• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.12B
• /
• pp.1081-1090
• /
• 2003

NGI(Next Generation Internet) is characterized by QoS(Quality of Service) and high speed transmission. Recently, DiffServ and MPLS become most influential NGI architecture. To guarantee end-to-end QoS, it is essential for NGI to interwork MPLS with DiffServ. Here, MPLS WG(Working Group) in IETF(Internet Engineering Task Force) proposed the method of E-LSP(EXP inferred-PSC LSPs) and L-LSP(Label-Only-Inferred PSC LSPs), but both of them have serious problems to satisfy perfect interworking. In this paper, we proposed an extended E-LSP architecture that supports Perfect DiffServ class and experimental function in MPLS such as ECN(Explicit Congestion Notification) capability. We verify that the proposed E-LSP architecture improves QoS in NGI by using ns2 simulator.

• The KIPS Transactions:PartC
• /
• v.9C no.6
• /
• pp.893-902
• /
• 2002

MPLS enables efficient explicit routing, and thus provides great advantages in supporting traffic engineering. Exploiting this capability, we Propose a load balancing scheme which deploys a multipath routing. It is named LBM (Load Balancing in MPLS networks), and targets at efficient network utilization as well as performance enhancement. LBM establishes multiple LSP (Label Switched Path)s between a pair of ingress-egress routers, and distributes traffic over these LSPs at the new level. Its routing decision is based on both the length and the utilization of the paths. In order to enhance the efficiency of a link usage, a link is limited to be used by shorter paths as its utilization becomes higher Longer paths are considered to be candidate alternative paths as the utilization of shorter paths becomes higher. Simulation experiments are performed in order to compare the performance of LBM to that of static shortest path only scheme as well as the other representative dynamic multipath traffic distribution approaches. The simulation results show that LBM outperforms the compared approaches, and the performance gain is more significant when the traffic distribution among the ingress-egress pairs is non-uniform.

• Journal of Electrical Engineering and Technology
• /
• v.10 no.2
• /
• pp.699-707
• /
• 2015

Although digit character recognition has got a significant improvement in recent years, it is still challenging to achieve satisfied result if the data contains an amount of distracting factors. This paper proposes a novel digit character recognition approach using a multi-layer hierarchical model, Hybrid Restricted Boltzmann Machines (HRBMs), which allows the learning architecture to be robust to background distracting factors. The insight behind the proposed model is that useful high-level features appear more frequently than distracting factors during learning, thus the high-level features can be decompose into hybrid hierarchical structures by using only small label information. In order to extract robust and compact features, a stochastic 0-1 layer is employed, which enables the model's hidden nodes to independently capture the useful character features during training. Experiments on the variations of Mixed National Institute of Standards and Technology (MNIST) dataset show that improvements of the multi-layer hierarchical model can be achieved by the proposed method. Finally, the paper shows the proposed technique which is used in a real-world application, where it is able to identify digit characters under various complex background images.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2011.07a
• /
• pp.437-438
• /
• 2011

Spectral segmentation is a major trend in image segmentation. Specially, constrained spectral segmentation, inspired by the user-given inputs, remains its challenging task. Since it makes use of the spectrum of the affinity matrix of a given image, its overall quality depends mainly on how to design the graphical model. In this work, we propose a sparse, multi-layer graphical model, where the pixels and the over-segmented regions are the graph nodes. Here, the graph affinities are computed by using the must-link and cannot-link constraints as well as the likelihoods that each node has a specific label. They are then used to simultaneously cluster all pixels and regions into visually coherent groups across all layers in a single multi-layer framework of Normalized Cuts. Although we incorporate only the adjacent connections in the multi-layer graph, the foreground object can be efficiently extracted in the spectral framework. The experimental results demonstrate the relevance of our algorithm as compared to existing popular algorithms.

• Journal of KIISE
• /
• v.44 no.8
• /
• pp.822-831
• /
• 2017

In this paper, we propose a Korean dependency parsing model using multi-task learning based pointer networks. Multi-task learning is a method that can be used to improve the performance by learning two or more problems at the same time. In this paper, we perform dependency parsing by using pointer networks based on this method and simultaneously obtaining the dependency relation and dependency label information of the words. We define five input criteria to perform pointer networks based on multi-task learning of morpheme in dependency parsing of a word. We apply a fine-tuning method to further improve the performance of the dependency parsing proposed in this paper. The results of our experiment show that the proposed model has better UAS 91.79% and LAS 89.48% than conventional Korean dependency parsing.

• Journal of Digital Convergence
• /
• v.10 no.3
• /
• pp.39-49
• /
• 2012

The purpose of this study is to analyzed that the perceived price and VMD fitness of the SPA(speciality store retailer of Private label Apparel) brand characteristics impacts on multidimensional commitment consist of affective commitment, calculative commitment and behavioral commitment. Using SPSS 18.0 and AMOS 18.0 by structural equation modeling analysis, we found that perceived price have effects on affective commitment, behavioral commitment and calculative commitment significantly, and VMD fitness have effects affective commitment significant only. Affective commitment and calculative commitment on behavioral commitment was affected significantly. Through this study we found that perceived price and VMD fitness SPA brand maintaining relationships between the consumer and the seller are an important factors.

• The Journal of the Acoustical Society of Korea
• /
• v.17 no.7
• /
• pp.50-56
• /
• 1998

본 논문은 자동 음소 분할기의 음소 경계 오류를 보상하기 위한 후처리 (Postprocessing)에 관한 연구이다. 이는 현재 음성 합성을 위한 음성/언어학적 연구, 운율 모델링, 합성단위 자동 생성 연구 등에 대량의 음소 단위 분절과 음소 레이블링된 데이터의 필요성에 따른 연구의 일환이다. 특히 수작업에 의한 분절 및 레이블링은 일관성의 유지가 어렵고 긴 시간이 소요되므로 자동 분절 기술이 더욱 중요시 되고 있다. 따라서, 본 논문은 자동 분절 경계의 오류 범위를 줄일 수 있는 후처리기를 제안하여 자동 분절 결과를 직접 합성 단위로 사용할 수 있고 대량의 합성용 운율 데이터 베이스 구축에 유용함을 기술한다. 제안된 후처리기는 수작업으로 조정된 데이터의 특징 벡터를 다층 신경회로망 (MLP:Multi-layer perceptron)을 통해 학습을 한 후, ETRI(Electronics and Telecommunication Research Institute)에서 개발된 음성 언어 번역 시스템을 이용한 자동 분절 결과와 후처리기인 MLP를 이용하여 새로운 음소 경계를 추출한다. 고립단어로 발성된 합성 데이터베이스에서 후처리기로 보정된 분절 결과는 음성 언어 번역 시스템의 분할율보 다 약 25%의 향상된 성능을 보였으며, 절대 오류(｜Hand label position-Auto label position ｜)는 약 39%가 향상되었다. 이는 MLP를 이용한 후처리기로 자동 분절 오류의 범위를 줄 일 수 있고, 대량의 합성용 운율 데이터 베이스 구축 및 합성 단위의 자동생성에 이용될 수 있음을 보이는 것이다.

• Journal of Communications and Networks
• /
• v.10 no.3
• /
• pp.351-361
• /
• 2008

This paper presents a delay-margin based traffic engineering (TE) approach to provide end-to-end quality of service (QoS) in multi-protocol label switching (MPLS) networks using differentiated services (DiffServ) at the link level. The TE, including delay, class, and route assignments, is formulated as a nonlinear optimization problem reflecting the inter-class and inter-link dependency introduced by DiffServ and end-to-end QoS requirements. Three algorithms are used to provide a solution to the problem: The first two, centralized offline route configuration and link-class delay assignment, operate in the convex areas of the feasible region to consecutively reduce the objective function using a per-link per-class decomposition of the objective function gradient. The third one is a heuristic that promotes/demotes connections at different links in order to deal with concave areas that may be produced by a trunk route usage of more than one class on a given link. Approximations of the three algorithms suitable for on-line distributed TE operation are also derived. Simulation is used to show that proposed approach can increase the number of users while maintaining end-to-end QoS requirements.

• The Journal of the Korea Contents Association
• /
• v.8 no.9
• /
• pp.34-41
• /
• 2008

Existing mobility support scheme like MIPv4 has the delay at data transfer and Triangle routing problem. This paper proposes the new mobility support scheme and method that can overcome the problem using MPLS LSP in Next Generation Network. The proposed scheme overcomes the problem by introducing the LMS(Location Management Server) managing the location of the mobile terminal and the MPLS LSP between FA for a tunneling. We shows on the mathematical analysis using the queuing theory that this proposed scheme is more superior to the existing method(MIP).

• ETRI Journal
• /
• v.27 no.6
• /
• pp.733-746
• /
• 2005

In the context of multi-protocol label switching (MPLS) traffic engineering, this paper proposes a scalable constraintbased shortest path first (CSPF) routing algorithm with multiple QoS metrics. This algorithm, called the multiple constraint-based shortest path first (M_CSPF) algorithm, provides an optimal route for setting up a label switched path (LSP) that meets bandwidth and end-to-end delay constraints. In order to maximize the LSP accommodation probability, we propose a link weight computation algorithm to assign the link weight while taking into account the future traffic load and link interference and adopting the concept of a critical link from the minimum interference routing algorithm. In addition, we propose a bounded order assignment algorithm (BOAA) that assigns the appropriate order to the node and link, taking into account the delay constraint and hop count. In particular, BOAA is designed to achieve fast LSP route computation by pruning any portion of the network topology that exceeds the end-to-end delay constraint in the process of traversing the network topology. To clarify the M_CSPF and the existing CSPF routing algorithms, this paper evaluates them from the perspectives of network resource utilization efficiency, end-to-end quality, LSP rejection probability, and LSP route computation performance under various network topologies and conditions.