used for the distance measure needed when clustering. And after calculating the mean and variance between the pair of lowest distance, the new Gaussians are created. The parameters for the new Gaussians are derived from the parameters of the Gaussians from which it is born. Experiments have been performed using the STOCKNAME (1,680) databases. And the test results show that the proposed method using Bhattacharyya distance measure maintains their recognition rate at $97.2\%$ and reduces the ratio of the number of mixture Gaussians by $1.0\%$. And the method using Euclidean distance measure shows that it maintains the recognition rate at $96.9\%$ and reduces the ratio of the number of mixture Gaussians by $1.0\%$. Then the methods can optimize the state tying model.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.26 no.1
• /
• pp.59-67
• /
• 2013

Complex-shaped tall building causes many structural challenges due to its structural characteristics regarding inclined members and complexed shape. This paper is aimed at development of design process using computational-platform which is effective design tool for responding frequent design changes, particularly as to overseas projects. StrAuto, a parametric structural modeling and optimizing system, provides the optimized alternatives according to design intent and realize a swift process converting a series of structural information necessary to nonlinear analytical models. The application of the process was to a 45-story hotel building in Ulanbator, Mongolia adopting shear wall and special moment frame with outrigger systems. To investigate the safety of lateral force resisting system against maximum considered earthquake(MCE), nonlinear response history analysis was conducted using StrAuto.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.17 no.11 s.114
• /
• pp.1120-1131
• /
• 2006

This paper describes the numerical analysis techniques using the Combined Integration/Matrix Method to calculate ground potential rise which can be occurred in the various grounding systems. Combined Integration/Matrix Method is used to reduce the error and computation time with the analytical integration equation and the proper segmentaion of earth embedded conductor. To do it, optimal segmentaion method for the buried conductors is presented through error analysis which is capable of applying the practical scaled various grounding systems. The optimum length of segmented element is much co-related with the buried depth of grounding electrode and the maximum length of buried electrode. As a result, less 3 precent errors was obtained by proposed model. The proposed model is applied to verify an effect of multi-grounding problems which was aroused much controversy with separated or common grounding between the high power grounding system and low power grounding system such as signal and telecommunication grounding.

• Journal of the Institute of Convergence Signal Processing
• /
• v.3 no.1
• /
• pp.8-14
• /
• 2002

In this paper, we realized a real-time text-independent speaker recognition system using gaussian mixture model, and applied frame level likelihood normalization method which shows its effects in verification system. The system has three parts as front-end, training, recognition. In front-end part, cepstral mean normalization and silence removal method were applied to consider speaker's speaking variations. In training, gaussian mixture model was used for speaker's acoustic feature modeling, and maximum likelihood estimation was used for GMM parameter optimization. In recognition, likelihood score was calculated with speaker models and test data at frame level. As test sentences, we used text-independent sentences. ETRI 445 and KLE 452 database were used for training and test, and cepstrum coefficient and regressive coefficient were used as feature parameters. The experiment results show that the frame-level likelihood method's recognition result is higher than conventional method's, independently the number of registered speakers.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2012.05a
• /
• pp.118-118
• /
• 2012

수자원의 효율적 이용 및 관리는 심화되고 있는 기후변화에 선제적으로 대응하고 발생가능한 물위기에 대비하기 위한 필수조건이다. 그 중 가장 핵심이 되는 요소는 댐에 저수된 물을 효과적으로 이용하는 것, 즉 댐 건설목적에 따라 시간 및 공간별로 적절하게 할당시키는 것이라고 할 수 있다(Kim, 1998). 그러나 단일 댐의 운영과는 달리 수계내 댐군의 연계 운영은 매우 복잡하고 어려운 문제이다. 연계된 댐들간 저수 상황을 고려하여 유역내 시 공간적인 용수 수요의 지속적인 충족을 위하여 유입량 예측의 정확성을 높이도록 하고, 상류 댐에 최대한 저류하도록 하며, 여수로 방류 같은 불필요한 방류를 최소화 하고, 서로 상충되는 목표를 갖고 있지만, 홍수용량 및 발전수위를 최대로 확보하도록 하여야 한다. 이처럼 댐 운영을 위한 실제 상황은 단일 목적에 의한 최적화와는 달리 여러 상충되는 목적 및 구성 요소들간의 타협, 조정을 필요로 한다. 댐군의 연계운영 문제는 1960년대 초부터 현재까지 활발히 연구가 진행되어 온 분야 중 하나이나 문제의 복잡성과 어려움으로 인해 아직까지도 최선의 방안을 제시하기 어려운 문제이다(ReVelle, 2000). 이를 위한 방법은 시뮬레이션 모형 활용기법과 최적화 모형 활용기법으로 대별할 수 있으며 각 방법의 서로 다른 구조적 특성과 장단점으로 인하여 이원화된 체계로 사용되는 것이 현재의 국내 실정이다. 대부분의 실무에서는 이해도도 쉽고, 비교적 결과를 빨리 도출할 수 있는 시뮬레이션 기반의 모형을 활용하며 대표적으로 HEC-5, K-ModSim, HEC-ResSim 등이 활용되어왔다. 반면, 학계에서는 DP, MIP, SLP, SDP 등 최적화기법을 댐운영에 활용 할 것을 제안하고 있지만, 활용에 대한 거부감이 남아있는 것이 현실이다. 본 연구에서는 시뮬레이션과 최적화기법의 원론적 비교를 통해 각 방법의 장단점과 한계점을 분석하고, 왜 이원화된 사용체계로 되었는지에 대한 고찰과 이에 대한 해결책으로 시뮬레이션모형의 장점과 최적화기법의 장점을 결합한 모형을 제안한다. 국내에는 Kim and Park(1998)이 시뮬레이션 기반의 최적화 모형 CoMOM(Coordinated Multi-Reservoir Operating Model)을 개발하였으며, 이후 21C프론티어 연구사업(2001-2011)에서 모형의 보완수정 검증을 통해 실무 활용도를 높여 왔다. 본 연구를 통하여 거부감의 원천을 추적해 보고, 타당한 이유가 있는지 대한 것을 심층 분석해보고, CoMOM모형과 시뮬레이션 모형, 다른 최적화 기법들과의 원론적 비교를 통해 각 방법들의 효율적인 활용방안과 최적화모형의 구체적인 활용방안을 제시하고자 한다.

• Journal of Biomedical Engineering Research
• /
• v.24 no.6 s.81
• /
• pp.515-522
• /
• 2003

Pharynx is a system transporting foods by peristaltic motion(contraction and expansion movement! into the esophagus and functioning as airway passages. In this study, structural changes of pharyngeal dysfunction are analyzed by biomechanical model using CT and FEM(finite clement method). Loading condition was assumed that equal pressure was loaded sequentially to inside of pharyngeal tissue. In order to analyze the pharyngeal muscular dysfunction by biomechanical model. the pharyngeal dysfunctions was classified into 3 cases. Taking into account the clinical complication by neuromuscular symptoms such as pharyngeal dysfunction after stroke. we assumed that a change of material property is caused by muscular tissue stiffness. A deformation of cross sectional area of the pharynx is analyzed increasing the stiffness $25\%,\;50\%,\;75\%$ in each case on the basis of stress-strain relationship. Based on three-dimensional reconstruction of pharyngeal structure using limited factor - techniques and the optimization procedure by means of inverse dynamic approach. the biomechanical model of the human pharynx is implemented. The results may be used as clinical index illustrating the degree of pharyngeal muscular dysfunction. This study may be used as useful diagnostic model in discovering early deglutitory impediment caused by physiological or pathological pharyngeal dysfunction.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_1
• /
• pp.1149-1161
• /
• 2021

With a trend of the utilization of computer vision for satellite images, cloud detection using deep learning also attracts attention recently. In this study, we conducted a U-Net cloud detection modeling using SPARCS (Spatial Procedures for Automated Removal of Cloud and Shadow) Cloud Dataset with the image data augmentation and carried out 10-fold cross-validation for an objective assessment of the model. Asthe result of the blind test for 1800 datasets with 512 by 512 pixels, relatively high performance with the accuracy of 0.821, the precision of 0.847, the recall of 0.821, the F1-score of 0.831, and the IoU (Intersection over Union) of 0.723. Although 14.5% of actual cloud shadows were misclassified as land, and 19.7% of actual clouds were misidentified as land, this can be overcome by increasing the quality and quantity of label datasets. Moreover, a state-of-the-art DeepLab V3+ model and the NAS (Neural Architecture Search) optimization technique can help the cloud detection for CAS500 (Compact Advanced Satellite 500) in South Korea.

• Korea Science and Art Forum
• /
• v.37 no.2
• /
• pp.101-112
• /
• 2019

Regardless of the size, the role and meaning of the space required for human daily life are the same. Especially, as the spatial design is small and the size is small, the careful design and problem solving are needed to enable comfortable and convenient life even in a narrow space. The purpose of this study is to design a convergence model that utilizes the advantages of BIM, which can simulate actual design based on a new one-person housing space design plan optimized for one person. This study applied the BIM design technology to one-person housing 2D design, and the suitability examination and the space optimization design of the interior design were carried out. As a result of the study, utility of space improvement, consideration of housing environment, interference check, application of eco-friendly housing system, and MEP design item were derived. Therefore, BIM space design in interior space has been confirmed as a way to overcome limit and inefficiency of 2D design which is applied to actual space by various space design elements. Based on the results of this study, the One-person housing space model, which is applied to the study, is a pure creation designed based on various one-person housing of social and cultural peculiarities derived from previous research. This design example was applied to BIM technology to confirm the detailed and practical design possibility.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.2
• /
• pp.404-411
• /
• 2019

In this study, the HEC-HMS was applied to determine rainfall-runoff processes for the Gokgyuchun basin. Several sub-basins have large-scale reservoirs for agricultural needs and they store large amounts of initial runoff. Three infiltration methods were implemented to reflect the effect of initial loss by reservoirs: 'SCS-CN'(Scheme I), 'SCS-CN' with simple surface method(Scheme II), and 'Initial and Constant rate'(Scheme III). Modeling processes include incorporating three different methods for loss due to infiltration, Clark's UH model for transformation, exponential recession model for baseflow, and Muskingum model for channel routing. The parameters were calibrated using an optimization technique with trial and error method. Performance measures, such as NSE, RAR, and PBIAS, were adopted to aid in the calibration processes. The model performance for those methods was evaluated at Gangcheong station, which is the outlet of study site. Good accuracy in predicting runoff volume and peak flow, and peak time was obtained using the Scheme II and III, considering the initial loss, whereas Scheme I showed low reliability for storms. Scheme III did not show good matches between observed and simulated values for storms with multi peaks. Conclusively, Scheme II provided better results for both single and multi-peak storms. The results of this study can provide a useful tool for decision makers to determine master plans for regional flood control management.