• Journal of the Korea Society of Computer and Information
• /
• v.9 no.4 s.32
• /
• pp.33-39
• /
• 2004

In this paper. we design the learning algorithm of LVQ which is used Forward Counter Propagation Networks to improve classification performance of LVQ networks. The weights of Forward Counter Propagation Networks which is between input layer and cluster layer can be learned to determine initial reference vectors by using SOM algorithm and to learn reference vectors by using LVQ algorithm. Finally. pattern vectors is classified into subclasses by neurons which is being in the cluster layer, and the weights of Forward Counter Propagation Networks which is between cluster layer and output layer is learned to classify the classified subclass, which is enclosed a class. Also. kr the number of classes is determined, the number of neurons which is being in the input layer, cluster layer and output layer can be determined. To prove the performance of the proposed learning algorithm. the simulation is performed by using training vectors and test vectors that ate Fisher's Iris data, and classification performance of the proposed learning method is compared with ones of the conventional LVQ, and it was a confirmation that the proposed learning method is more successful classification than the conventional classification.

• Proceedings of the KIEE Conference
• /
• 2002.07a
• /
• pp.108-110
• /
• 2002

In this paper, a fault location algorithm is suggested for line to line faults in distribution networks. Conventional fault location algorithms use the symmetrical component transformation, a very useful tool for transmission network analysis. However, its application is restricted to balanced network only. Distribution networks are, in general, operated in unbalanced manners, therefore, conventional methods cannot be applied directly, which is the reason why there are few research results on fault location in distribution networks. Especially, the line to line fault is considered as a more difficult subject. The proposed algorithm uses direct 3-phase circuit analysis, which means it can be applied not only to balanced networks but also to unbalanced networks like distribution a network. The comparisons of simulation results between one of conventional methods and the suggested method are presented to show its effectiveness and accuracy.

• Journal of Communications and Networks
• /
• v.4 no.3
• /
• pp.161-169
• /
• 2002

Ad hoc wireless networks involving large populations of scattered communication nodes will play a key role in the development of low power, high capacity, interactive, multimedia communication networks. Such networks must support arbitrary network connections and provide coverage anywhere and anytime. This paper partitions such arbitrarily connected network architectures into three distinct groups, identifies the associated dual network architectures and counts the number of network architectures assuming there exist N network nodes. Connectivity between network nodes is characterized as a random event. Defining the link availability P as the probability that two arbitrary network nodes in an ad hoc network are directly connected, the network connection probability $ \integral_n$(p) that any two network nodes will be directly or indirectly connected is derived. The network connection probability $ \integral_n$(p) is evaluated and graphically demonstrated as a function of p and N. It is shown that ad hoc wireless networks containing a large number of network nodes possesses the same network connectivity performance as does a fixed network, i.e., for p>0, $lim_{N\to\infty} Integral_n(p)$ = 1. Furthermore, by cooperating with fixed networks, the ad hoc network connection probability is used to derive the global network connection probability for hybrid networks. These probabilities serve to characterize network connectivity performance for users of wireless ad hoc and hybrid networks, e.g., IEEE 802.11, IEEE 802.15, IEEE 1394-95, ETSI BRAN HIPERLAN, Bluetooth, wireless ATM and the world wide web (WWW).

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.7
• /
• pp.1638-1656
• /
• 2013

The demand for multimedia streaming services is increasing rapidly. To meet this demand, there has been much research and many practical developments for providing multimedia services. A push-to-talk (PTT) service is one of the multimedia streaming services that have been deployed not only over IP multimedia subsystem (IMS) but also in peer-to-peer (P2P) overlay networks. The benefit of PTT has been demonstrated in the literature. However, the need for using PTT service in communication can be arbitrary among users, regardless what kind of PTT services they use. This demand does not support current PTT systems, so an expansion of PTT services still be limited. Moreover, the combination of PTT services in IMS and P2P networks will help operators to provide more scalable PTT services. Therefore, in this paper, we proposed a model to support PTT service interworking between IMS and P2P overlay networks. We also introduced our system design and some interworking service scenarios. We confirmed our architecture through implementation and testing.

• The KIPS Transactions:PartC
• /
• v.13C no.4 s.107
• /
• pp.481-490
• /
• 2006

In a super-peer based P2P network, The network is clustered and each cluster is managed by a special peer, called a super-peer which has information of all peers in its cluster. This clustered P2P model is known to have efficient information search and less traffic load. In this paper, we first estimate the message traffic cost caused by peer's query, join and update actions within a cluster as well as between the clusters and with these values, we present the optimal number of super-peers that minimizes the traffic cost for the various size of super-peer based P2P networks.rks.

• The Journal of Society for e-Business Studies
• /
• v.23 no.3
• /
• pp.207-224
• /
• 2018

In this article, an empirical study was conducted by using public dataset from Lending Club Corporation, the largest online peer-to-peer (P2P) lending in the world. We explore significant predictor variables related to P2P lending default that housing situation, length of employment, average current balance, debt-to-income ratio, loan amount, loan purpose, interest rate, public records, number of finance trades, total credit/credit limit, number of delinquent accounts, number of mortgage accounts, and number of bank card accounts are significant factors to loan funded successful on Lending Club platform. We developed online P2P lending default prediction models using discriminant analysis, logistic regression, neural networks, and decision trees (i.e., CART and C5.0) in order to predict P2P loan default. To verify the feasibility and effectiveness of P2P lending default prediction models, borrower loan data and credit data used in this study. Empirical results indicated that neural networks outperforms other classifiers such as discriminant analysis, logistic regression, CART, and C5.0. Neural networks always outperforms other classifiers in P2P loan default prediction.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.61 no.5
• /
• pp.744-752
• /
• 2012

In this study, the Polynomial-based Radial Basis Function Neural Networks is proposed as an one of the recognition part of overall face recognition system that consists of two parts such as the preprocessing part and recognition part. The design methodology and procedure of the proposed pRBFNNs are presented to obtain the solution to high-dimensional pattern recognition problems. In data preprocessing part, Principal Component Analysis(PCA) which is generally used in face recognition, which is useful to express some classes using reduction, since it is effective to maintain the rate of recognition and to reduce the amount of data at the same time. However, because of there of the whole face image, it can not guarantee the detection rate about the change of viewpoint and whole image. Thus, to compensate for the defects, Linear Discriminant Analysis(LDA) is used to enhance the separation of different classes. In this paper, we combine the PCA&LDA algorithm and design the optimized pRBFNNs for recognition module. The proposed pRBFNNs architecture consists of three functional modules such as the condition part, the conclusion part, and the inference part as fuzzy rules formed in 'If-then' format. In the condition part of fuzzy rules, input space is partitioned with Fuzzy C-Means clustering. In the conclusion part of rules, the connection weight of pRBFNNs is represented as two kinds of polynomials such as constant, and linear. The coefficients of connection weight identified with back-propagation using gradient descent method. The output of the pRBFNNs model is obtained by fuzzy inference method in the inference part of fuzzy rules. The essential design parameters (including learning rate, momentum coefficient and fuzzification coefficient) of the networks are optimized by means of Differential Evolution. The proposed pRBFNNs are applied to face image(ex Yale, AT&T) datasets and then demonstrated from the viewpoint of the output performance and recognition rate.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.2 s.34
• /
• pp.113-121
• /
• 2005

In this paper, we design the hybrid learning algorithm of LVQ which is to perform EMG pattern recognition. The proposed hybrid LVQ learning algorithm is the modified Counter Propagation Networks(C.p Net. ) which is use SOM to learn initial reference vectors and out-star learning algorithm to determine the class of the output neurons of LVa. The weights of the proposed C.p. Net. which is between input layer and subclass layer can be learned to determine initial reference vectors by using SOM algorithm and to learn reference vectors by using LVd algorithm, and pattern vectors is classified into subclasses by neurons which is being in the subclass layer, and the weights which is between subclass layer and class layer of C.p. Net. is learned to classify the classified subclass. which is enclosed a class . To classify the pattern vectors of EMG. the proposed algorithm is simulated with ones of the conventional LVQ, and it was a confirmation that the proposed learning method is more successful classification than the conventional LVQ.

• The KIPS Transactions:PartA
• /
• v.11A no.3
• /
• pp.189-198
• /
• 2004

We consider the fault hamiltonian properties of m ${\times}$ n meshes with two wraparound links on the first row and the last row, denoted by M$_2$(m,n), (m$\geq$2, n$\geq$3). M$_2$(m,n), which is bipartite, with a single faulty link has a fault-free path of length mn-l(mn-2) between arbitrary two nodes if they both belong to the different(same) partite set. Compared with the previous works of P$_{m}$ ${\times}$C$_{n}$ , it also has these hamiltonian properties. Our result show that two additional wraparound links are sufficient for an m${\times}$n mesh to have such properties rather than m wraparound links. Also, M$_2$(m,n) is a spanning subgraph of many interconnection networks such as multidimensional meshes, recursive circulants, hypercubes, double loop networks, and k-ary n-cubcs. Thus, our results can be applied to discover fault-hamiltonicity of such interconnection networks. By applying hamiltonian properties of M$_2$(m,n) to 3-dimensional meshes, recursive circulants, and hypercubes, we obtain fault hamiltonian properties of these networks.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.4C
• /
• pp.451-457
• /
• 2006

A peer-to-peer(P2P) network is inherently vulnerable to malicious attacks from participating peers because of its open, flat, and autonomous nature. This paper addresses the problem of effectively defending from active attacks of malicious peers at bootstrapping phase and at online phase, respectively. We propose a secure membership handling protocol to protect the assignment of ID related things to a newly joining peer with the aid of a trusted entity in the network. The trusted entities are only consulted when new peers are joining and are otherwise uninvolved in the actions of the P2P networks. For the attacks in online phase, we present a novel message structure applied to each message transmitted on the P2P overlay. It facilitates the detection of message alteration, replay attack and a message with wrong information. Taken together, the proposed techniques deter malicious peers from cheating and encourage good peers to obey the protocol of the network. The techniques assume a basic P2P overlay network model, which is generic enough to encompass a large class of well-known P2P networks, either unstructured or not.