• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.453-458
• /
• 2018

본 논문은 한국어 자소를 인식 단위로 사용한 hybrid CTC-Attention 모델 기반 end-to-end speech recognition을 제안한다. End-to-end speech recognition은 기존에 사용된 DNN-HMM 기반 음향 모델과 N-gram 기반 언어 모델, WFST를 이용한 decoding network라는 여러 개의 모듈로 이루어진 과정을 하나의 DNN network를 통해 처리하는 방법을 말한다. 본 논문에서는 end-to-end 모델의 출력을 추정하기 위해 자소 단위의 출력구조를 사용한다. 자소 기반으로 네트워크를 구성하는 경우, 추정해야 하는 출력 파라미터의 개수가 11,172개에서 49개로 줄어들어 보다 효율적인 학습이 가능하다. 이를 구현하기 위해, end-to-end 학습에 주로 사용되는 DNN 네트워크 구조인 CTC와 Attention network 모델을 조합하여 end-to-end 모델을 구성하였다. 실험 결과, 음절 오류율 기준 10.05%의 성능을 보였다.

• Journal of the Korea Society of Computer and Information
• /
• v.22 no.3
• /
• pp.131-138
• /
• 2017

New drug development is time-consuming and costly. Hence, it is necessary to repurpose old drugs for finding new indication. We suggest the way that repurposing old drug using massive literature data and biological network. We supposed a disease-drug relationship can be available if signal pathways of the relationship include significant genes identified in literature data. This research is composed of three steps-identifying significant gene using co-occurrence in literature; analyzing the shortest path on biological network; and scoring a relationship with comparison between the significant genes and the shortest paths. Based on literatures, we identify significant genes based on the co-occurrence frequency between a gene and disease. With the network that include weight as possibility of interaction between genes, we use shortest paths on the network as signal pathways. We perform comparing genes that identified as significant gene and included on signal pathways, calculating the scores and then identifying the candidate drugs. With this processes, we show the drugs having new possibility of drug repurposing and the use of our method as the new method of drug repurposing.

• International Journal of Computer Science & Network Security
• /
• v.21 no.8
• /
• pp.111-118
• /
• 2021

Banking systems are sensitive to data privacy since users' data, if not well protected, may be used to perform fake transactions. Blockchains, public and private, are frequently used in such systems thanks to their efficiency and high security. Public blockchains fail to fully protect users' data, despite their power in the accuracy of the transactions. The private blockchain is better used to protect the privacy of the sensitive data. They are not open and they apply authorization to login into the blockchain. However, they have a lower security compared to public blockchain. We propose in this paper a hybrid public-private architecture that profits from network virtualization. The main novelty of this proposal is the use of network virtualization that helps to reduce the complexity and efficiency of the computations. Simulations have been conducted to evaluate the performance of the proposed solution. Findings prove the efficiency of the scheme in reducing complexity and enhancing data privacy by guarantee high security. The contribution conducted by this proposal is that the results are verified by the centralized controller that ensures a correct validation of the resulted blockchains. In addition, computation complexity is to be reduced by profiting from the cooperation performed by the virtual agents.

• International Journal of Computer Science & Network Security
• /
• v.21 no.3
• /
• pp.212-218
• /
• 2021

Low-power and Lossy Networks (LLNs) have become the common network infrastructure for a wide scope of Internet of Things (IoT) applications. For efficient routing in LLNs, IETF provides a standard solution, namely the IPv6 Routing Protocol for LLNs (RPL). It enables effective interconnectivity with IP networks and flexibly can meet the different application requirements of IoT deployments. However, it still suffers from different open issues, particularly in large-scale setups. These include the node unreachability problem which leads to increasing routing losses at RPL sink nodes. It is a result of the event of memory overflow at LLNs devices due to their limited hardware capabilities. Although this can be alleviated by the establishment of multi-sink topologies, RPL still lacks the support for effective load balancing among multiple sinks. In this paper, we address the need for an efficient multi-sink load balancing solution to enhance the performance of PRL in large-scale scenarios and alleviate the node unreachability problem. We propose a new RPL objective function, Multi-Sink Load Balancing Objective Function (MSLBOF), and introduce the Memory Utilization metrics. MSLBOF enables each RPL node to perform optimal sink selection in a way that insure better memory utilization and effective load balancing. Evaluation results demonstrate the efficiency of MSLBOF in decreasing packet loss and enhancing network stability, compared to MRHOF in standard RPL.

• Journal of The Institute of Information and Telecommunication Facilities Engineering
• /
• v.1 no.1
• /
• pp.26-39
• /
• 2002

This paper describes design and implementation of home network control protocol for networked home appliances. The proposed network protocol has four-layered protocol structure and device-modem interface structure for the flexibility of modems based on power line communication. The standard message set is specified to guarantee the interoperability between various home appliances The proposed protocol can be easily implemented because it has minimum network overhead.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2011.06a
• /
• pp.367-368
• /
• 2011

본 논문에서는 PON(Passive Optical Network)망의 장애진단을 위한 임베디드 네트워크 제어기를 설계하고 구현하였다. 본 연구에서 구현한 임베디드 네트워크 제어기는 PON망의 가장 말단인 수용가의 인터넷 공유기에 부착되어 수용가의 인터넷 선로장애를 진단할 수 있도록 구현함으로서 인터넷 서비스 제공자(ISP)의 NMS가 점검할 수 없는 영역까지 장애를 진단할 수 있다. 또한, 임베디드 네트워크 제어기는 PON망의 장애진단 뿐만 아니라 수용가의 가전제품 전원제어나 다양한 센서를 부착하여 제어할 수 있도록 제작하여 간단한 홈오토메이션 제어기로 활용할 수 있도록 설계하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.05a
• /
• pp.1433-1436
• /
• 2004

The ad hoc network is a collection of wireless mobile nodes dynamically forming a temporary network without the use of any existing network infrastructure of centralized administration. Load-Balanced Ad hoc Routing(LBAR) protocol is an on-demand routing protocol intended for delay-sensitive applications where users are most concern with packet transmission delay. Although LBAR mechanism is a novel load balancing routing protocol for ad hoc network, it has own limitation in route path maintenance phase. Therefore, in this paper, we propose Advanced Load-Balanced Ad hoc Routing(A-LBAR) that is delay-sensitive and has an efficient path maintenance scheme. The robust path maintenance scheme is maintained by considering about nodal loads all over network and misbehavior of overloaded or selfish nodes. The proposed scheme provides good performance over DSR and AODV in terms of packet delay and packet loss rate when some misbehaving nodes exist in the network.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.7 no.1
• /
• pp.22-37
• /
• 2013

Content distribution to a large number of concurrent clients stresses both server and network. While the server limitation can be circumvented by deploying server clusters, the network limitation is far less easy to cope with, due to the difficulty in measuring and balancing network load. In this paper, we use two useful network load metrics, the worst link stress (WLS) and the degree of interference (DOI), and formulate the problem as partitioning the clients into disjoint subsets subject to the server capacity constraint so that the WLS and the DOI are reduced for each session and also well balanced across the sessions. We present a network load-aware partition algorithm, which is practicable and effective in achieving the design goals. Through experiments on PlanetLab, we show that the proposed scheme has the remarkable advantages over existing schemes in reducing and balancing the network load. We expect the algorithm and performance metrics can be easily applied to various Internet applications, such as media streaming, multicast group member selection.

• Journal of the Korea Society of Computer and Information
• /
• v.24 no.10
• /
• pp.205-213
• /
• 2019

Network robustness is one of the most important characteristics needed as the network. Over the military tactical communication network, robustness is a key function for maintaining attack phase constantly. Tactical Information Communication Network, called TICN, has mixed characteristics of lattice- and tree-type network topology, which looks somewhat weak in the viewpoint of network robustness. In this paper, we search articulation points and bridges in a current Tactical Information Communication Network using graph theory. To improve the weak points empirically searched, we try to add links to create the concrete network and then observe the change of network-based verification values through diminishing nodes. With these themes, we evaluate the generated networks through SNA techniques. Experimental results show that the generated networks' robustness is improved compared with current network structure.

• Proceedings of the Korea Contents Association Conference
• /
• 2009.05a
• /
• pp.8-11
• /
• 2009

NAC(Network Access Control) technology is intended for authentication of internal networks access through various paths. In this paper, we build computer network platform using NAC and Extreme switch and confirm authentication for the platform. The platform consists of PIX, NAC and authentication server.