• Journal of information and communication convergence engineering
• /
• v.3 no.4
• /
• pp.209-212
• /
• 2005

This study proposes hybrid case-based reasoning and genetic algorithms model for customer classification. In this study, vertical and horizontal dimensions of the research data are reduced through integrated feature and instance selection process using genetic algorithms. We applied the proposed model to customer classification model which utilizes customers' demographic characteristics as inputs to predict their buying behavior for the specific product. Experimental results show that the proposed model may improve the classification accuracy and outperform various optimization models of typical CBR system.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.9 no.3
• /
• pp.67-72
• /
• 2010

This paper presents biometrical driver verification system in car experiment through analysis of speech, and face information. We have used Mel-scale Frequency Cesptral Coefficients (MFCCs) for speaker verification using speech information. For face verification, face region is detected by AdaBoost algorithm and dimension-reduced feature vector is extracted by using principal component analysis only from face region. In this paper, we apply the extracted speech- and face feature vectors to an SVM kernel with Gaussian Mixture Models(GMM) supervector. The experimental results of the proposed approach show a clear improvement compared to a simple GMM or SVM approach.

• International Journal of Computer Science & Network Security
• /
• v.21 no.8
• /
• pp.79-86
• /
• 2021

Alzheimer's is a chronic progressive disease which exhibits varied symptoms and behavioural traits from person to person. The deterioration in cognitive abilities is more noticeable through their Activities and Instrumental Activities of Daily Living rather than biological markers. This information discussed in social media communities was collected and features were extracted by using the proposed fuzzy logic based algorithm to address the uncertainties and imprecision in the data reported. The data thus obtained is used to train machine learning models in order to predict the longevity of the patients. Models built on features extracted using the proposed algorithm performs better than models trained on full set of features. Important findings are discussed and Support Vector Regressor with RBF kernel is identified as the best performing model in predicting the longevity of Alzheimer's patients. The results would prove to be of high value for healthcare practitioners and palliative care providers to design interventions that can alleviate the trauma faced by patients and caregivers due to chronic diseases.

• Asia pacific journal of information systems
• /
• v.29 no.1
• /
• pp.35-49
• /
• 2019

The paper attempts to document the application of relevant Machine Learning (ML) models on Yelp (a crowd-sourced local business review and social networking site) dataset to analyze, predict and recommend business. Strategically using two cloud platforms to minimize the effort and time required for this project. Seven machine learning algorithms in Azure ML of which four algorithms are implemented in Databricks Spark ML. The analyzed Yelp business dataset contained 70 business attributes for more than 350,000 registered business. Additionally, review tips and likes from 500,000 users have been processed for the project. A Recommendation Model is built to provide Yelp users with recommendations for business categories based on their previous business ratings, as well as the business ratings of other users. Classification Model is implemented to predict the popularity of the business as defining the popular business to have stars greater than 3 and unpopular business to have stars less than 3. Text Analysis model is developed by comparing two algorithms, uni-gram feature extraction and n-feature extraction in Azure ML studio and logistic regression model in Spark. Comparative conclusions have been made related to efficiency of Spark ML and Azure ML for these models.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2015.08a
• /
• pp.61.1-61.1
• /
• 2015

Nanoscale semiconductor plasma processing has become one of the most challenging issues due to the limits of physicochemical fabrication routes with its inherent complexity. The mission of future and emerging plasma processing for development of next generation semiconductor processing is to achieve the ideal nanostructures without abnormal profiles and damages, such as 3D NAND cell array with ultra-high aspect ratio, cylinder capacitors, shallow trench isolation, and 3D logic devices. In spite of significant contributions of research frontiers, these processes are still unveiled due to their inherent complexity of physicochemical behaviors, and gaps in academic research prevent their predictable simulation. To overcome these issues, a Korean plasma consortium began in 2009 with the principal aim to develop a realistic and ultrafast 3D topography simulator of semiconductor plasma processing coupled with zero-D bulk plasma models. In this work, aspects of this computational tool are introduced. The simulator was composed of a multiple 3D level-set based moving algorithm, zero-D bulk plasma module including pulsed plasma processing, a 3D ballistic transport module, and a surface reaction module. The main rate coefficients in bulk and surface reaction models were extracted by molecular simulations or fitting experimental data from several diagnostic tools in an inductively coupled fluorocarbon plasma system. Furthermore, it is well known that realistic ballistic transport is a simulation bottleneck due to the brute-force computation required. In this work, effective parallel computing using graphics processing units was applied to improve the computational performance drastically, so that computer-aided design of these processes is possible due to drastically reduced computational time. Finally, it is demonstrated that 3D feature profile simulations coupled with bulk plasma models can lead to better understanding of abnormal behaviors, such as necking, bowing, etch stops and twisting during high aspect ratio contact hole etch.

• The KIPS Transactions:PartA
• /
• v.17A no.6
• /
• pp.281-288
• /
• 2010

In this paper, we describe the development of a system for generating a 3-dimensional human face using 3D scanned facial data and photo images, and morphing animation. The system comprises a facial feature input tool, a 3-dimensional texture mapping interface, and a 3-dimensional facial morphing interface. The facial feature input tool supports texture mapping and morphing animation - facial morphing areas between two facial models are defined by inputting facial feature points interactively. The texture mapping is done first by means of three photo images - a front and two side images - of a face model. The morphing interface allows for the generation of a morphing animation between corresponding areas of two facial models after texture mapping. This system allows users to interactively generate morphing animations between two facial models, without programming, using 3D scanned facial data and photo images.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.37 no.10
• /
• pp.1305-1313
• /
• 2013

In this study, a method to apply feature-based simplification to boundary representation models is proposed. For feature-based simplification, a volume decomposition tree is created from a boundary representation model. The volume decomposition tree is represented by regularized Boolean operations of additive volumes, subtractive volumes, and fillet/round/chamfer volumes, and it is generated by stepwise volume decomposition, which consists of fillet/round/chamfer decomposition, wrap-around decomposition, volume split decomposition, and cell-based decomposition. After the volume decomposition tree is transformed to an infix expression, the CAD model can be simplified by reordering the volumes. To verify the proposed method, a prototype system was implemented, and experiments on test cases were conducted. From the results of the experiments, it is verified that the proposed method is useful for simplifying CAD models based on boundary representation.

• Journal of Korea Multimedia Society
• /
• v.13 no.5
• /
• pp.722-729
• /
• 2010

Prediction problems are widely used in medical domains. For example, computer aided diagnosis or prognosis is a key component in a CDSS (Clinical Decision Support System). SVMs with nonlinear kernels like RBF kernels, have shown superior accuracy in prediction problems. However, they are not preferred by physicians for medical prediction problems because nonlinear SVMs are difficult to visualize, thus it is hard to provide intuitive interpretation of prediction results to physicians. Nomogram was proposed to visualize SVM classification models. However, it cannot visualize nonlinear SVM models. Localized Radial Basis Function (LRBF) was proposed which shows comparable accuracy as the RBF kernel while the LRBF kernel is easier to interpret since it can be linearly decomposed. This paper presents a new tool named VRIFA, which integrates the nomogram and LRBF kernel to provide users with an interactive visualization of nonlinear SVM models, VRIFA visualizes the internal structure of nonlinear SVM models showing the effect of each feature, the magnitude of the effect, and the change at the prediction output. VRIFA also performs nomogram-based feature selection while training a model in order to remove noise or redundant features and improve the prediction accuracy. The area under the ROC curve (AUC) can be used to evaluate the prediction result when the data set is highly imbalanced. The tool can be used by biomedical researchers for computer-aided diagnosis and risk factor analysis for diseases.

• Journal of the Korean Chemical Society
• /
• v.53 no.6
• /
• pp.653-662
• /
• 2009

Developing effective tools for predicting absorption, distribution, metabolism, excretion properties and toxicity (ADME/T) of new chemical entities in the early stage of drug design is one of the most important tasks in drug discovery and development today. As one of these attempts, support vector machines (SVM) has recently been exploited for the prediction of ADME/T related properties. However, two problems in SVM modeling, i.e. feature selection and parameters setting, are still far from solved. The two problems have been shown to be crucial to the efficiency and accuracy of SVM classification. In particular, the feature selection and optimal SVM parameters setting influence each other, which indicates that they should be dealt with simultaneously. In this account, we present an integrated practical solution, in which genetic-based algorithm (GA) is used for feature selection and grid search (GS) method for parameters optimization. hERG ion-channel inhibitor classification models of ADME/T related properties has been built for assessing and testing the proposed GA-GS-SVM. We generated 6 different models that are 3 different single models and 3 different ensemble models using training set - 1891 compounds and validated with external test set - 175 compounds. We compared single model with ensemble model to solve data imbalance problems. It was able to improve accuracy of prediction to use ensemble model.

• Journal of KIISE:Software and Applications
• /
• v.36 no.12
• /
• pp.996-1008
• /
• 2009

This paper proposes a novel approach for the model-based pose matching of a human body using Active Shape Models. To improve the processing time of model creation and registration, we use a skeleton-type model instead of the conventional silhouette-based models. The skeleton model defines feature information that is used to match the human pose. Images used to make the model are for 600 human bodies, and the model has 17 landmarks which indicate the body junction and key features of a human pose. When applying primary Active Shape Models to the skeleton-type model in the matching process, a problem may occur in the proximal joints of the arm and leg due to the color variations on a human body and the insufficient information for the fore-rear directions of profile normals. This problem is solved by using the background subtraction information of a body region in the input image and adding a 4-directions feature of the profile normal in the proximal parts of the arm and leg. In the matching process, the maximum iteration is less than 30 times. As a result, the execution time is quite fast, and was observed to be less than 0.03 sec in an experiment.