• Journal of KIISE:Software and Applications
• /
• v.30 no.3_4
• /
• pp.251-258
• /
• 2003

A support vector machine (SVM) is supposed to provide a good generalization performance, but the actual performance of a actually implemented SVM is often far from the theoretically expected level. This is largely because the implementation is based on an approximated algorithm, due to the high complexity of time and space. To improve this limitation, we propose ensemble of SVMs by using Bagging (bootstrap aggregating) and Boosting. By a Bagging stage each individual SVM is trained independently using randomly chosen training samples via a bootstrap technique. By a Boosting stage an individual SVM is trained by choosing training samples according to their probability distribution. The probability distribution is updated by the error of independent classifiers, and the process is iterated. After the training stage, they are aggregated to make a collective decision in several ways, such ai majority voting, the LSE(least squares estimation) -based weighting, and double layer hierarchical combining. The simulation results for IRIS data classification, the hand-written digit recognition and Face detection show that the proposed SVM ensembles greatly outperforms a single SVM in terms of classification accuracy.

• The Journal of the Korea Contents Association
• /
• v.8 no.1
• /
• pp.259-271
• /
• 2008

We propose a protein function finding algorithm that is able to predict specific molecular function for unannotated proteins through domain analysis from protein-protein network. To do this, we first construct protein-protein interaction(PPI) network in Saccharomyces cerevisiae from MIPS databases. The PPI network(proteins; 3,637, interactions; 10,391) shows the characteristics of a scale-free network and a hierarchical network that proteins with a number of interactions occur in small and the inherent modularity of protein clusters. Protein-protein interaction databases obtained from a Y2H(Yeast Two Hybrid) screen or a composite data set include random false positives. To filter the database, we reconstruct the PPI networks based on the cellular localization. And then we analyze Hub proteins and the network structure in the reconstructed network and define structural modules from the network. We analyze protein domains from the structural modules and derive functional modules from them. From the derived functional modules with high certainty, we find tentative functions for unannotated proteins.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.9 no.7
• /
• pp.1588-1593
• /
• 2005

In coagulant control of water treatment plants, rule extraction, one of datamining categories, was performed for coagulant control of a water treatment plant. Clustering methods were applied to extract control rules from data. These control rules can be used for fully automation of water treatment plants instead of operator's knowledge for plant control. To perform fuzzy clustering, there are some coefficients to be determined and these kinds of studies have been performed over decades such as clustering indices. In this study, statistical indices were taken to calculate the number of clusters. Simultaneously, seed points were found out based on hierarchical clustering. These statistical approaches give information about features of clusters, so it can reduce computing cost and increase accuracy of clustering. The proposed algorithm can play an important role in datamining and knowledge discovery.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2007.06a
• /
• pp.61-64
• /
• 2007

MPLS networks architectures have been prevailed as scalable approach to provide quality of service in the Internet. Many researches have been mainly focused on scalability and multi-path calculation scheme for failure restoration. However, the MPLS network have an scalability problem about traffic, and the existing failure restoration methods are wasted resources, and it has the problem that loss of a packet by a lot of delay occurs in too. Therefore in this paper, we propose a H-MPLS (Hierachical-Multiprotocol Label Switching)network for rapidly failure restorations and effective management of network extended. The proposed H-MPLS Network apply LSP multi-path routing algorithm and consists of several MPLS. After comparing the performance among the existing failure restoration methods with various network models, we use NS simulator in order to analyze the performance. Finally, we present an improvement scheme of the efficiency and scalability.

• Journal of Korea Multimedia Society
• /
• v.14 no.10
• /
• pp.1287-1302
• /
• 2011

Clustering protocol that is used in wireless sensor network is an efficient method that extends the lifetime of the network. However, when this method is applied to an environment in which collected data of the sensor node easily overlap, sensor nodes unnecessarily consumes energy. In the case of clustering technique that uses a threshold, the lifetime of the network is extended but the degree of accuracy of collected data is low. Therefore it is hard to trust the data and improvement is needed. In addition, it is hard for the clustering protocol that uses multi-hop transmission to normally collect data because the selection of a cluster head node occurs at random and therefore the link of nodes is often disconnected. Accordingly this paper suggested a cluster-formation algorithm that reduces unnecessary energy consumption and that works with an alleviated link disconnection. According to the result of performance analysis, the suggested method lets the nodes consume less energy than the existing clustering method and the transmission efficiency is increased and the entire lifetime is prolonged by about 30%.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.2
• /
• pp.136-141
• /
• 2004

There has been enormous growth in the amount of commercial and scientific data, such as retail transactions, protein sequences, and web-logs. Such datasets consist of sequence data that have an inherent sequential nature. However, few clustering algorithms consider sequentiality. In this paper, we study how to cluster sequence datasets. We propose a new similarity measure to compute the similarity between two sequences. We also present an efficient method for determining the similarity measure and develop a clustering algorithm. Due to the high computational complexity of hierarchical clustering algorithms for clustering large datasets, a new clustering method is required. Therefore, we propose a new scalable clustering method using sampling and a k-nearest-neighbor method. Using a real dataset and a synthetic dataset, we show that the quality of clusters generated by our proposed approach is better than that of clusters produced by traditional algorithms.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.59 no.5
• /
• pp.972-980
• /
• 2010

In this study, we introduce the design methodology of an optimized type-2 fuzzy cascade controller with the aid of hierarchical fair competition-based genetic algorithm(HFCGA) for ball & beam system. The ball & beam system consists of servo motor, beam and ball, and remains mutually connected in line in itself. The ball & beam system determines the position of ball through the control of a servo motor. Consequently the displacement change of the position of the moving ball and its ensuing change of the angle of the beam results in the change of the position angle of a servo motor. The type-2 fuzzy cascade controller scheme consists of the outer controller and the inner controller as two cascaded fuzzy controllers. In type-2 fuzzy logic controller(FLC) as the expanded type of type-1 fuzzy logic controller(FLC), we can effectively improve the control characteristic by using the footprint of uncertainty(FOU) of membership function. The control parameters(scaling factors) of each fuzzy controller using HFCGA which is a kind of parallel genetic algorithms(PGAs). HFCGA helps alleviate the premature convergence being generated in conventional genetic algorithms(GAs). We estimated controller characteristic parameters of optimized type-2 fuzzy cascade controller applied ball & beam system such as maximum overshoot, delay time, rise time, settling time and steady-state error. For a detailed comparative analysis from the viewpoint of the performance results and the design methodology, the proposed method for the ball & beam system which is realized by the fuzzy cascade controller based on HFCGA, is presented in comparison with the conventional PD cascade controller based on serial genetic algorithms.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.6
• /
• pp.783-788
• /
• 2004

Many reported methods assume that the faces in an image or an image sequence have been identified and localization. Face detection from image is a challenging task because of the variability in scale, location, orientation and pose. The difficulties in visual detection and recognition are caused by the variations in viewpoint, viewing distance, illumination. In this paper, we present an efficient linear discriminant for multi-view face detection and face location. We define the training data by using the Fisher`s linear discriminant in an efficient learning method. Face detection is very difficult because it is influenced by the poses of the human face and changes in illumination. This idea can solve the multi-view and scale face detection problems. In this paper, we extract the face using the Fisher`s linear discriminant that has hierarchical models invariant size and background. The purpose of this paper is to classify face and non-face for efficient Fisher`s linear discriminant.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.37 no.1
• /
• pp.66-75
• /
• 2000

In this paper, we present the efficient image transmission system using designed source-channel combined coder in W-CDMA mobile communication system. In proposed schemes, we decompose the wavelet transformed hierarchical band-images into some types of different size blocks which have different properties in error sensitivity. The RS(Reed-Solomon) coder with different coding rate is used for each decomposed source blocks which has different importance. In addition, we combine retransmitted error frames in Truncated Hybrid Type I ARQ. The proposed algorithm shows efficient image transmission methods because it is not much degraded in PSNR compared with the existing not combined source-channel coder in erroneous wireless channel.

• 전자공학회논문지 IE
• /
• v.43 no.2
• /
• pp.87-92
• /
• 2006

Many watermarking methods for protecting the intellectual property right and authentication multimedia data have been proposed in recent years. In image authentication watermarking, watermark is inserted into an image to detect any malicious alteration. So, watermark for authentication and integrity should be erased easily when the image is changed by scaling or cropping etc. We propose in this paper a fragile watermarking algorithm for image integrity verification and tamper localization proofing using special hierarchical structure. In the proposed method, the image to be watermarked is divided into blocks in a multi-level hierarchy and calculating block digital signatures in thus hierarchy. At each level of the hierarchy, a digital signature for each block is calculated using the seven most significant bit(MSBs)-plane values of all pixels within the block. And the resulting signature is incorporated into the LSBs of selected pixels within the block. We provide experimental results to demonstrate the effectiveness of the proposed method.