• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.346-346
• /
• 2023

한국은 기상·수문정보의 예측이 기상 및 기후 측면에서 주도적으로 이루어지고 있다. 그러나 단기 및 중기 수자원 평가 및 분석을 위해 필요한 시공간적 규모, 정확도, 평가체계를 고려한 기상 기후 예측정보의 활용 방안이 마련될 필요가 있다. 이에 본 연구에서는 미래 수자원 평가 및 분석을 위한 방안을 마련하고자 국내 경안천 유역을 대상으로 하천유량을 예측하고 평가하였다. 이를 위해, 우리는 세계기상기구(World Meteorological Organization, WMO)에서 회원국을 대상으로 배포 중인 수자원 평가 도구인 동적수자원평가시스템(Dynamic Water resources Assessment Tool, DWAT)을 경안천 유역에 대하여 구축하고, 과거 관측 기상 및 유량 자료를 이용하여 매개변수를 보정하였다. 앙상블 하천유량 예측을 위해서 전지구적인 기후 패턴과 국내 기상 특성 간의 상관성 분석 후 이를 예측인자로 활용하여 다중회귀모형과 인공신경망 모형으로부터 생성된 1,000개의 앙상블 강우 및 기온 예측정보를 DWAT의 입력자료로 이용하였다. 2022년에 대한 앙상블예측정보를 DWAT의 입력자료로 사용하여 앙상블 하천유량이 예측되었다. 예측된 일-단위 하천유량은 실제 관측유량과 차이를 보이나 이는 예측된 앙상블 강우 및 기온정보의 오차에 기인하는 것으로 보인다. 이러한 결과는 수문 모형 결과의 오차는 강제 자료의 오차에 큰 영향을 받는 한계를 다시 한번 확인시켜준다. 따라서 단기·중기 수자원 평가 및 분석을 월-단위 하천유량으로 변환하여 월별 통계치를 분석하는 방향을 고려할 필요가 있다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.6
• /
• pp.1213-1220
• /
• 2011

In this paper, a new ensemble system using dynamic weighting method with added weight information into classifiers is proposed. The weights used in the traditional ensemble system are those after the training phase. Once extracted, the weights in the traditional ensemble system remain fixed regardless of the test data set. One way to circumvent this problem in the gating networks is to update the weights dynamically by adding processes making architectural hierarchies, but it has the drawback of added processes. A simple method to update weights dynamically, without added processes, is proposed, which can be applied to the already established ensemble system without much of the architectural modification. Experiment shows that this method performs better than AdaBoost.

• Journal of KIISE
• /
• v.44 no.8
• /
• pp.842-853
• /
• 2017

Streaming data is a sequence of data samples that are consistently generated over time. The data distribution or concept can change over time, and this change becomes a factor to reduce the performance of a classification model. Adaptive incremental learning can maintain the classification performance by updating the current classification model with the weight adjusted according to the degree of concept drift. However, selecting the proper weight value depending on the degree of concept drift is difficult. In this paper, we propose a dynamic ensemble method based on adaptive weight adjustment according to the degree of concept drift. Experimental results demonstrate that the proposed method shows higher performance than the other compared methods.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.546-551
• /
• 2023

일반적으로 딥러닝 모델은 모든 라벨에 데이터 수가 균형을 이룰 때 가장 좋은 성능을 보인다. 그러나 현실에서는 특정라벨에 대한 데이터가 부족한 경우가 많으며 이로 인해 불균형 데이터 문제가 발생한다. 이에 대한 해결책으로 오버샘플링과 가중치 손실과 같은 데이터 불균형 기법이 연구되었지만 이러한 기법들은 데이터가 적은 라벨의 성능을 개선하는 동시에 데이터가 많은 라벨의 성능을 저하시키는 부작용을 가지고 있다. 본 논문에서는 이 문제를 완화시키고자 어텐션 기반의 앙상블 기법을 제안한다. 어텐션 기반의 앙상블은 데이터 불균형 기법을 적용한 모델과 적용하지 않은 모델의 출력 값을 가중 평균하여 최종 예측을 수행한다. 이때 가중치는 어텐션 메커니즘을 통해 동적으로 조절된다. 그로므로 어텐션 기반의 앙상블 모델은 입력 데이터 특성에 따라 가중치를 조절할 수가 있다. 실험은 에세이 자동 평가 데이터를 대상으로 수행하였다. 실험 결과로는 제안한 모델이 데이터 불균형 기법의 부작용을 완화하고 성능이 개선되었다.

• The KIPS Transactions:PartB
• /
• v.11B no.4
• /
• pp.465-470
• /
• 2004

Web Information Predictive Systems have the restriction such as they need users profiles and visible feedback information for obtaining the necessary information. For overcoming this restrict, this study designed and implemented Dynamic Web Information Predictive System using Ensemble Support Vector Machine to be able to predict the web information and provide the relevant information every user needs most by click stream data and user feedback information, which have some clues based on the data. The result of performance test using Dynamic Web Information Predictive System using Ensemble Support Vector Machine against the existing Web Information Predictive System has preyed that this study s method is an excellence solution.

• Journal of Intelligence and Information Systems
• /
• v.16 no.1
• /
• pp.93-116
• /
• 2010

Ensemble approach is applied to the detection modeling of illegal cash accommodation (ICA) that is the well-known type of fraudulent usages of credit cards in far east nations and has not been addressed in the academic literatures. The performance of fraud detection model (FDM) suffers from the imbalanced data problem, which can be remedied to some extent using an ensemble of many classifiers. It is generally accepted that ensembles of classifiers produce better accuracy than a single classifier provided there is diversity in the ensemble. Furthermore, recent researches reveal that it may be better to ensemble some selected classifiers instead of all of the classifiers at hand. For the effective detection of ICA, we adopt ensemble size reduction technique that prunes the ensemble of all classifiers using accuracy and diversity measures. The diversity in ensemble manifests itself as disagreement or ambiguity among members. Data imbalance intrinsic to FDM affects our approach for ICA detection in two ways. First, we suggest the training procedure with over-sampling methods to obtain diverse training data sets. Second, we use some variants of accuracy and diversity measures that focus on fraud class. We also dynamically calculate the diversity measure-Forward Addition and Backward Elimination. In our experiments, Neural Networks, Decision Trees and Logit Regressions are the base models as the ensemble members and the performance of homogeneous ensembles are compared with that of heterogeneous ensembles. The experimental results show that the reduced size ensemble is as accurate on average over the data-sets tested as the non-pruned version, which provides benefits in terms of its application efficiency and reduced complexity of the ensemble.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2004.05b
• /
• pp.272-276
• /
• 2004

이수기와 같이 장기적인 관점에서 저수지운영을 해야 하는 관리자는 해당 기간동안의 이익을 최대화하는 전략을 필요로 한다. 이를 위해서는 미래 유입량의 불확실성을 고려한 최적화 모형에 근거한 운영률을 수립해야 할 것이다. 본 연구에서는 금강수계의 이수기를 대상으로 추계학적 최적화 기법인 표본 추계학적 동적계획법(Sampling Stochastic Dynamic Programming)을 적용하여 최적 연계운영 시스템을 개발하였다. 본 연구를 통해 개발된 모형은 상용프로그램인 CSUDP와의 비교를 통해 검증되었으며 이를 기반으로 과거자료를 이용한 SSDP/Hist모형과 앙상블 유량예측(Ensemble Streamflow Prediction)을 이용한 SSDP/ESP모형을 개발하여 두 모형의 장${\cdot}$단점을 비교 분석하였다. 발전부분은 두 모형이 비슷하였으나 용수공급 측면에서는 SSDP/ESP가 SSDP/Hist 보다 우수함을 알 수 있었다.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.4
• /
• pp.221-228
• /
• 2024

Predicting insurance claims is a key task for insurance companies to manage risks and maintain financial stability. Accurate insurance claim predictions enable insurers to set appropriate premiums, reduce unexpected losses, and improve the quality of customer service. This study aims to enhance the performance of insurance claim prediction models by applying ensemble learning techniques. The predictive performance of models such as Random Forest, Gradient Boosting Machine (GBM), XGBoost, Stacking, and the proposed Dynamic Weighted Ensemble (DWE) model were compared and analyzed. Model performance was evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), and the Coefficient of Determination (R²). Experimental results showed that the DWE model outperformed others in terms of evaluation metrics, achieving optimal predictive performance by combining the prediction results of Random Forest, XGBoost, LR, and LightGBM. This study demonstrates that ensemble learning techniques are effective in improving the accuracy of insurance claim predictions and suggests the potential utilization of AI-based predictive models in the insurance industry.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.298-301
• /
• 2024

클라우드 환경에서 컨테이너 사용이 증가하면서 컨테이너 환경을 대상으로 하는 여러 보안 위협이 증가하고 있다. 대표적인 악성 컨테이너는 크립토재킹 컨테이너로, 인스턴스 소유자의 승인 없이 리소스를 탈취하여 암호화폐를 채굴하는 공격이다. 이러한 공격은 리소스 낭비를 초래할 뿐 아니라 자원을 공유하는 정상 컨테이너나 호스트 인프라에까지도 영향을 미칠 수 있다. 따라서 본 논문에서는 크립토재킹 컨테이너를 탐지하기 위한 앙상블 기반의 크립토재킹 컨테이너 탐지 프레임워크 설계를 제안한다. 또한, 앙상블 모델 학습을 위한 데이터 수집에 있어 크립토재킹 컨테이너의 동적 특징을 나타내는 시스템 콜 및 네트워크 플로우 기반의 특성 활용 가능성을 사례 연구를 통해 분석하였다.

• The Korean Journal of Nuclear Medicine
• /
• v.38 no.6
• /
• pp.486-491
• /
• 2004

Purpose: factor analysis and independent component analysis (ICA) has been used for handling dynamic image sequences. Theoretical advantages of a newly suggested ICA method, ensemble ICA, leaded us to consider applying this method to the analysis of dynamic myocardial $H_2^{15}O$ PET data. In this study, we quantified patients' blood flow using the ensemble ICA method. Materials and Methods: Twenty subjects underwent $H_2^{15}O$ PET scans using ECAT EXACT 47 scanner and myocardial perfusion SPECT using Vertex scanner. After transmission scanning, dynamic emission scans were initiated simultaneously with the injection of $555{\sim}740$ MBq $H_2^{15}O$. Hidden independent components can be extracted from the observed mixed data (PET image) by means of ICA algorithms. Ensemble learning is a variational Bayesian method that provides an analytical approximation to the parameter posterior using a tractable distribution. Variational approximation forms a lower bound on the ensemble likelihood and the maximization of the lower bound is achieved through minimizing the Kullback-Leibler divergence between the true posterior and the variational posterior. In this study, posterior pdf was approximated by a rectified Gaussian distribution to incorporate non-negativity constraint, which is suitable to dynamic images in nuclear medicine. Blood flow was measured in 9 regions - apex, four areas in mid wall, and four areas in base wall. Myocardial perfusion SPECT score and angiography results were compared with the regional blood flow. Results: Major cardiac components were separated successfully by the ensemble ICA method and blood flow could be estimated in 15 among 20 patients. Mean myocardial blood flow was $1.2{\pm}0.40$ ml/min/g in rest, $1.85{\pm}1.12$ ml/min/g in stress state. Blood flow values obtained by an operator in two different occasion were highly correlated (r=0.99). In myocardium component image, the image contrast between left ventricle and myocardium was 1:2.7 in average. Perfusion reserve was significantly different between the regions with and without stenosis detected by the coronary angiography (P<0.01). In 66 segment with stenosis confirmed by angiography, the segments with reversible perfusion decrease in perfusion SPECT showed lower perfusion reserve values in $H_2^{15}O$ PET. Conclusions: Myocardial blood flow could be estimated using an ICA method with ensemble learning. We suggest that the ensemble ICA incorporating non-negative constraint is a feasible method to handle dynamic image sequence obtained by the nuclear medicine techniques.