• International Journal of Contents
• /
• 제14권4호
• /
• pp.24-29
• /
• 2018

EEG-based brain-computer interfaces has focused on explicitly expressed intentions to assist physically impaired patients. For EEG-based-computer interfaces to function effectively, it should be able to understand users' implicit information. Since it is hard to gather EEG signals of human brains, we do not have enough training data which are essential for proper classification performance of implicit intention. In this paper, we improve the subject independent classification of implicit intention through the generation of additional training data. In the first stage, we perform the PCA (principal component analysis) of training data in a bid to remove redundant components in the components within the input data. After the dimension reduction by PCA, we train ICA (independent component analysis) network whose outputs are statistically independent. We can get additional training data by adding Gaussian noises to ICA outputs and projecting them to input data domain. Through simulations with EEG data provided by CNSL, KAIST, we improve the classification performance from 65.05% to 66.69% with Gamma components. The proposed sample generation method can be applied to any machine learning problem with fewer samples.

• International Journal of Computer Science & Network Security
• /
• 제22권6호
• /
• pp.127-130
• /
• 2022

The number of published journals, conferences, and research papers in computer science is increasing rapidly, which has led to a challenge in coming up with new and unique ideas for research. To alleviate the issue, this paper uses artificial neural networks (ANNs) to generate new computer science research ideas. It does so by using a dataset collected from IEEE published journals and conferences to train an ANN model. The results reveal that the model has a 14% success rate in generating usable ideas. The outcome of this paper has implications for helping both new and experienced researchers come up with novel research topics.

• 한국컴퓨터그래픽스학회논문지
• /
• 제29권3호
• /
• pp.79-84
• /
• 2023

이미지 생성 모델은 다양한 분야에 적용되어 데이터 부족 문제와 시간 및 비용 문제를 극복하고 있다. 그러나 규칙적 패턴 이미지에서의 이미지 생성과 해당 데이터의 결함 검출에는 한계를 가진다. 본 논문에서는 이미지 생성 모델의 패턴 이미지 생성의 가능성을 확인하고 OLED 패널의 결함 검출을 위한 데이터 증강에 적용하였다. OLED 결함 검출 모델을 학습하기 위해 필요한 데이터는 OLED 패널의 높은 비용 문제로 실제 데이터 세트를 확보하기 어렵다. 그렇기 때문에 해당 데이터 세트를 확보한다 하더라도 여러가지 결함 유형을 정의하고 분류하는 작업이 필요하다. 이를 위한 가상의 기반 데이터 세트를 획득할 OLED 패널 결함 데이터 획득 시스템을 소개하고, 이미지 생성 모델로 해당 데이터를 증강한다. 또, 확산모델에서의 패턴 이미지 생성의 어려움을 확인하여 가능성을 제안하고, 이미지 생성 모델 이용한 데이터 증강 및 결함 검출 데이터 증강의 제한 사항을 개선하였다.

• 한국추진공학회:학술대회논문집
• /
• 한국추진공학회 2008년 영문 학술대회
• /
• pp.612-621
• /
• 2008

For the purpose of improving accuracy in jet noise prediction and investigating its generation mechanism, high subsonic jets were computed by using compressible Large Eddy Simulation(LES), wherein the inflow forcing or disturbance added in the inflow shear layer was incorporated. The far-field Sound Pressure Levels(SPL) as well as the flow field resulted in good agreement with available experimental data by applying only the high azimuthal modes among the inflow forcing parameters. We found that this result was due to an important role of the inflow forcing upon breaking down the axiymmetric vortices that caused high amplitude velocity and pressure fluctuations. In order to examine generation mechanism of the dominant noise component, wavelet transformation was introduced to reveal the presence of a well-organized structure of pressure fluctuations that originated mainly from vortex motions near the end of the jet potential core. This structure took a train of alternately positive and negative wavelet-transformed pressure regions along the jet distance, spreading towards the downstream with advection and propagation. It was concluded that this structure and its dynamic motion are the reason why a high subsonic jet produces the dominant noise with a particular downstream directivity.

• 정보보호학회논문지
• /
• 제32권2호
• /
• pp.381-390
• /
• 2022

기계학습 이미지 인식 기술의 발전에 따라 이를 악성코드 검출에 적용하는 방법이 연구되고 있다. 그 대표적인 접근법으로 악성코드 파일을 이미지로 변환하고 이를 CNN과 같은 딥러닝 네트워크에 학습시켜 악성코드 검출과 분류를 수행하는 연구가 진행되어 의미 있는 결과가 발표되고 있다. 본 연구에서는 기계학습을 사용한 악성코드 검출에 효과적인 이미지 생성방법을 제시하고자 한다. 이를 위하여 이미지 생성의 여러 선택 요소에 따른 악성코드 검출의 성능을 실험하고 분석하였으며, 그 결과를 반영하여 명령어 흐름의 특성을 좀 더 명확하게 나타낼 수 있는 선형적 이미지 생성방법을 제시하고 이 방법이 악성코드 검출의 정밀도를 높일 수 있음을 실험을 통하여 보였다.

• 대한기계학회논문집A
• /
• 제35권5호
• /
• pp.567-574
• /
• 2011

알미늄 압출재는 고속열차의 경량화를 위하여 기존의 주름강판을 대체하여 널리 사용된다. 알미늄 압출재는 고속열차 적층재 가운데 가장 큰 차음 기여도를 보이나, 동일한 중량의 평판과 비교할 때, 국부공진 주파수 대역에서 투과손실이 크게 떨어진다. 이 연구에서는 차세대 400km/h급 고속철도 차량용 알미늄 압출재를 대상으로 차음 문제를 검토하고, 차음성능의 향상 방안을 제시한다. 코어 구조를 변경시켜 국부공진 대역을 높이고, 우레탄 폼을 코어에 충진시킬 때의 차음성능 향상효과를 실험적으로 확인한다. 최종적으로 제시된 방법이 바닥 적층재의 총 투과손실을 어느 정도 개선시키는가를 평가한다.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2011년도 정기총회 및 추계학술대회 논문집
• /
• pp.556-562
• /
• 2011

Railroad started from invention of stream locomotive at the beginning of the industrial revolution. Development of diesel and electric locomotives in railroad induced science and technology since the industrial revolution. Korean high-speed train so-called high-speed train project, is localized 86.7%(compared to price), 92%(compared to number of parts). This technology has been contributed to KTX since 2004. High-speed railway was developed to integrated system, which is contained high technology such as civil, mechanical and electrical technology. This effects to advanced technology across the board science, technology, industry. High-speed railway is fast, safe, comfortable, eco-friendly as next generation's representative transportation. Furthermore it is inaugurate the era of the new railroad. Korea is opened the high-speed railway following Japan, France, Germany, Spain. Korea became the fifth nation of the high-speed railway. It joins the rank of developed countries operating at the speed of 300km/h. Especially, KTX is emerging as advanced transportation with reducing logistics costs significantly due to time advantage and transportation revolution. The purpose of this study is to present the strategy through internal and external railway market's analysis for Korea railway system's globalization expansion(overseas expansion).

• 소음진동
• /
• 제10권3호
• /
• pp.444-450
• /
• 2000

Wheel-rail noise is normally classified into three catagories : rolling impact and squeal noise. In this paper rolling noise caused by the irregularity between a wheel and a rail is analysed as follows: The irregularity between the wheel and the rail is assumed as linear superposition of sinusoidal profiles. Wheel-rail contact stiffness is linearized by using Hertzian contact theory and then contact force between the wheel and the rail is calculated. vibration of the rail and the wheel is calculated theoretically by receptance method or FEM depending on the geometry of the wheel or the rail for the frequency range of 100-500 Hz important for noise generation. The radiation noise caused by those vibration response is computed by BEM To verify this analysis tools rolling noise is calculated by proposed analysis steps using typical roughness data and these results are compared with experimental rolling noise data. This analysis tools show reasonable results and finally used for the prediction of the Korean high speed train rolling noise.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2003년도 춘계학술대회 논문집
• /
• pp.737-741
• /
• 2003

The wind power system is spotlighted as one of the no-pollution power generation systems. The system uses winds as power source that are rotated the blade and the rotating power from blade generate the electricity power. Gearbox needs to transfer the wind powers that have the high-torque-low-speed characteristics to generator that have the low-torque-high-speed characteristics. Because the wind power system generally locates the remote place like seaside or mountainside and the gearbox installs on the limited and high placed space, the gearbox of the wind power system is required the optimal space design and high reliability. In this paper, the structure of the gearbox is proposed to achieve the optimal space and efficiency by compounding the planetary gear train that has the high power density and parallel type gear train that has the long service life. The design parameters that are affected the service life are studied. The gear ratio and face width are investigated as an affected parameter for design sensitivity of service life.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2007년도 춘계학술대회 논문집
• /
• pp.1121-1128
• /
• 2007

Subway routes $1{\sim}4$ were constructed in gravel roadbed track structure in consideration of the technological capabilities, construction cost and duration at the time of the initial construction. As such, 224.8km, approximately 81.2% of entire length, of total length of railway track at 276.9km was constructed on gravel roadbed. However, improvement of gravel roadbed to concreted roadbed began in 1998 due to problems including frequent repair works and limited time application for such works caused by occurrence of tract distortion during operation as well as lowering of roadbed functions and generation of dust caused by frictional power, impact absorption capabilities, abrasion and crushing of gravel on roadbed. Currently, this improvement is continuing with target of converting entire route into concreted roadbed structure. Therefore, this Study modifies formula for setting cant, analyze the correlation between wearing of rail side of the curvature and cant insufficiency following increasing of the running speed of the train, and to present the directions for fundamental review for adjustment of cant insufficiencies at the time of improvement of gravel roadbed to concreted roadbed that is being implemented on the operational tracks of the railway trains.