• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.6
• /
• pp.43-49
• /
• 2004

The classification time required by conventional multi-class SVMs(Support Vector Machines) greatly increases as the number of pattern classes increases. This is due to the fact that the needed set of binary class SVMs gets quite large. In this paper, we propose a method to reduce the number of classes by using nearest neighbor rule (NNR) in the principle component analysis and linear discriminant analysis (PCA+LDA) feature subspace. The proposed method reduces the number of face classes by selecting a few classes closest to the test data projected in the PCA+LDA feature subspace. Results of experiment show that our proposed method has a lower error rate than nearest neighbor classification (NNC) method. Though our error rate is comparable to the conventional multi-class SVMs, the classification process of our method is much faster.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.10
• /
• pp.73-81
• /
• 2015

The paper proposes a prototype selection method and evaluates the generalization performance of standard algorithms and prototype based classification learning. The proposed prototype classifier defines multidimensional spheres with variable radii within class areas and generates a small set of training data. The nearest-neighbor classifier uses the new training set for predicting the class of test data. By decomposing bias and variance of the mean expected error value, we compare the generalization errors of k-nearest neighbor, Bayesian classifier, prototype selection using fixed radius and the proposed prototype selection method. In experiments, the bias-variance changing trends of the proposed prototype classifier are similar to those of nearest neighbor classifiers with all training data and the prototype selection rates are under 27.0% on average.

• Journal of the Korean Society for Precision Engineering
• /
• v.16 no.12
• /
• pp.119-125
• /
• 1999

This paper gives an overview of the various approaches to designing statistical pattern recognition scheme based on Bayes discrimination rule and the artificial neural networks for rotating machine condition classification. Concerning to Bayes discrimination rule, this paper contains the linear discrimination rule applied to classification into several multivariate normal distributions with common covariance matrices, the quadratic discrimination rule under different covariance matrices. Also we discribes k-nearest neighbor method to directly estimate a posterior probability of each class. Five features are extracted in time domain vibration signals. Employing these five features, statistical pattern classifier and neural networks have been established to detect defects on rotating machine. Four different cases of rotation machine were observed. The effects of k number and neural networks structures on monitoring performance have also been investigated. For the comparison of diagnosis performance of these two method, their recognition success rates are calculated form the test data. The result of experiment which classifies the rotating machine conditions using each method presents that the neural networks shows the highest recognition rate.

• Journal of Korea Multimedia Society
• /
• v.10 no.7
• /
• pp.856-868
• /
• 2007

We propose an effective method for block classification in a document image. The gray level document image is converted to the binary image for a block segmentation. This binary image would be smoothed to find the locations and sizes of each block. And especially during this smoothing, the inner block heights of each block are obtained. The gray level image is divided to several blocks by these location informations. The SGLDM(spatial gray level dependence matrices) are made using the each gray-level document block and the seven second-order statistical texture features are extracted from the (0,1) direction's SGLDM which include the document attributes. Document image blocks are classified to two groups, text and non-text group, by the inner block height of the block at the nearest neighbor rule. The seven texture features(that were extracted from the SGLDM) are used for the five detail categories of small font, large font, table, graphic and photo blocks. These document blocks are available not only for structure analysis of document recognition but also the various applied area.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.29B no.7
• /
• pp.23-29
• /
• 1992

In this paper, a method of representing the pattern classifier as a collection of hypercubic regions is proposed. This representation has following advantages over the conventional ones : 1) a simple form of human knowledge can be used in designing the classifier, 2) the form of the classifier is suit for the rule-based system, and 3) this can reduce the classification time. A method of synthesis of the classifier under this representation is also proposed and the experimental result shows that the proposed method is faster than the well-known nearest neighbor classifier.

• Journal of the Korean Society for Precision Engineering
• /
• v.17 no.7
• /
• pp.98-105
• /
• 2000

In this study, we developed rotating machine vibration condition monitoring system based upon Windows NT and DSP Board. Developed system includes signal analysis module, trend monitoring and simple diagnosis using threshold value. Trend analysis and report generation are offered with database management tool which was developed in MS-ACCESS environment. Post-processor, based upon Matlab, is developed for vibration signal analysis and fault detection using statistical pattern recognition scheme based upon Bayes discrimination rule and neural networks. Concerning to Bayes discrimination rule, the developed system contains the linear discrimination rule with common covariance matrices and the quadratic discrimination rule under different covariance matrices. Also the system contains k-nearest neighbor method to directly estimate a posterior probability of each class. The result of case studies with the data acquired from Pyung-tak LNG pump and experimental setup show that the system developed in this research is very effective and useful.

• Korean Journal of Remote Sensing
• /
• v.31 no.4
• /
• pp.303-319
• /
• 2015

The objectives of this paper are to measure surface imperviousness using three different classification methods: per-pixel, sub-pixel, and object-oriented classification. They are tested on high-spatial resolution QuickBird data at 2.4 meters (four spectral bands and three principal component bands) as well as a medium-spatial resolution Landsat TM image at 30 meters. To measure impervious surfaces, we selected 30 sample sites with different land uses and residential densities across image representing the city of Phoenix, Arizona, USA. For per-pixel an unsupervised classification is first conducted to provide prior knowledge on the possible candidate spectral classes, and then a supervised classification is performed using the maximum-likelihood rule. For sub-pixel classification, a Linear Spectral Mixture Analysis (LSMA) is used to disentangle land cover information from mixed pixels. For object-oriented classification several different sets of scale parameters and expert decision rules are implemented, including a nearest neighbor classifier. The results from these three methods show that the object-oriented approach (accuracy of 91%) provides more accurate results than those achieved by per-pixel algorithm (accuracy of 67% and 83% using Landsat TM and QuickBird, respectively). It is also clear that sub-pixel algorithm gives more accurate results (accuracy of 87%) in case of intensive and dense urban areas using medium-resolution imagery.

• Journal of Intelligence and Information Systems
• /
• v.24 no.3
• /
• pp.21-44
• /
• 2018

In recent years, the rapid development of internet technology and the popularization of smart devices have resulted in massive amounts of text data. Those text data were produced and distributed through various media platforms such as World Wide Web, Internet news feeds, microblog, and social media. However, this enormous amount of easily obtained information is lack of organization. Therefore, this problem has raised the interest of many researchers in order to manage this huge amount of information. Further, this problem also required professionals that are capable of classifying relevant information and hence text classification is introduced. Text classification is a challenging task in modern data analysis, which it needs to assign a text document into one or more predefined categories or classes. In text classification field, there are different kinds of techniques available such as K-Nearest Neighbor, Naïve Bayes Algorithm, Support Vector Machine, Decision Tree, and Artificial Neural Network. However, while dealing with huge amount of text data, model performance and accuracy becomes a challenge. According to the type of words used in the corpus and type of features created for classification, the performance of a text classification model can be varied. Most of the attempts are been made based on proposing a new algorithm or modifying an existing algorithm. This kind of research can be said already reached their certain limitations for further improvements. In this study, aside from proposing a new algorithm or modifying the algorithm, we focus on searching a way to modify the use of data. It is widely known that classifier performance is influenced by the quality of training data upon which this classifier is built. The real world datasets in most of the time contain noise, or in other words noisy data, these can actually affect the decision made by the classifiers built from these data. In this study, we consider that the data from different domains, which is heterogeneous data might have the characteristics of noise which can be utilized in the classification process. In order to build the classifier, machine learning algorithm is performed based on the assumption that the characteristics of training data and target data are the same or very similar to each other. However, in the case of unstructured data such as text, the features are determined according to the vocabularies included in the document. If the viewpoints of the learning data and target data are different, the features may be appearing different between these two data. In this study, we attempt to improve the classification accuracy by strengthening the robustness of the document classifier through artificially injecting the noise into the process of constructing the document classifier. With data coming from various kind of sources, these data are likely formatted differently. These cause difficulties for traditional machine learning algorithms because they are not developed to recognize different type of data representation at one time and to put them together in same generalization. Therefore, in order to utilize heterogeneous data in the learning process of document classifier, we apply semi-supervised learning in our study. However, unlabeled data might have the possibility to degrade the performance of the document classifier. Therefore, we further proposed a method called Rule Selection-Based Ensemble Semi-Supervised Learning Algorithm (RSESLA) to select only the documents that contributing to the accuracy improvement of the classifier. RSESLA creates multiple views by manipulating the features using different types of classification models and different types of heterogeneous data. The most confident classification rules will be selected and applied for the final decision making. In this paper, three different types of real-world data sources were used, which are news, twitter and blogs.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.3
• /
• pp.59-66
• /
• 2016

This paper presents an empirical evaluation on dimensionality reduction strategies by which dissimilarity-based classifications (DBC) can be implemented efficiently. In DBC, classification is not based on feature measurements of individual objects (a set of attributes), but rather on a suitable dissimilarity measure among the individual objects (pair-wise object comparisons). One problem of DBC is the high dimensionality of the dissimilarity space when a lots of objects are treated. To address this issue, two kinds of solutions have been proposed in the literature: prototype selection (PS)-based methods and dimension reduction (DR)-based methods. In this paper, instead of utilizing the PS-based or DR-based methods, a way of performing DBC in Eigen spaces (ES) is considered and empirically compared. In ES-based DBC, classifications are performed as follows: first, a set of principal eigenvectors is extracted from the training data set using a principal component analysis; second, an Eigen space is expanded using a subset of the extracted and selected Eigen vectors; third, after measuring distances among the projected objects in the Eigen space using $l_p$-norms as the dissimilarity, classification is performed. The experimental results, which are obtained using the nearest neighbor rule with artificial and real-life benchmark data sets, demonstrate that when the dimensionality of the Eigen spaces has been selected appropriately, compared to the PS-based and DR-based methods, the performance of the ES-based DBC can be improved in terms of the classification accuracy.

• Journal of Intelligence and Information Systems
• /
• v.17 no.1
• /
• pp.139-152
• /
• 2011

Recommender systems have become an important research field since the emergence of the first paper on collaborative filtering in the mid-1990s. In general, recommender systems are defined as the supporting systems which help users to find information, products, or services (such as books, movies, music, digital products, web sites, and TV programs) by aggregating and analyzing suggestions from other users, which mean reviews from various authorities, and user attributes. However, as academic researches on recommender systems have increased significantly over the last ten years, more researches are required to be applicable in the real world situation. Because research field on recommender systems is still wide and less mature than other research fields. Accordingly, the existing articles on recommender systems need to be reviewed toward the next generation of recommender systems. However, it would be not easy to confine the recommender system researches to specific disciplines, considering the nature of the recommender system researches. So, we reviewed all articles on recommender systems from 37 journals which were published from 2001 to 2010. The 37 journals are selected from top 125 journals of the MIS Journal Rankings. Also, the literature search was based on the descriptors "Recommender system", "Recommendation system", "Personalization system", "Collaborative filtering" and "Contents filtering". The full text of each article was reviewed to eliminate the article that was not actually related to recommender systems. Many of articles were excluded because the articles such as Conference papers, master's and doctoral dissertations, textbook, unpublished working papers, non-English publication papers and news were unfit for our research. We classified articles by year of publication, journals, recommendation fields, and data mining techniques. The recommendation fields and data mining techniques of 187 articles are reviewed and classified into eight recommendation fields (book, document, image, movie, music, shopping, TV program, and others) and eight data mining techniques (association rule, clustering, decision tree, k-nearest neighbor, link analysis, neural network, regression, and other heuristic methods). The results represented in this paper have several significant implications. First, based on previous publication rates, the interest in the recommender system related research will grow significantly in the future. Second, 49 articles are related to movie recommendation whereas image and TV program recommendation are identified in only 6 articles. This result has been caused by the easy use of MovieLens data set. So, it is necessary to prepare data set of other fields. Third, recently social network analysis has been used in the various applications. However studies on recommender systems using social network analysis are deficient. Henceforth, we expect that new recommendation approaches using social network analysis will be developed in the recommender systems. So, it will be an interesting and further research area to evaluate the recommendation system researches using social method analysis. This result provides trend of recommender system researches by examining the published literature, and provides practitioners and researchers with insight and future direction on recommender systems. We hope that this research helps anyone who is interested in recommender systems research to gain insight for future research.