• 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.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.12
• /
• pp.928-932
• /
• 2009

Most diseases are caused by complex processes of various factors. Although previous researches have tried to identify the causes of the disease, there are still lots of limitations to clarify the complex factors. Here, we present a disease classification model based on an evolutionary learning approach of combinatorial features using the data sets from the genetics and cohort studies. We implemented a system for finding the combinatorial risk factors and visualizing the results. Our results show that the proposed method not only improves classification accuracy but also identifies biologically meaningful sets of risk factors.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.6
• /
• pp.1550-1556
• /
• 1997

In this paper, we apply the evolutionary algorithm to the combinatorial optimization problem. Evolutionary algorithm useful for the optimization of the large space problem. This paper propose a method for the reuse of wastes of light water in atomic reactor system. These wastes contain several reusable elements, and they should be carefully selected and blended to satisfy requirements as an input material to the heavy water atomic reactor system. This problem belongs to an NP-hard like the 0/1 knapsack problem. Two evolutionary strategies are used as approximation algorithms in the highly constrained combinatorial optimization problem. One is the traditional strategy, using random operator with evaluation function, and the other is heuristic based search that uses the vector operator reducing between goal and current status. We also show the method which perform the feasible test and solution evaluation by using the vectored knowledge in problem domain. Finally, We compare the simulation results of using random operator and vector operator for such combinatorial optimization problems.

• 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 Intelligence and Information Systems
• /
• v.3 no.2
• /
• pp.1-9
• /
• 1997

These papers propose an evolutionary algorithm for re-using output of waste fuel of light water reactor system in nuclear power plants. Evolutionary algorithm is useful for optimization of the large space problem. The wastes contain several re-useable elements, and they should be carefully selected and blended to satisfy requirements as input material to the heavy water nuclear reactor system. This problem belongs to a NP-hard like the 0/1 Knapsack problem. Two evolutionary strategies are used as a, pp.oximation algorithms in the highly constrained combinatorial optimization problem. One is the traditional strategy, using random operator with evaluation function, and the other is heuristic based search that uses the vector operator reducing between goal and current status. We also show the method, which performs the feasible teat and solution evaluation by using the vectorized data in problem. Finally, We compare the simulation results of using random operator and vector operator for such combinatorial optimization problems.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.10a
• /
• pp.463-466
• /
• 2016

Sharing of product and process information with partners is a basic activity and key requirement which ensures success of distribution. ePedigree that encapsulates all of the event data from manufacturer to retail shop provides a flexible mechanism of storing and sharing traceable information to the partners of supply chain and credibility of shared information through digital signature based on Public Key Infrastructure (PKI). To generate the signature that can be authenticated through PKI, the partners of supply chain should pay for PKI certificates from Certificate Authority (CA). In case of agrifood cooperatives which consist of petty merchants or farmers, it is hard to pay for the PKI certificate for all members and is a big obstacle for the ePedigree to be applied to the supply chain. This paper proposes a private certification method of ePedigree for cooperatives. Cooperatives can apply the ePedigree using the proposed method to all the members at small cost and the proposed method can assure the credibility of information at the same level of the previous ePedigree.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.7
• /
• pp.1749-1756
• /
• 2014

Knowledge about protein subcellular localization provides important information about protein function. This paper improves a label power-set multi-label classification for the accurate prediction of subcellular localization of proteins which simultaneously exist at multiple subcellular locations. Among multi-label classification methods, label power-set method can effectively model the correlation between subcellular locations of proteins performing certain biological function. With constrained optimization, this paper calculates combination weights which are used in the linear combination representation of a multi-label by other multi-labels. Using these weights, the prediction probabilities of multi-labels are combined to give final prediction results. Experimental results on human protein dataset show that the proposed method achieves higher performance than other prediction methods for protein subcellular localization. This shows that the proposed method can successfully enrich the prediction probability of multi-labels by exploiting the overlapping information between multi-labels.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2013.05a
• /
• pp.675-676
• /
• 2013

This paper presents a design method for multiple-output combinational digital logic systems using time domain based on multiplexing and common multi-terminal extension decision diagrams. The common multi-terminal extension decision diagrams represents extension valued multiple-output functions, while time domain based on multiplexing systems transmit several signals on a single lines. The proposed method can reduce the 1)hardware, 2)logic levels and 3)pins. In the logic system design, we use two types of decision diagrams, that is the common binary decision diagrams and common multi-terminal extension decision diagrams.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.11b
• /
• pp.847-849
• /
• 2005

지문의 특이점은 인식이나 분류를 수행할 때 중요한 특징이다. 특이점 추출은 영상의 품질에 민감하기 때문에, 정확한 추출을 위해 영상의 품질을 향상시키는 것이 필요하다. 다양한 이미지 필터를 적절히 적용하면 개별필터를 적용하는 것에 비해 더 높은 품질을 얻을 수 있지만, 영상 필터가 다양할수록 성능이 좋은 필터 조합을 찾는 것이 어렵다. 본 논문에서는 유전자 알고리즘을 이용하여 특이점 추출율이 좋은 필터 조합을 찾는 방법을 제안한다. 진화에 의해 성능이 좋은 필터 조합을 자동으로 찾아주며, 전문가의 지식이 필요하지 않기 때문에 영상 개선의 여러 분야에 적용될 수 있다. 제안하는 방법을 NIST DB 4에 적용한 결과 기존의 영상 개선 방법보다 높은 성능을 획득하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.10a
• /
• pp.165-166
• /
• 2015

Recently, IoT is a hot issue in the IT research and industry area. Wireless Sensor plays a key roll in the in the IoT. Usually the device of the wireless sensor has tinyOS and small battery power. Hence there are a lot of previous research that want to extend the life time. In this paper, we will introduce the scheduling method based on the combinatorial theory.