• 한국지능정보시스템학회:학술대회논문집
• /
• 한국지능정보시스템학회 2005년도 공동추계학술대회
• /
• pp.571-578
• /
• 2005

Corporate credit rating analysis has drawn a lot of research interests in previous studies, and recent studies have shown that machine learning techniques achieved better performance than traditional statistical ones. This paper applies support vector machines (SVMs) to the corporate credit rating problem in an attempt to suggest a new model with better explanatory power and stability. To serve this purpose, the researcher uses a grid-search technique using 5-fold cross-validation to find out the optimal parameter values of kernel function of SVM. In addition, to evaluate the prediction accuracy of SVM, the researcher compares its performance with those of multiple discriminant analysis (MDA), case-based reasoning (CBR), and three-layer fully connected back-propagation neural networks (BPNs). The experiment results show that SVM outperforms the other methods.

• 자원리싸이클링
• /
• 제26권2호
• /
• pp.46-55
• /
• 2017

본 연구에서는 레이저유도붕괴분광(Laser Induced Breakdown Spectroscopy, LIBS)을 이용하여 방사형 기저함수 신경회로망(Radial Basis Function Neural Networks, RBFNNs) 분류기 설계방법론을 개발하고 실제 폐소형가전제품의 플라스틱 분류 시스템에 적용하였다. ABS, PP, PS와 같은 검정색 플라스틱을 구별하기 위해, 지능형 알고리즘 중 하나인 방사형 기저함수 신경회로망 분류기를 설계하였다. 획득한 입력변수는 주성분 분석법(Principal Component Analysis, PCA)을 이용하여 축소시켰으며, 군집화기법 중 하나인 K-means 클러스터링 방법을 이용해 여러 그룹으로 분할하였다. 전체 데이터는 학습 데이터와 테스트 데이터를 4:1의 비율로 나누었으며, 제안된 분류기의 성능 및 신뢰도를 평가하기 위하여 5-FCV(5-Fold Cross Validation) 기법을 사용하였다. 입력변수와 클러스터의 개수가 각각 5개인 경우, 제안된 분류기의 분류 성능은 96.78%로 나타났다. 또한, 제안된 분류기는 다른 분류기들과 비교하였을 경우 분류 성능의 관점에서 우수성을 보여주었다.

• 한국지능시스템학회논문지
• /
• 제26권6호
• /
• pp.477-484
• /
• 2016

선풍기, 오디오, 전기밥솥 등의 소형 산업가전제품들은 대부분 ABS, PP, PS 등의 재질로 이루어져 있다. 색깔이 있는 플라스틱은 근적외선(NIR) 분광기에 의해 분류가 가능하지만, 반면에 검은색 플라스틱은 빛을 흡수하는 특성으로 인해 분류하기가 어렵다. 그래서 본 연구에서는 LIBS(Laser Induced Breakdown Spectroscopy) 분광기를 통해 폐소형가전 플라스틱을 선별하는 RBFNNs(Radial Basis Function Neural Networks) 패턴 분류기를 소개한다. 전처리부분에는 차원축소 알고리즘 중 하나인 PCA(Principal Component Analysis)를 사용해 처리 속도를 향상시킬 뿐만 아니라 효과적인 데이터의 특성을 추출한다. 조건부에는 FCM(Fuzzy C-Means) 클러스터링을 사용한다. 결론부에는 다항식의 형태 중 하나인 1차 선형식을 연결가중치로서 사용한다. PSO와 5-fold cross validation은 성능의 신뢰도를 향상시키고, 분류율을 높이는데 사용된다. 제안된 분류기의 성능은 최적화한 것과 최적화하지 않은 것 두 가지의 관점에서 보여준다.

• Journal of Korea Trade
• /
• 제24권7호
• /
• pp.1-28
• /
• 2020

Purpose - This paper empirically investigates the predictors and main determinants of consumers' ratings of mobile applications in the Google Play Store. Using a linear and nonlinear model comparison to identify the function of users' review, in determining application rating across countries, this study estimates the direct effects of users' reviews on the application rating. In addition, extending our modelling into a sentimental analysis, this paper also aims to explore the effects of review polarity and subjectivity on the application rating, followed by an examination of the moderating effect of user reviews on the polarity-rating and subjectivity-rating relationships. Design/methodology - Our empirical model considers nonlinear association as well as linear causality between features and targets. This study employs competing theoretical frameworks - multiple regression, decision-tree and neural network models - to identify the predictors and main determinants of app ratings, using data from the Google Play Store. Using a cross-validation method, our analysis investigates the direct and moderating effects of predictors and main determinants of application ratings in a global app market. Findings - The main findings of this study can be summarized as follows: the number of user's review is positively associated with the ratings of a given app and it positively moderates the polarity-rating relationship. Applying the review polarity measured by a sentimental analysis to the modelling, it was found that the polarity is not significantly associated with the rating. This result best applies to the function of both positive and negative reviews in playing a word-of-mouth role, as well as serving as a channel for communication, leading to product innovation. Originality/value - Applying a proxy measured by binomial figures, previous studies have predominantly focused on positive and negative sentiment in examining the determinants of app ratings, assuming that they are significantly associated. Given the constraints to measurement of sentiment in current research, this paper employs sentimental analysis to measure the real integer for users' polarity and subjectivity. This paper also seeks to compare the suitability of three distinct models - linear regression, decision-tree and neural network models. Although a comparison between methodologies has long been considered important to the empirical approach, it has hitherto been underexplored in studies on the app market.

• 한국기계가공학회지
• /
• 제19권7호
• /
• pp.7-15
• /
• 2020

Recently, the fourth industrial revolution has been presented as a new paradigm and additive manufacturing (AM) has become one of the most important topics. For this reason, process monitoring for each cross-sectional layer of additive metal manufacturing is important. Particularly, deep learning can train a machine to analyze, optimize, and repair defects. In this paper, image classification is proposed by learning images of defects in the metal cross sections using the convolution neural network (CNN) image labeling algorithm. Defects were classified into three categories: crack, porosity, and hole. To overcome a lack-of-data problem, the amount of learning data was augmented using a data augmentation algorithm. This augmentation algorithm can transform an image to 180 images, increasing the learning accuracy. The number of training and validation images was 25,920 (80 %) and 6,480 (20 %), respectively. An optimized case with a combination of fully connected layers, an optimizer, and a loss function, showed that the model accuracy was 99.7 % and had a success rate of 97.8 % for 180 test images. In conclusion, image labeling was successfully performed and it is expected to be applied to automated AM process inspection and repair systems in the future.

• 한국콘텐츠학회논문지
• /
• 제20권9호
• /
• pp.511-519
• /
• 2020

본 연구의 목적은 수정된 노팅엄 감각평가(rNSA)를 한국판으로 번역하고 검증하는 것이었다. 수정된 rNSA를 한국어로 번역하기 위해 번안/역번안 과정을 사용하여 단면 연구가 수행되었고, 한국판 rNSA의 유효성을 조사하기 위하여 평가자간 및 평가자내 신뢰도, 내적일관성 및 동시타당도를 조사하였다. 한국판 rNSA는 평가자간 신뢰도(r=0.92-1.00), 평가자내 신뢰도(r=0.93-1.00)로 높은 신뢰도를 확인하였고, 한국판 Fugl Meyer의 감각항목과의 유의한 상관관계(r=0.96)를 확인하였다. 한국판 rNSA의 내적일치도에 대한 Cronbach 값은 0.73-0.90의 범위이고, 한국판 Fugl Meyer의 감각항목의 Cronbach 값은 0.70-0.88의 범위였다. 이 결과로, 한국판 rNSA의 신뢰도 및 타당도를 확인하였으며, 임상환경에서 뇌졸중 이후 감각기능을 평가하는 것이 가능할 수 있다고 생각한다.

• Imaging Science in Dentistry
• /
• 제54권3호
• /
• pp.257-263
• /
• 2024

Purpose: The purpose of this study was to classify mandibular molar furcation involvement (FI) in periapical radiographs using a deep learning algorithm. Materials and Methods: Full mouth series taken at East Carolina University School of Dental Medicine from 2011-2023 were screened. Diagnostic-quality mandibular premolar and molar periapical radiographs with healthy or FI mandibular molars were included. The radiographs were cropped into individual molar images, annotated as "healthy" or "FI," and divided into training, validation, and testing datasets. The images were preprocessed by PyTorch transformations. ResNet-18, a convolutional neural network model, was refined using the PyTorch deep learning framework for the specific imaging classification task. CrossEntropyLoss and the AdamW optimizer were employed for loss function training and optimizing the learning rate, respectively. The images were loaded by PyTorch DataLoader for efficiency. The performance of ResNet-18 algorithm was evaluated with multiple metrics, including training and validation losses, confusion matrix, accuracy, sensitivity, specificity, the receiver operating characteristic (ROC) curve, and the area under the ROC curve. Results: After adequate training, ResNet-18 classified healthy vs. FI molars in the testing set with an accuracy of 96.47%, indicating its suitability for image classification. Conclusion: The deep learning algorithm developed in this study was shown to be promising for classifying mandibular molar FI. It could serve as a valuable supplemental tool for detecting and managing periodontal diseases.

• 한국정보통신학회논문지
• /
• 제21권9호
• /
• pp.1718-1724
• /
• 2017

최근 IoT기반으로 가전제품을 실시간 모니터링을 하는 스마트 플러그가 활성화 되고 있다. 이를 통해 상시 실시간 에너지 소비 모니터링을 통한 소비자의 에너지 절약 유도를 하고, 소비자 설정 기반의 알람 기능을 통해 소비전력을 절감하는 효과를 보고 있다. 본 논문에서는 이러한 실시간 모니터링을 위해 벽 전원 콘센트에서 나오는 교류 전류를 측정한다. 이때, 가전제품마다의 전류 패턴을 분류하고 어떤 제품이 동작하는지 판단을 위해 딥러닝(Deep learning)으로 실험하였다. 전류 패턴의 학습으로 제품의 종류에 따른 인식 성능을 검증하기 위하여, 교차 검증 방법과 붓스트랩(Bootstrap) 검증 방법을 이용하였다. 또한 Cost function과 학습 성공률(Accuracy)이 Train 데이터와 Test 데이터가 동일함을 확인하였다.

• Interdisciplinary Bio Central
• /
• 제3권4호
• /
• pp.15.1-15.11
• /
• 2011

Background: Histone deacetylase (HDAC) 8 is one of its family members catalyzes the removal of acetyl groups from N-terminal lysine residues of histone proteins thereby restricts transcription factors from being expressed. Inhibition of HDAC8 has become an emerging and effective anti-cancer therapy for various cancers. Application computational methodologies may result in identifying the key components that can be used in developing future potent HDAC8 inhibitors. Results: Facilitating the discovery of novel and potential chemical scaffolds as starting points in the future HDAC8 inhibitor design, quantitative structure-activity relationship models were generated with 30 training set compounds using genetic function approximation (GFA) and Bayesian algorithms. Six GFA models were selected based on the significant statistical parameters calculated during model development. A Bayesian model using fingerprints was developed with a receiver operating characteristic curve cross-validation value of 0.902. An external test set of 54 diverse compounds was used in validating the models. Conclusions: Finally two out of six models based on their predictive ability over the test set compounds were selected as final GFA models. The Bayesian model has displayed a high classifying ability with the same test set compounds and the positively and negatively contributing molecular fingerprints were also unveiled by the model. The effectively contributing physicochemical properties and molecular fingerprints from a set of known HDAC8 inhibitors were identified and can be used in designing future HDAC8 inhibitors.

• 한국전자파학회논문지
• /
• 제29권2호
• /
• pp.118-127
• /
• 2018

본 논문에서는 합성개구레이다(Synthetic Aperture Radar: SAR) 장비의 성능 분석을 위하여 레이다 표적모의장치를 설계, 제작한 결과를 선보인다. 먼저 표적모의장치의 기능과 성능에 대해 설명하고, SAR 장비와 연동하기 위한 표적 시나리오 생성방법에 대해 설명하였다. 또한 제작된 표적모의장치를 이용하여 SAR 장비의 시간지연 값을 간단하고 정확하게 측정하고 보상하는 방법을 개발하였다. 시간지연이 보상된 SAR 장비와 표적모의장치를 이용하여 SAR 영상에서 점표적을 획득하고, IRF(Impulse Response Function) 분석을 통해 성능을 분석하였다. 그 결과, 방위방향 IRF의 PSLR(Peak to Side Lobe Ratio)은 -13.25 dB, 해상도는 0.49 m로 이론값과 매우 유사한 값을 보였다.