• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.221-224
• /
• 2019

딥러닝을 이용한 Neural Machine Translation(NMT)의 등장으로 기계번역 분야에서 기존의 규칙 기반,통계기반 방식을 압도하는 좋은 성능을 보이고 있다. 본 논문은 기계번역 모델도 중요하지만 무엇보다 중요한 것은 고품질의 학습데이터를 구성하는 일과 전처리라고 판단하여 이에 관련된 다양한 실험을 진행하였다. 인공신경망 기계번역 시스템의 학습데이터 즉 병렬 코퍼스를 구축할 때 양질의 데이터를 확보하는 것이 무엇보다 중요하다. 그러나 양질의 데이터를 구하는 일은 저작권 확보의 문제, 병렬 말뭉치 구축의 어려움, 노이즈 등을 이유로 쉽지 않은 상황이다. 본 논문은 고품질의 학습데이터를 구축하기 위하여 병렬 코퍼스 필터링 기법을 제시한다. 병렬 코퍼스 필터링이란 정제와 다르게 학습 데이터에 부합하지 않다고 판단되며 소스, 타겟 쌍을 함께 삭제 시켜 버린다. 또한 기계번역에서 무엇보다 중요한 단계는 바로 Subword Tokenization 단계이다. 본 논문은 다양한 실험을 통하여 한-영 기계번역에서 가장 높은 성능을 보이는 Subword Tokenization 방법론을 제시한다. 오픈 된 한-영 병렬 말뭉치로 실험을 진행한 결과 병렬 코퍼스 필터링을 진행한 데이터로 만든 모델이 더 좋은 BLEU 점수를 보였으며 본 논문에서 제안하는 형태소 분석 단위 분리를 진행 후 Unigram이 반영된 SentencePiece 모델로 Subword Tokenization를 진행 하였을 시 가장 좋은 성능을 보였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.11a
• /
• pp.181-184
• /
• 2003

가격대 성능 비를 고려할 때 저가격으로 병렬시스템을 제작할 수 있다는 특징으로, 시작된 클러스터시스템이 구성 장비의 특수화 및 전체 시스템의 대규모화로 인하여 더 이상 보조적인 소규모 시스템이 아닌, 슈퍼컴퓨터의 한 종류를 이루는 비중있는 시스템으로 자리매김하고있다. 이처럼 클러스터시스템 개발의 전반적인 방향이 대규모화를 지향하는 점을 고려할 때, 각 구성 요소의 무결성, 즉 안정성 점검은 시스템의 정상적인 운영을 위해서 중요한 부분이다. 본 논문에서는 클러스터시스템을 구성하는 각 계산노드의 메모리의 이상 유무을 관리 서버 측면에서 종합적으로 진단하기 위한 관리 도구의 개발을 다루고 있다.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.44 no.12
• /
• pp.1043-1053
• /
• 2016

In this paper, the application of Parallel Distributed Compensation (PDC) controller for fixed pitch rotary wing system was studied. For nonlinear modeling, T-S fuzzy model was utilized to advance system control including the tilt type UAV. PDC controller was designed through the Linear Matrix Inequality (LMI). Experiments for determining the applicability and feasibility of PDC were performed using the 1 axis attitude control equipment and simulation. To verify the performance and characteristics of the controller, Mathworks Co. Simulink was used. After then, the PDC controller performance was verified and the results with developed controller using a 1 axis attitude control equipment were compared. Verification of the feasibility of PDC controller for the fixed pitch rotary wing system and identification of the overall performance and improvement analysis was conducted based on the experimental results.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2000.10b
• /
• pp.1259-1262
• /
• 2000

반도체 제조장비로 구성된 통신 시스템은 일반적으로 직렬통신, 병렬통신, TCP/IP 통신 또는 별도의 개발 시스템을 사용한다. 기존의 단일장비 모니터링 시스템은 동일한 통신시스템을 사용하는 장비들의 그룹만을 모니터링 하는 시스템이다. 그러나 다양한 반도체 제조장비의 모니터링을 위해서는 통신 장비들의 추가 변경이 가능한 구조가 요구된다. 본 논문은 기존의 단일 장비 모니터링 시스템의 한계를 보완하여 직렬통신장비, 병렬통신장비, iQ 망 통신, TCP/IP 통신을 하는 반도체 제조장비의 통합 모니터링을 위한 통신 시스템 구조를 제시한다.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.1
• /
• pp.1-10
• /
• 2013

In this paper, we propose a CUDA(Common Unified Device Architecture) based SW(software) design method for CPU(Central Processing Unit) and GPU(Graphic Processing Unit) parallel structure to implement real-time process in 3D Laser ladar(LADAR) imaging system. LADAR is a complex system to generate 3-dimensional image based on the laser ranging information, and requires massive process resources in each phase. Therefore, designing and implementing parallel structure are crucial to realize a real-time process within limited system resource. As a conclusion, we can meet the speed of required real-time process allocating separable work load to CUDA GPU by analyzing process algorithm in each phase and confirm the process speed increase by 46%.

• Journal of KIISE:Databases
• /
• v.28 no.3
• /
• pp.418-429
• /
• 2001

Recently, a GIS has been applicable to the most important computer applications such as urban information systems and transportation information systems. These applications require spatial operations for an efficient management of a large volume of data. In particular, a spatial join among basic operations has the property that its response time is increased exponentially according to the number of spatial objects included in the operation. Therefore, it is not proper to the systems demanding the fast response time. To satisfy these requirements, the efficient parallel processing of spatial joins has been required. In this paper, the efficient method for creating and allocating tasks to balance statically the load of each processor in a parallel spatial join is presented. A task graph is developed in which a vertex weight is calculated by the cost model I have proposed. Then, it is partitioned through a graph partitioning algorithm. According to the experiments in CC16 parallel machine, our method made an improvement in the static load balance by decreasing the variance of a task execution time on each processor.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.9 no.1
• /
• pp.39-45
• /
• 2009

OFDM is used for achieving a high-speed data transmission in mobile wireless communication systems. Conventionally, fast Fourier transform that is the main signal processing of OFDM is implemented using digital signal processing. The DSP FFT LSI requires large power consumption. Current-mode FFT LSI with analog signal processing is one of the best solutions for high speed and low power consumption. However, for the operation of current-mode FFT LSI that has the structure of parallel-input and parallel-output, current-mode serial-to-parallel and parallel-to-serial converter are indispensable. We propose a novel current-mode SPC and PSC and full chip simulation results agree with experimental data. The proposed current-mode SPC and PSC promise the wide application of the current-mode analog signal processing in the field of low power wireless communication LSI.

• Journal of KIISE:Computing Practices and Letters
• /
• v.12 no.5
• /
• pp.286-299
• /
• 2006

Due to the continuing technical progress in the capabilities of modeling, simulation, and sensor devices, huge volume data with very high resolution are common. In scientific visualization, various interactive real-time techniques on high performance parallel computers to effectively render such large scale volume data sets have been proposed. In this paper, we present a mobile volume visualization system that consists of mobile clients, gateways, and parallel rendering servers. The mobile clients allow to explore the regions of interests adaptively in higher resolution level as well as specify rendering / viewing parameters interactively which are sent to parallel rendering server. The gateways play a role in managing requests / responses between mobile clients and parallel rendering servers for stable services. The parallel rendering servers visualize the specified sub-volume with rendering contexts from clients and then transfer the high quality final images back. This proposed system lets multi-users with PDA simultaneously share commonly interesting parts of huge volume, rendering contexts, and final images through CSCW(Computer Supported Cooperative Work) mode.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.10A
• /
• pp.776-783
• /
• 2009

Current-cut circuit is an effective method to obtain low power consumption in wireless communication systems as high speed OFDM. For the operation of current-mode FFT LSI with analog signal processing essentially requires current-mode serial-to-parallel/parallel-to-serial converter with multi input and output structure. However, the Hold-mode operation of current-mode serial-to-parallel/parallel-to-serial converter has unnecessary power consumption. We propose a novel current-mode serial-to-parallel/parallel-to-serial converter with current-cut circuit and full chip simulation results agree with experimental data of low power consumption. The proposed current-mode serial-to-parallel/parallel-to-serial converter promise the wide application of the current-mode analog signal processing in the field of low power wireless communication LSI.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.8 no.4
• /
• pp.815-820
• /
• 2004

The parallel operation of voltage source inverters using stator windings of high power three-phase induction motors was proposed in this paper. Most current high power induction motors with more than 4 electric poles have their external terminals installed so that windings of each phase can be approached from the outside. High power induction motors can be driven by parallel-operating several voltage source inverters through these external terminals. This way, in case a certain inverter breaks down, the operation torque will get decreased but the system can maintain its operation with the other inverters, so it can cope more effectively with breakdowns. Moreover, by providing phase difference to the switching movements of each inverter, it can increase equivalent switching frequency, which helps achieve good characteristics such as the reduction in the ripple of output torque, the reduction in the ripple of input current, and the reduction in the size of DC capacitors. Besides, since power is divided into each inverter, it can also decrease the ifluence of EMI occurring in the system. The characteristics of the proposed method were proved through computer simulations in this paper.