• Journal of Korea Multimedia Society
• /
• v.25 no.11
• /
• pp.1643-1652
• /
• 2022

Most of the existing machine reading research has used Recurrent Neural Network (RNN) and Convolutional Neural Network (CNN) algorithms as networks. Among them, RNN was slow in training, and Question Answering Network (QANet) was announced to improve training speed. QANet is a model composed of CNN and self-attention. CNN extracts semantic and syntactic information well from the local corpus, but there is a limit to extracting the corresponding information from the global corpus. Graph Convolutional Networks (GCN) extracts semantic and syntactic information relatively well from the global corpus. In this paper, to take advantage of this strength of GCN, we propose I-QANet, which changed the CNN of QANet to GCN. The proposed model performed 1.2 times faster than the baseline in the Stanford Question Answering Dataset (SQuAD) dataset and showed 0.2% higher performance in Exact Match (EM) and 0.7% higher in F1. Furthermore, in the Korean Question Answering Dataset (KorQuAD) dataset consisting only of Korean, the learning time was 1.1 times faster than the baseline, and the EM and F1 performance were also 0.9% and 0.7% higher, respectively.

• Journal of Korean Society of societal Security
• /
• v.2 no.3
• /
• pp.47-53
• /
• 2009

This study performed the construction of test bed for the combination of the health monitoring system In Yongjong grand bridge with the safety management network system in order to explore the feasibility of combining the existing monitoring system with the safety management network system. For this purpose, connection program which consists of DAQ sender server program and DAQ receiver server program has been developed. DAQ sender server program installed in the connection server, which is located on the DMZ zone of the existing monitoring system, plays the role of communicating with the monitoring system. On the other hand, DAQ receiver server program installed in the connection server, which is located on the integrated operation center of the safety management network system, plays the part of gathering real time and tim history data of the existing monitoring system and storing them in the integrated operation server through the communication with the DAQ sender. Test bed demonstrated the feasibility of combining the existing monitoring system with the safety management network system.

• The KIPS Transactions:PartA
• /
• v.13A no.6 s.103
• /
• pp.533-540
• /
• 2006

This paper describes the design and implementation of a grid system META (Metacomputing Environment using Test-run of Application) which facilitates the execution of a CFD (Computational Fluid Dynamics) analysis program on distributed environment. The grid system META allows the CFD program developers can access the computing resources distributed over the network just like one computer system. The research issues involved in the grid computing include fault-tolerance, computing resource selection, and user-interface design. In this paper, we exploits an automatic resource selection scheme for executing the parallel SPMD (Single Program Multiple Data) application written in MPI (Message Passing Interface). The proposed resource selection scheme is informed from the network latency time and the elapsed time of the kernel loop attained from test-run. The network latency time highly influences the executional performance when a parallel program is distributed and executed over several systems. The elapsed time of the kernel loop can be used as an estimator of the whole execution time of the CFD Program due to a common characteristic of CFD programs. The kernel loop consumes over 90% of the whole execution time of a CFD program.

• International Journal of Aeronautical and Space Sciences
• /
• v.10 no.2
• /
• pp.34-42
• /
• 2009

Gas turbine performance diagnostics is a method for detecting, isolating and quantifying faults in gas turbine gas path components. On-line precise fault diagnosis can promote greatly reliability and availability of gas turbine in real time operation. This work proposes a GUI-type on-line diagnostic program using SIMULINK and Fuzzy-Neuro algorithms for a helicopter turboshaft engine. During development of the diagnostic program, a look-up table type base performance module are used for reducing computer calculating time and a signal generation module for simulating real time performance data. This program is composed of the on-line condition monitoring program to monitor on-line measuring performance condition, the fuzzy inference system to isolate the faults from measuring data and the neural network to quantify the isolated faults. Evaluation of the proposed on-line diagnostic program is performed through application to the helicopter engine health monitoring.

• Journal of Korea Game Society
• /
• v.1 no.1
• /
• pp.61-67
• /
• 2001

This paper describes a program which learns good strategies for two-poison, deterministic, zero-sum board games of perfect information. The program learns by simply playing the game against either a human or computer opponent. The results of the program's teaming of a lot of games are reported. The program consists of search kernel and a move generator module. Only the move generator is modified to reflect the rules of the game to be played. The kernel uses a temporal difference procedure combined with a backpropagation neural network to team good evaluation functions for the game being played. Central to the performance of the program is the search procedure. This is a the capture tree search used in most successful janggi playing programs. It is based on the idea of using search to correct errors in evaluations of positions. This procedure is described, analyzed, tested, and implemented in the game-teaming program. Both the test results and the performance of the program confirm the results of the analysis which indicate that search improves game playing performance for sufficiently accurate evaluation functions.

• KIEE International Transactions on Power Engineering
• /
• v.3A no.3
• /
• pp.139-147
• /
• 2003

The paper presents a novel approach of constructing the state matrix of the multi-machine power system for SSR (subsynchronous resonance) analysis using the linearized equations of individual devices including electrical transmission network dynamics. The machine models in the local d-q reference frame are integrated with the network models in the common R-I reference frame by simply transforming their output equations into the R-I frame where the transformed output is used as the input to the network dynamics or vice versa. The salient feature of the formulation is that it allows for modular construction of various component models without rearranging the overall state space formulation. This universal SSR small signal stability program provides a flexible tool for systematic analyses of SSR small-signal stability impacts of both conventional devices such as generation systems and novel devices such as power electronic apparatus and their controllers. The paper also presents its application results to IEEE benchmark models.

• Journal of Information Processing Systems
• /
• v.9 no.2
• /
• pp.259-270
• /
• 2013

A speaker's intentions can be represented by domain actions (domain-independent speech act and domain-dependent concept sequence pairs). Therefore, it is essential that domain actions be determined when implementing dialogue systems because a dialogue system should determine users' intentions from their utterances and should create counterpart intentions to the users' intentions. In this paper, a neural network model is proposed for classifying a user's domain actions and planning a system's domain actions. An integrated neural network model is proposed for simultaneously determining user and system domain actions using the same framework. The proposed model performed better than previous non-integrated models in an experiment using a goal-oriented dialogue corpus. This result shows that the proposed integration method contributes to improving domain action determination performance.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.20 no.5
• /
• pp.1426-1435
• /
• 1996

In this paper. a network theory for generaltransmission systme was extended considering the direction of power flow. Also, a modified network model was suggested for a node with 4 shafts in order to verify the power flow. Based on the extended network theory, a simulation program was developed to analyze a hydro-mecaanical tranmission(HMT) system consistion of two hydrostatic pump motors, severeal planetary gear trains steer differential gear. The simulation result showed that the extendednotwork analysis program develped can predict the power circulation as well as the magnitude of torque and speed for each transmission element and can be used design tool for genaral power transmission system.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 1996.04a
• /
• pp.89-92
• /
• 1996

It is generally recognized that Video On Demand (VOD) service will become a promising interactive service in B-ISDN. We consider a hierarchical two level network architecture for VOD service. The first level consists of connected Local Video Servers(LVSs) in which a limited number of high vision programs are stored. The second level has one Central Video Server(CVS) containing all the programs served in the network. The purpose of this paper is to analyze the network traffic and to propose the storage policy for LVSs. For this purpose, we present an analysis of program storage amount in each LVS, transmission traffic volume between LVSs, and link traffic volume between CVS and LVSs, according to changing the related factors such as demand, the number of LVSs is also presented on the basis of the tradeoffs among program storage cost, link traffic cost, and transmission cost.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.11C
• /
• pp.1040-1045
• /
• 2002

This paper focuses on a hash-based, distributed Wet caching network that eliminates inter-cache communication. An agent program on cache servers, a mapping program on the DNS server, and other components comprised in a distributed Web caching network were modified and developed to implement a so-called "consistent" hashing. Also, a dynamic load balancing algorithm is proposed to address the load-balancing problem that is a key performance issue on distributed architectures. This algorithm effectively balances the load among cache servers by distributing the calculated amount of mapping items that have higher popularity than others. Therefore, this developed network can resolve the imbalanced load that is caused by a variable page popularity, a non-uniform distribution of a hash-based mapping, and a variation of cache servers.