• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제9권6호
• /
• pp.187-194
• /
• 2020

본 논문은 클라이언트의 익명성과 개인 정보를 보장하는 토르 네트워크에서 앙상블 학습을 이용한 웹사이트 핑거프린팅 방법을 제안한다. 토르네트워크에서 수집된 트래픽 패킷들로부터 웹사이트 핑거프린팅을 위한 훈련 문제를 구성하며, 트리 기반 앙상블 모델을 적용한 웹사이트 핑거프린팅 시스템의 성능을 비교한다. 훈련 특징 벡터는 트래픽 시퀀스에서 추출된 범용 정보, 버스트, 셀 시퀀스 길이, 그리고 셀 순서로부터 준비하며, 각 웹사이트의 특징은 고정 길이로 표현된다. 실험 평가를 위해 웹사이트 핑거프린팅의 사용에 따른 4가지 학습 문제(Wang14, BW, CW_T, CW_H)를 정의하고, CUMUL 특징 벡터를 사용한 지지 벡터 기계 모델과 성능을 비교한다. 실험 평가에서, BW 경우를 제외하고 제안하는 통계 기반 훈련 특징 표현이 CUMUL 특징 표현보다 우수하다.

• 경영정보학연구
• /
• 제22권4호
• /
• pp.21-39
• /
• 2020

딥러닝(Deep learning) 기법은 패턴분석, 이미지분류 등 다양한 분야에서 높은 성과를 나타내고 있다. 특히, 주식시장 분석문제는 머신러닝 연구분야에서도 어려운 분야이므로 딥러닝이 많이 활용되는 영역이다. 본 연구에서는 패턴분석과 분류능력이 높은 딥러닝의 일종인 합성곱신경망(Convolutional Neural Network) 모델을 활용하여 주가방향 예측방법을 제안한다. 추가적으로 합성곱신경망 모델을 효율적으로 학습시키기 위한 속성선택(Feature Selection, FS)방법이 적용된다. 합성곱신경망 모델의 성과는 머신러닝 단일 분류기와 앙상블 분류기를 벤치마킹하여 객관적으로 검증된다. 본 연구에서 벤치마킹한 분류기는 로지스틱 회귀분석(Logistic Regression), 의사결정나무(Decision Tree), 인공신경망(Neural Network), 서포트 벡터머신(Support Vector Machine), 아다부스트(Adaboost), 배깅(Bagging), 랜덤포레스트(Random Forest)이다. 실증분석 결과, 속성선택을 적용한 합성곱신경망이 다른 벤치마킹 분류기보다 분류 성능이 상대적으로 높게 나타났다. 이러한 결과는 합성곱신경망 모델과 속성선택방법을 적용한 예측방법이 기업의 재무자료에 내포된 가치를 보다 정교하게 분석할 수 있는 가능성이 있음을 실증적으로 확인할 수 있었다.

• 정보관리학회지
• /
• 제23권1호
• /
• pp.279-299
• /
• 2006

학술지 논문에 디스크립터를 자동부여하기 위하여 기계학습 기반의 접근법을 적용하였다. 정보학 분야의 핵심 학술지를 선정하여 지난 11년간 수록된 논문들을 대상으로 문헌집단을 구성하였고, 자질 선정과 학습집합의 크기에 따른 성능을 살펴보았다. 그 결과, 자질 선정에서는 카이제곱 통계량(CHI)과 고빈도 선호 자질 선정 기준들(COS, GSS, JAC)을 사용하여 자질을 축소한 다음, 지지벡터기계(SVM)로 학습한 결과가 가장 좋은 성능을 보였다. 학습집합의 크기에서는 지지벡터기계(SVM)와 투표형 퍼셉트론(VPT)의 경우에는 상당한 영향을 받지만 나이브 베이즈(NB)의 경우에는 거의 영향을 받지 않는 것으로 나타났다.

• 한국멀티미디어학회논문지
• /
• 제3권4호
• /
• pp.416-423
• /
• 2000

기존 Kohonen의 자기조직화 지도(self-organizing feature map)는 학습시 많은 입력 패턴이 필요하며 이에 따른 학습 시간 역시 증가하는 단점이 있다. 이러한 단점을 보완하기 위해 B. Bavarian은 위상학적 위치에 따라 각기 다른 학습률(learning rate)을 갖도록 하였으나 자기조직화가 정밀하게 되지 않는 단점을 갖고 있다. 본 논문에서는 자기조직화 지도의 학습시 계산량이 많은 가우시안 함수를 근사곡선(approximate curve)으로 변형하여 수렴속도를 향상시켰고 학습 횟수에 따라 근사곡선의 폭을 동적으로 변화시킴으로써 자기조직화지도의 수렴도를 개선하였다.

• Nuclear Engineering and Technology
• /
• 제56권6호
• /
• pp.2266-2273
• /
• 2024

A machine learning platform is proposed for the diagnosis of a severe accident progression in a nuclear power plant. To predict the key parameters for accident management including lost signals, a long short term memory (LSTM) network is proposed, where multiple accident scenarios are used for training. Training and test data were produced by MELCOR simulation of the Fukushima Daiichi Nuclear Power Plant (FDNPP) accident at unit 3. Feature variables were selected among plant parameters, where the importance ranking was determined by a recursive feature elimination technique using RandomForestRegressor. To answer the question of whether a reduced order ML model could predict the complex transient response, we performed a systematic sensitivity study for the choices of target variables, the combination of training and test data, the number of feature variables, and the number of neurons to evaluate the performance of the proposed ML platform. The number of sensitivity cases was chosen to guarantee a 95 % tolerance limit with a 95 % confidence level based on Wilks' formula to quantify the uncertainty of predictions. The results of investigations indicate that the proposed ML platform consistently predicts the target variable. The median and mean predictions were close to the true value.

• International Journal of Computer Science & Network Security
• /
• 제21권1호
• /
• pp.220-225
• /
• 2021

Human population growth rate is an important parameter for real-world planning. Common approaches rely upon fixed parameters like human population, mortality rate, fertility rate, which is collected historically to determine the region's population growth rate. Literature does not provide a solution for areas with no historical knowledge. In such areas, machine learning can solve the problem, but a multitude of machine learning algorithm makes it difficult to determine the best approach. Further, the missing feature is a common real-world problem. Thus, it is essential to compare and select the machine learning techniques which provide the best and most robust in the presence of missing features. This study compares 17 machine learning techniques (base learners and ensemble learners) performance in predicting the human population growth rate of the country. Among the 17 machine learning techniques, random forest outperformed all the other techniques both in predictive performance and robustness towards missing features. Thus, the study successfully demonstrates and compares machine learning techniques to predict the human population growth rate in settings where historical data and feature information is not available. Further, the study provides the best machine learning algorithm for performing population growth rate prediction.

• 한국HCI학회:학술대회논문집
• /
• 한국HCI학회 2008년도 학술대회 1부
• /
• pp.504-509
• /
• 2008

Support Vector Machine (SVM) is one of powerful learning machine and has been applied to varying task with generally acceptable performance. The success of SVM for classification tasks in one domain is affected by features which represent the instance of specific class. Given the representative and discriminative features, SVM learning will give good generalization and consequently we can obtain good classifier. In this paper, we will assess the problem of feature choices for human detection tasks and measure the performance of each feature. Here we will consider HOG-family feature. As a natural extension of SVM, we combine SVM with Principal Component Analysis (PCA) to reduce dimension of features while retaining most of discriminative feature vectors.

• Journal of Computing Science and Engineering
• /
• 제10권4호
• /
• pp.103-110
• /
• 2016

Emotion recognition is an important component of affective computing, and is significant in the implementation of natural and friendly human-computer interaction. An effective approach to recognizing emotion from text is based on a machine learning technique, which deals with emotion recognition as a classification problem. However, in emotion recognition, the texts involved are usually very short, leaving a very large, sparse feature space, which decreases the performance of emotion classification. This paper proposes to resolve the problem of feature sparseness, and largely improve the emotion recognition performance from short texts by doing the following: representing short texts with word cluster features, offering a novel word clustering algorithm, and using a new feature weighting scheme. Emotion classification experiments were performed with different features and weighting schemes on a publicly available dataset. The experimental results suggest that the word cluster features and the proposed weighting scheme can partly resolve problems with feature sparseness and emotion recognition performance.

• Journal of information and communication convergence engineering
• /
• 제20권1호
• /
• pp.31-40
• /
• 2022

Coronary heart disease (CHD) is a comorbidity of COVID-19; therefore, routine early diagnosis is crucial. A large number of examination attributes in the context of diagnosing CHD is a distinct obstacle during the pandemic when the number of health service users is significant. The development of a precise machine learning model for diagnosis with a minimum number of examination attributes can allow examinations and healthcare actions to be undertaken quickly. This study proposes a CHD diagnosis model based on feature selection, data balancing, and ensemble-based classification methods. In the feature selection stage, a hybrid SVM-GA combined with fast correlation-based filter (FCBF) is used. The proposed system achieved an accuracy of 94.60% and area under the curve (AUC) of 97.5% when tested on the z-Alizadeh Sani dataset and used only 8 of 54 inspection attributes. In terms of performance, the proposed model can be placed in the very good category.

• 전력전자학회논문지
• /
• 제27권4호
• /
• pp.332-338
• /
• 2022

This study proposes a battery state-of-health estimation method by applying a feature extraction technique. The technique that can improve estimation performance is the process of identifying and extracting meaningful data. To apply a data-driven-based aging state estimation method to batteries, health indicators are used as training data. However, limitations occur in extracting health indicators from charge/discharge cycles. This study proposes a deep-learning-based battery state-of-health estimation method that applies feature extraction techniques to compensate for this problem. According to the performance evaluation result of the proposed method, it has a low estimation error of 0.3887% based on an absolute error evaluation method.