• The Journal of the Korea Contents Association
• /
• v.20 no.12
• /
• pp.49-57
• /
• 2020

This study predicts the ratio of added value, which represents the competitiveness of export industries in South Korea, using various machine learning techniques. To enhance the accuracy and stability of prediction, forecast combination technique was applied to predicted values of machine learning techniques. In particular, this study improved the efficiency of the prediction process by selecting key variables out of many variables using recursive feature elimination method and applying them to machine learning techniques. As a result, it was found that the predicted value by the forecast combination method was closer to the actual value than the predicted values of the machine learning techniques. In addition, the forecast combination method showed stable prediction results unlike volatile predicted values by machine learning techniques.

• Journal of the Korea Convergence Society
• /
• v.11 no.11
• /
• pp.257-267
• /
• 2020

In this study, we predicted the employment ratio by the export industry using various machine learning methods and verified whether the prediction performance is improved by applying the constrained forecast combination method to these predicted values. In particular, the constrained forecast combination method is known to improve the prediction accuracy and stability by imposing the sum of predicted values' weights up to one. In addition, this study considered various variables affecting the employment ratio of each industry, and so we adopted recursive feature elimination method that allows efficient use of machine learning methods. As a result, the constrained forecast combination showed more accurate prediction performance than the predicted values of the machine learning methods, and in particular, the stability of the prediction performance of the constrained forecast combination was higher than that of other machine learning methods.

• Journal of the Korea Convergence Society
• /
• v.13 no.5
• /
• pp.237-247
• /
• 2022

This study predicts the economic activity level of the elderly in Korea using various machine learning methods. While the previous studies mainly focused on testing the relationship between the economic activity level and the life satisfaction or the social security system, this study aims at the accurate prediction on the economic activity level of the elderly using various machine learning methods and the forecast combination. Dependent variables such as the activity rate, employment rate, etc and independent variables such as the income, average wage, etc compose the dataset in this study. Five different machine learning methods and two forecast combinations are applied to the given dataset. The prediction performances of the machine learning method and the forecast combination varied across the dependent variables and prediction intervals, but it was found that the forecast combination was relatively superior to other methods in terms of the stability of prediction. This study has significance in that it accurately predicted the economic activity level of the elderly and achieved the stability of the prediction, raising practicality from a policy perspective.

• Journal of KIISE:Software and Applications
• /
• v.36 no.10
• /
• pp.851-860
• /
• 2009

Recently, among the computational methods of protein-protein interaction prediction, vast amounts of domain based methods originated from domain-domain relation consideration have been developed. However, it is true that multi domains collaboration is avowedly ignored because of computational complexity. In this paper, we implemented a protein interaction prediction system based the Interaction Significance matrix, which quantified an influence of domain combination pair on a protein interaction. Unlike conventional domain combination methods, IS matrix contains weighted domain combinations and domain combination pair power, which mean possibilities of domain collaboration and being the main body on a protein interaction. About 63% of sensitivity and 94% of specificity were measured when we use interaction data from DIP, IntAct and Pfam-A as a domain database. In addition, prediction accuracy gradually increased by growth of learning set size, The prediction software and learning data are currently available on the web site.

• Journal of KIISE:Computing Practices and Letters
• /
• v.10 no.4
• /
• pp.299-308
• /
• 2004

In this paper, we propose a probabilistic framework to predict the interaction probability of proteins. The notion of domain combination and domain combination pair is newly introduced and the prediction model in the framework takes domain combination pair as a basic unit of protein interactions to overcome the limitations of the conventional domain pair based prediction systems. The framework largely consists of prediction preparation and service stages. In the prediction preparation stage, two appearance probability matrices, which hold information on appearance frequencies of domain combination pairs in the interacting and non-interacting sets of protein pairs, are constructed. Based on the appearance probability matrix, a probability equation is devised. The equation maps a protein pair to a real number in the range of 0 to 1. Two distributions of interacting and non-interacting set of protein pairs are obtained using the equation. In the prediction service stage, the interaction probability of a Protein pair is predicted using the distributions and the equation. The validity of the prediction model is evaluated for the interacting set of protein pairs in Yeast organism and artificially generated non-interacting set of protein pairs. When 80% of the set of interacting protein pairs in DIP database are used as teaming set of interacting protein pairs, very high sensitivity(86%) and specificity(56%) are achieved within our framework.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.10 no.3 s.49
• /
• pp.65-75
• /
• 2006

Nonlinear pushover analysis is used to evaluate the earthquake response of building structures. To accurately predict the inelastic response of a structure, the prescribed story load profile should be able to describe the earthquake force profile which actually occurs during the time-history response of the structure. In the present study, a new modal combination method was developed to predict the earthquake load profiles of building structures. In the proposed method, multiple story load profiles are predicted by combining the modal spectrum responses multiplied by the modal combination factors. Parametric studies were performed far moment-resisting frames and walls. Based on the results. the modal combination factors were determined according to the hierarchy of each mode affecting the dynamic responses of structures. The proposed modal combination method was applied to prototype buildings with and without vertical irregularity. The results showed that the proposed method predicts the actual story load profiles which occur during the time-history responses of the structures.

• Journal of Korea Water Resources Association
• /
• v.44 no.7
• /
• pp.523-536
• /
• 2011

The objective of this study is to develop the data driven model for the flood forecasting that are improved the problems of the existing hydrological model for flood forecasting in medium and small streams. Neuro-Fuzzy flood forecasting model which linked the Takagi-Sugeno fuzzy inference theory with neural network, that can forecast flood only by using the rainfall and flood level and discharge data without using lots of physical data that are necessary in existing hydrological rainfall-runoff model is established. The accuracy of flood forecasting using this model is determined by temporal distribution and number of used rainfall and water level as input data. So first of all, the various combinations of input data were constructed by using rainfall and water level to select optimal input data combination for applying Neuro-Fuzzy flood forecasting model. The forecasting results of each combination are compared and optimal input data combination for real-time flood forecasting is determined.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.358-358
• /
• 2023

본 연구에서는 그동안 수자원분야 강우유출 해석분야에 활용되었던 대표적인 머신러닝&딥러닝(ML&DL) 모델을 활용하여 모델의 하이퍼파라미터 튜닝뿐만 아니라 모델의 특성을 고려한 기상 및 수문데이터의 조합과 전처리(lag-time, 이동평균 등)를 통하여 데이터 특성과 ML&DL모델의 조합시나리오에 따른 일 유입량 예측성능을 비교 검토하는 연구를 수행하였다. 이를 위해 소양강댐 유역을 대상으로 1974년에서 2021년까지 축적된 기상 및 수문데이터를 활용하여 1) 강우, 2) 유입량, 3) 기상자료를 주요 영향변수(독립변수)로 고려하고, 이에 a) 지체시간(lag-time), b) 이동평균, c) 유입량의 성분분리조건을 적용하여 총 36가지 시나리오 조합을 ML&DL의 입력자료로 활용하였다. ML&DL 모델은 1) Linear Regression(LR), 2) Lasso, 3) Ridge, 4) SVR(Support Vector Regression), 5) Random Forest(RF), 6) LGBM(Light Gradient Boosting Model), 7) XGBoost의 7가지 ML방법과 8) LSTM(Long Short-Term Memory models), 9) TCN(Temporal Convolutional Network), 10) LSTM-TCN의 3가지 DL 방법, 총 10가지 ML&DL모델을 비교 검토하여 일유입량 예측을 위한 가장 적합한 데이터 조합 특성과 ML&DL모델을 성능평가와 함께 제시하였다. 학습된 모형의 유입량 예측 결과를 비교·분석한 결과, 소양강댐 유역에서는 딥러닝 중에서는 TCN모형이 가장 우수한 성능을 보였고(TCN>TCN-LSTM>LSTM), 트리기반 머신러닝중에서는 Random Forest와 LGBM이 우수한 성능을 보였으며(RF, LGBM>XGB), SVR도 LGBM수준의 우수한 성능을 나타내었다. LR, Lasso, Ridge 세가지 Regression모형은 상대적으로 낮은 성능을 보였다. 또한 소양강댐 댐유입량 예측에 대하여 강우, 유입량, 기상계열을 36가지로 조합한 결과, 입력자료에 lag-time이 적용된 강우계열의 조합 분석에서 세가지 Regression모델을 제외한 모든 모형에서 NSE(Nash-Sutcliffe Efficiency) 0.8이상(최대 0.867)의 성능을 보였으며, lag-time이 적용된 강우와 유입량계열을 조합했을 경우 NSE 0.85이상(최대 0.901)의 더 우수한 성능을 보였다.

• Fire Science and Engineering
• /
• v.28 no.4
• /
• pp.35-43
• /
• 2014

The prediction performance of 9 model sets, which combine 3 turbulent models and 3 combustion models, was investigated numerically for turbulent partially-premixed jet flame. The standard ${\kappa}-{\varepsilon}$ (SKE), Realizable ${\kappa}-{\varepsilon}$ (RKE) and Reynolds stress model (RSM) were used as a turbulence model, and the eddy dissipation concept (EDC), steady laminar flamelet (SLF) and unsteady laminar flamelet model (ULF) were also adopted as a combustion model. The prediction performance of those 9 model sets was evaluated quantitatively and qualitatively for Sandia D flame of which flame structure was measured precisely. The flame length was predicted as, from longest to shortest, RSM > SKE > RKE, and the RKE predicted the flame length of the jet flame much shorter than experiment. The flame temperature was over predicted by the combination of RSM + SLF or RSM + ULF while the flame length obtained by RSM + SLF and RSM + ULF was well agreed with the experiment. The combination of SKE + SLF and SKE + ULF predicts well the flame length as well as the temperature distribution. The SKE turbulence model was most superior to the other turbulent models, and SKE + ULF showed the best prediction performance for the structure of turbulent partially-premixed jet flame.

• Proceedings of the Korean Information Science Society Conference
• /
• 2007.06b
• /
• pp.476-481
• /
• 2007

큰 소수를 빠르게 생성하기 위한 다양한 소수 검사 방법이 개발되었으며, 가장 많이 쓰이는 소수 검사 방법은 trial division과 Fermat (또는 Miller-Rabin) 검사를 조합한 방법과 gcd 연산과 Fermat (또는 Miller-Rabin) 검사를 조합한 방법이다. 이 중 trial division과 조합한 방법에 대해서는 확률적 분석을 이용하여 수행시간을 예측하고 수행시간을 최적화 하는 방법이 개발되었다. 하지만, gcd 연산과 조합한 방법에 대해서는 아무런 연구결과도 제시되어 있지 않다. 본 논문에서는 gcd 연산을 이용한 조합 소수 검사 방법에 대해 확률적 분석을 이용하여 수행시간을 예측하고 수행시간을 최적화 하는 방법을 제안한다.