• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.7
• /
• pp.2633-2648
• /
• 2015

The problem of visual words' synonymy and ambiguity always exist in the conventional bag of visual words (BoVW) model based object category methods. Besides, the noisy visual words, so-called "visual stop-words" will degrade the semantic resolution of visual dictionary. In view of this, a novel bag of visual words method based on PLSA and chi-square model for object category is proposed. Firstly, Probabilistic Latent Semantic Analysis (PLSA) is used to analyze the semantic co-occurrence probability of visual words, infer the latent semantic topics in images, and get the latent topic distributions induced by the words. Secondly, the KL divergence is adopt to measure the semantic distance between visual words, which can get semantically related homoionym. Then, adaptive soft-assignment strategy is combined to realize the soft mapping between SIFT features and some homoionym. Finally, the chi-square model is introduced to eliminate the "visual stop-words" and reconstruct the visual vocabulary histograms. Moreover, SVM (Support Vector Machine) is applied to accomplish object classification. Experimental results indicated that the synonymy and ambiguity problems of visual words can be overcome effectively. The distinguish ability of visual semantic resolution as well as the object classification performance are substantially boosted compared with the traditional methods.

• Physical Therapy Korea
• /
• v.19 no.1
• /
• pp.76-85
• /
• 2012

The purpose of this study was to identify the factors determining the participation restriction of chronic stroke patients based on international classification of functioning, disability, and health (ICF) model. Sixty-eight stroke patients participated. The participants were assessed participation restriction using the Korean version of London handicap scale (K-LHS), modified Barthel index (K-MBI) to measure activities of daily living, Berg balance scale (K-BBS) to assess balance, and the center for epidemiologic studies depression (K-CES-D) to gauge depression. Also, 3 minutes walking test (3MWT), gait velocity, asymmetric posture, and family support were assessed. A stepwise multiple regression analysis was used to explore the factors determining participation restriction. There were no significant different in the K-LHS and K-MBI results by gender (p>.05). Correlations between the K-LHS and K-MBI (r=-.656), K-BBS (r=-.543), K-CES-D (r=.266), 3MWT (r=-.363), gait velocity (r=.348), and family support (r=-.389) were significant (p<.05). Also, the K-MBI and family support were the factors that determined participation restriction (p<.05) and that 40.2% of the variation in the K-LHS can be explained. Therefore, it is suggested that evaluation and intervention of patient's activity level and extent of family support is necessary to reduce participation restriction of chronic stroke patients.

• Journal of Information Technology Application
• /
• v.3 no.2
• /
• pp.1-24
• /
• 2001

This purpose of this study is to drive important criteria for improving customer relationship of R institute using data mining techniques. The focus of this research is to consider patterns and interactions of research variables from research management database of R institute, and to classify the outside organizations and the inside organizations for research contract organizations, and to decide the directions of customer relationship management through analyzing the research type and research cost of research topics. In order to drive criteria variables through pattern analysis of the research database, decision tree algorithm is employed. The results show that determinant variables of 17 input variables are research period, overhead cost, R ＆ D cost as variables to classify the outside and inside contract organization.

• Journal of Animal Science and Technology
• /
• v.51 no.2
• /
• pp.97-104
• /
• 2009

Complement of test standard, evaluation methods and models are needed to improve national competitiveness and to exchange superior genetic resources through the comparison of genetic evaluation score among nations in dairy cattle. Therefore, this study was conducted for the application of international standard to Korea considering domestic circumstance by changing linear-classification test score system of 50 classes which is currently used in Korea to system of 9 classes which is used in advanced nations of dairy. 15,230 of holstein cow linear type records with first parity records for the fifteen linear type and one total score from 2001 to 2006 and pedigree data which were collected by the Korean Animal Improvement Association were used in this study. Population classified by 9 levels was more normal distributed than 50 levels. Correlation coefficients between 50 and 9 score system showed over 0.98 by each classification scheme. Therefore, the 50 point system can be substituted with 9 point system due to their highly positive correlation. However, scores in all traits were still very contingent on classifier under the 9 point system (p<0.001), and F values between foot angle and front teat attachment showed high fluctuation depending on classifier. It means that subjective opinions of classifier would influence on linear type score as ever even if class scheme transformed to system of 9 class. Therefore, the relevance of transformation to the 9 point system should be assessed after analyses about various environmental factors.

• Journal of Internet Computing and Services
• /
• v.19 no.1
• /
• pp.27-35
• /
• 2018

This paper is a study to classify malicious applications in Android environment. And studying the threat and behavioral analysis of malicious Android applications. In addition, malicious apps classified by machine learning were performed as experiments. Android behavior analysis can use dynamic analysis tools. Through this tool, API Calls, Runtime Log, System Resource, and Network information for the application can be extracted. We redefined the properties extracted for machine learning and evaluated the results of machine learning classification by verifying between the overall features and the main features. The results show that key features have been improved by 1~4% over the full feature set. Especially, SVM classifier improved by 10%. From these results, we found that the application of the key features as a key feature was more effective in the performance of the classification algorithm than in the use of the overall features. It was also identified as important to select meaningful features from the data sets.

• The Journal of the Korea Contents Association
• /
• v.14 no.1
• /
• pp.364-373
• /
• 2014

The relationship between a service provided via the information system and user satisfaction has been thought of as an important factor for the development of a new service for the information system. In this study, the twelve new key services that are applicable to national R&D report system were derived by web environment changes in step with IT technology developments in order to support the new service for the user. The twelve new key services are as follows; semantic search service for national R&D report, associated report service, RSS service, mesh-up service, topic-map service, open API service, personalized service, collective intelligence service, SNS service, unstructured data service, detailed search service, mailing service. To assess the quality attribute of the twelve new key services in the national R&D report system, a survey was performed. In conclusion, a stepwise service plan for the national R&D report system was proposed which would use the satisfaction coefficient and the results of the service classification. The following step-by-step service should be developed by in this way. The unstructured data service, personalized service, associated report service, topic-map service, open API service, and the collective intelligence service are needed to develop the first step and RSS service, mesh-up service, semantic search service for the national R&D report, mailing service, detailed search service, and SNS service are needed to develop the second step.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.64 no.1
• /
• pp.113-120
• /
• 2015

In this paper, we propose a feature extraction method using the stacked autoencoders which consist of restricted Boltzmann machines. The stacked autoencoders is a sort of deep networks. Restricted Boltzmann machines (RBMs) are probabilistic graphical models that can be interpreted as stochastic neural networks. In terms of pattern classification problem, the feature extraction is a key issue. We use the stacked autoencoders networks to extract new features which have a good influence on the improvement of the classification performance. After feature extraction, fuzzy k-nearest neighbors algorithm is used for a classifier which classifies the new extracted data set. To evaluate the classification ability of the proposed pattern classifier, we make some experiments with several machine learning data sets.

• ETRI Journal
• /
• v.35 no.6
• /
• pp.1134-1143
• /
• 2013

The rising pace of technological change in information and communications technology (ICT) has provoked technological convergence by providing a new mode of diversification. This paper investigates the nature of ICT-based converging technologies by examining comparative empirical evidence on converging versus nonconverging technologies in relation to the following issues: patent application trends, concentration across technologies, the concentration of patenting activity across firms, R&D efforts, and a technology impact index. For this study, a new operational definition of ICT-based converging technology is derived, and a massive quantity of patents, up to around 600,000, is analyzed. This study follows the International Patent Classification as well as the modified European Commission's industry classification system for the classification of technologies and industries, respectively.

• The Journal of Korea Robotics Society
• /
• v.2 no.2
• /
• pp.191-195
• /
• 2007

Abstract MOCIE(Ministry of Commerce, Industry and Energy) handles 6 Projects for Intelligent Robot, whose budget is around 40 Million dollars per year. In this paper we have tried to analyze the state of robot technology of the projects. Each sub-projects has been divided according to the technological classification. Two major projects of Next Generation Growth Engine and 21C Frontier show different state each other. The former is focused on the product while the latter on the technology. Output of 21C Frontier should be linked to the Next Generation Growth Engine, otherwise, it will fail to advance. The project management handles only the quantitative performance such as business results, number of prototype, and number of patents and papers. Technological Capability is essential and it should be managed. This paper proposes efficient classification of robot technology and technology index.

• Journal of the Korean Society for Heat Treatment
• /
• v.36 no.2
• /
• pp.69-76
• /
• 2023

In this study, we present a system for identifying equivalent grades of standardized wire rod steel based on alloy composition using machine learning techniques. The system comprises two models, one based on a supervised multi-class classification algorithm and the other based on unsupervised autoencoder algorithm. Our evaluation showed that the supervised model exhibited superior performance in terms of prediction stability and reliability of prediction results. This system provides a useful tool for non-experts seeking similar grades of steel based on alloy composition.