Cheong Hee Park

Chungnam National University, Daiden, Daejeon, South Korea

Are you Cheong Hee Park?

Claim your profile

Publications (31)11.18 Total impact

  • Yunji Kim, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Query by Singing/Humming(QBSH) is to retrieve songs in the music database using user's singing or humming. Open-End Dynamic Time Warping (OEDTW) is one of the methods which are commonly used in QBSH studies. This paper proposes a method for improving OEDTW performance using optimal scaling factor taken from Linear scaling phase. The proposed method finds an optimal global scale of a query for each target song by linear scaling and key shifting of the query and target song over the obtained scale can complement DTW or OEDTW. The experimental results demonstrate that the proposed method can improve Top-1 hit rate by 32% compared with the original DTW.
    Proceedings of the 2013 International Conference on Signal-Image Technology & Internet-Based Systems; 12/2013
  • Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: In semi-supervised learning, when the number of data samples with class label information is very small, information from unlabeled data is utilized in the learning process. Many semi-supervised learning methods have been presented and have exhibited competent performance. Active learning also aims to overcome the shortage of labeled data by obtaining class labels for some selected unlabeled data from experts. However, the selection process for the most informative unlabeled data samples can be demanding when the search is performed over a large set of unlabeled data. In this paper, we propose a method for batch mode active learning in graph-based semi-supervised learning. Instead of acquiring class label information of one unlabeled data sample at a time, we obtain information about several data samples at once, reducing time complexity while preserving the beneficial effects of active learning. Experimental results demonstrate the improved performance of the proposed method.
    International Journal of Pattern Recognition and Artificial Intelligence 11/2013; 27(07). · 0.56 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: We propose a user pose verification technique using depth information to compare their pose with expert riders. Xtion sensor by Asus is used for gathering a depth data in real world. The user pose verification algorithm is divided into two categories: user segmentation and user pose verification. In user segmentation step, body parts are segmented based on the region growing algorithm from a head point (i.e. seed point). Then, a simple algorithm is used to generate skeletal joints in the segmented body parts. Finally, the user pose is investigated by the standard pose of experts in the same situation.
    RO-MAN, 2013 IEEE; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we measured and analyzed the driver's EEG for lane change behavior to measure and quantify drivers' workload. The experiment took place in the real road. We extracted the reference and the lane change behavior section and calculated the EEG variation rates from the EEG values. We performed paired-sample T-test to know EEG variation rate of reference and lane change section. Also, we performed ANOVA analysis to know the difference of EEG variation rate dependent on driving road type. We performed independent samples T-test to know the difference of EEG variation rate dependent on gender types.
    ICT Convergence (ICTC), 2013 International Conference on; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a number recognition method to quickly and accurately get bus route information for the visually-impaired from natural scenes. We take a boosting approach based on Modified Census Transform (MCT) features, applied with success to number detection. Then, a gradient-based feature extraction algorithm is used to recognize the route number from detected samples. Finally, the recognition result is spoken by the Text-To-Speech (TTS) engine via the speaker embedded in tablet PC so that this information can be used when they take a bus without someone's help. In the experiment, we evaluate the performance of number detection and recognition in our dataset. We achieve high accuracy in both detection and recognition to provide reliable information of the bus routes.
    Ubiquitous Robots and Ambient Intelligence (URAI), 2013 10th International Conference on; 01/2013
  • Hoyoung Woo, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Active learning aims to select data samples which would be the most informative to improve classification performance so that their class labels are obtained from an expert. Recently, an active learning method based on locally linear reconstruction(LLR) has been proposed and the performance of LLR was demonstrated well in the experiments comparing with other active learning methods. However, the time complexity of LLR is very high due to matrix operations required repeatedly for data selection. In this paper, we propose an efficient active learning method based on random sampling and backward deletion. We select a small subset of data samples by random sampling from the total data set, and a process of deleting the most redundant points in the subset is performed iteratively by searching for a pair of data samples having the smallest distance. The distance measure using a graph-based shortest path distance is utilized in order to consider the underlying data distribution. Experimental results demonstrate that the proposed method has very low time complexity, but the prediction power of data samples selected by our method outperforms that by LLR.
    Proceedings of the third Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering; 10/2012
  • Hoyoung Woo, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Ensemble learning has been widely used in data mining and pattern recognition. However, when the number of labeled data samples is very small, it is difficult to train a base classifier for ensemble learning, therefore, it is necessary to utilize an abundance of unlabeled data effectively. In most semi-supervised ensemble methods, the label prediction of unlabeled data and their use as pseudo-label data are common processes. However, the low accuracy of the label prediction of unlabeled data limits the ability to obtain improved ensemble members. In this paper, we propose effective ensemble learning methods for semi-supervised classification that combine label propagation and ensemble learning. We show that accurate ensemble members can be constructed using class labels predicted by a label propagation method, and unlabeled data samples are fully utilized for diverse ensemble member construction. Extensive experimental results demonstrate the performance of the proposed methods.
    Computer and Information Technology (CIT), 2012 IEEE 12th International Conference on; 01/2012
  • Young Tae Lee, Yong Joon Shin, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Linear discriminant analysis(LDA) is a traditional dimension reduction method which finds projective directions to maximize separability between classes. However, when the number of labeled data points is small, the performance of LDA is degraded severely. In this paper, we propose two improved methods for LDA which utilizes abundant unlabeled data. Instead of using all the unlabeled data as in most of semi- supervised dimension reduction methods, we select confident unlabeled data and develop extended LDA algorithms. In the first method, a graph-based LDA method is developed to utilize confidence scores for chosen unlabeled data so that unlabeled data with a low confidence score contributes smaller than unlabeled data with a high confidence score. In the second method, selected unlabeled data points are used to modify the centroids of classes in an objective function of LDA. Extensive experimental results in text classification demonstrates the effectiveness of the proposed methods compared with other semi-supervised dimension reduction methods.
    01/2011;
  • Source
    Yong Joon Shin, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Dimension reduction is an important topic in data mining and machine learning. Especially dimension reduction combined with feature fusion is an effective preprocessing step when the data are described by multiple feature sets. Canonical Correlation Analysis (CCA) and Discriminative Canonical Correlation Analysis (DCCA) are feature fusion methods based on correlation. However, they are different in that DCCA is a supervised method utilizing class label information, while CCA is an unsupervised method. It has been shown that the classification performance of DCCA is superior to that of CCA due to the discriminative power using class label information. On the other hand, Linear Discriminant Analysis (LDA) is a supervised dimension reduction method and it is known as a special case of CCA. In this paper, we analyze the relationship between DCCA and LDA, showing that the projective directions by DCCA are equal to the ones obtained from LDA with respect to an orthogonal transformation. Using the relation with LDA, we propose a new method that can enhance the performance of DCCA. The experimental results show that the proposed method exhibits better classification performance than the original DCCA.
    Applied Mathematics and Computer Science. 01/2011; 21:549-558.
  • Cheong Hee Park, Hongsuk Shim
    IJPRAI. 01/2010; 24:1-14.
  • Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Dimension reduction has been applied in various areas of pattern recognition and data mining. While a traditional dimension reduction method, Principal Component Analysis (PCA) finds projective directions to maximize the global scatter in data, Locality Preserving Projection (LPP) pursues linear dimension reduction to minimize the local scatter. However, the discriminative power by either global or local scatter optimization is not guaranteed to be effective for classification. A recently proposed method, Unsupervised Discriminant Projection (UDP) aims to minimize the local scatter among near points and maximize the global scatter of distant points at the same time. Although its performance has been proven to be comparable to other dimension reduction methods, PCA preprocessing step due to the singularity of global and local scatter matrices may degrade the performance of UDP. In this paper, we propose several algorithms to improve the performances of UDP greatly. An improved algorithm for UDP is presented which applies the Generalized Singular Value Decomposition (GSVD) to overcome singularities of scatter matrices in UDP. Two-dimensional UDP and nonlinear extension of UDP are also proposed. Extensive experimental results demonstrate superiority of the proposed algorithms.
    International Journal of Pattern Recognition and Artificial Intelligence 01/2010; 24:193-206. · 0.56 Impact Factor
  • Cheong Hee Park, Moonhwi Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: Associative classification has been recently proposed which combines association rule mining and classification, and many studies have shown that associative classifiers give high prediction accuracies compared with other traditional classifiers such as a decision tree. However, in order to apply association rule mining to classification problems, data transformation into the form of transaction data should be preceded before applying association rule mining. In this paper, we propose a discretization method based on Support vector machines, which can improve the performance of association classification greatly. The proposed method finds optimal class boundaries by using SVM, and discretization utilizing distances to the boundaries is performed. Experimental results demonstrate that performing SVM-based discretization for continuous attributes makes associative classification more effective in that it reduces the number of classification rules mined and also improves the prediction accuracies at the same time.
    Expert Syst. Appl. 01/2009; 36:4784-4787.
  • Gaksoo Lim, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Semi-supervised learning aims to utilize unlabeled data in the process of supervised learning. In particular, combining semi-supervised learning with dimension reduction can reduce overfitting caused by small sample size in high dimensional data. By graph representation with similarity edge weights among data samples including both labeled and unlabeled data, statistical and geometricstructures in data are utilized to explore clustering structure of a small number of labeled data samples. However, most of semi-supervised dimension reduction methods use the information induced from unlabeled data points to modify only within-class scatter of labeled data, since unlabeled data can not give any information about distance between classes. In this paper, we propose semi-supervised dimension reduction which reinforcebetween-class distance by using a penalty graph and minimize within-class scatter by using a similarity graph. We apply our approach to extend linear dimension reduction methods such as Linear discriminant analysis (LDA) and Maximum margin criterion (MMC) and demonstrate that modifying between-class distance as well can make great impacts on classification performance.
    Ninth IEEE International Conference on Computer and Information Technology, Xiamen, China, CIT 2009, 11-14 October 2009, Proceedings, Volume I; 01/2009
  • Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Dimension reduction is a preprocessing step by which small number of optimal features are extracted. Among several statistical dimension reduction methods, Linear discriminant analysis (LDA) performs dimension reduction to maximize class separability in the reduced dimensional space. However, in multi-labeled problems, data samples belonging to multiple classes cause contradiction between the maximization of the distances between classes and the minimization of the scatter within classes, since they are placed in the overlapping area of multiple classes. In this paper, we show that in multi-labeled text categorization, the outputs from multiple linear methods can be used to compose new features for low dimensional representation. Especially, we apply least squares regression and a linear support vector machine (SVM) for multiple binary-class problems constructed from a multi-labeled problem and obtain optimal features in a low dimensional space which are fed into another classification algorithm. Extensive experimental results in text categorization are presented comparing with other dimension reduction methods and multi-label classification algorithms.
    Computer and Information Technology, 2008. CIT 2008. 8th IEEE International Conference on; 08/2008
  • Cheong Hee Park, Moonhwi Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: Associative classification has been recently proposed which combines association rule mining and classification, and many studies have shown that associative classifiers have high prediction accuracies. In order to apply an asso- ciation rule mining to classification problem, data transfor- mation into the form of transaction data should be preceded before applying association rule mining. In this paper, we propose a discretization method based on Support vector machines, which is very effective for association classifica- tion. The proposed method finds optimal class boundaries by using SVM, and discretization utilizing distances to the boundaries is performed. Experimental results demonstrate that performing SVM-based discretization for continuous attributes makes associative classification more effective in that it reduces the number of association rules mined and also improves the prediction accuracies at the same time.
    Computational Intelligence and Security, 2007 International Conference on; 01/2008
  • Source
    IADIS European Conference on Data Mining 2008, Amsterdam, The Netherlands, July 24-26, 2008. Proceedings; 01/2008
  • Source
    Cheong Hee Park, Moonhwi Lee
    [Show abstract] [Hide abstract]
    ABSTRACT: Linear discriminant analysis (LDA) is one of the most popular dimension reduction methods, but it is originally focused on a single-labeled problem. In this paper, we derive the formulation for applying LDA for a multi-labeled problem. We also propose a generalized LDA algorithm which is effective in a high dimensional multi-labeled problem. Experimental results demonstrate that by considering multi-labeled structure, LDA can achieve computational efficiency and also improve classification performances.
    Pattern Recognition Letters. 01/2008;
  • Moonhwi Lee, Cheong Hee Park
    [Show abstract] [Hide abstract]
    ABSTRACT: While much progress has been made in face recognition over the last decades, changes in illumination directions still remain as a difficult problem. In this paper, we propose an efficient image normalization method which can overcome illumination effects effectively. The proposed method is based on intensity distribution transformation. However, instead of applying it globally, transformation in intensity distribution is performed for each column independently using one frontal mean face as a reference image. Since it does not require image processing such as image segmentation, the computational complexity is very low and it can circumvent boundary discontinuity caused by region segmentation. Extensive experimental results using Feret database and extended Yale B database demonstrate the competence of the proposed method.
    Proceedings of the 1st ACM SIGMM International Conference on Multimedia Information Retrieval, MIR 2008, Vancouver, British Columbia, Canada, October 30-31, 2008; 01/2008
  • Source
    Cheong Hee Park, Haesun Park
    [Show abstract] [Hide abstract]
    ABSTRACT: Linear discriminant analysis (LDA) is a dimension reduction method which finds an optimal linear transformation that maximizes the class separability. However, in undersampled problems where the number of data samples is smaller than the dimension of data space, it is difficult to apply LDA due to the singularity of scatter matrices caused by high dimensionality. In order to make LDA applicable, several generalizations of LDA have been proposed recently. In this paper, we present theoretical and algorithmic relationships among several generalized LDA algorithms and compare their computational complexities and performances in text classification and face recognition. Towards a practical dimension reduction method for high dimensional data, an efficient algorithm is proposed, which reduces the computational complexity greatly while achieving competitive prediction accuracies. We also present nonlinear extensions of these LDA algorithms based on kernel methods. It is shown that a generalized eigenvalue problem can be formulated in the kernel-based feature space, and generalized LDA algorithms are applied to solve the generalized eigenvalue problem, resulting in nonlinear discriminant analysis. Performances of these linear and nonlinear discriminant analysis algorithms are compared extensively.
    Pattern Recognition 01/2008; · 2.63 Impact Factor
  • Cheong Hee Park, Hongsuk Shim
    [Show abstract] [Hide abstract]
    ABSTRACT: Most of classifiers implicitly assume that data samples belong to at least one class among predefined classes. How- ever, all data patterns may not be known at the time of data collection or a new pattern can be emerging over time. Hence ideal classifiers need to be able to recognize an emerging pattern. In this paper, we explore the per- formances and limitations of the existing classification sys- tems in detecting a new class. Also a new method is pro- posed that can monitor the change in class distribution and detect an emerging class. It works under the supervised learning model where along with classification an emerg- ing class with new characteristic is detected so that classi- fication model can be adapted systematically. For detection of an emerging class, we design statistical significance test- ing for signaling change of class distribution. When the alarm for new class generation is set on, candidates for new class members are retrieved for close examination by experts. Our experimental results demonstrate competent performances of the proposed method.
    Granular Computing, 2007. GRC 2007. IEEE International Conference on; 12/2007

Publication Stats

221 Citations
11.18 Total Impact Points

Institutions

  • 2006–2013
    • Chungnam National University
      • Department of Computer Science and Engineering
      Daiden, Daejeon, South Korea
  • 2008
    • Georgia Institute of Technology
      • College of Computing
      Atlanta, GA, United States
  • 2004–2005
    • University of Minnesota Twin Cities
      • Department of Computer Science and Engineering
      Minneapolis, MN, United States
    • University of Minnesota Duluth
      • Department of Computer Science
      Duluth, MN, United States