• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.14 no.2
• /
• pp.534-539
• /
• 2010

In genomic studies, thousands of features are collected on relatively few samples. One of the goals of these studies is to build classifiers to predict the outcome of future observations. There are three inherent steps to build classifiers: a significant gene selection, model selection and prediction assessment. In the paper, with a focus on prediction assessment, we normalize microarray data with quantile-normalization methods that adjust quartile of all slide equally and then design a system comparing several methods to estimate 'true' prediction error of a prediction model in the presence of feature selection and compare and analyze a prediction error of them. LOOCV generally performs very well with small MSE and bias, the split sample method and 2-fold CV perform with small sample size very pooly. For computationally burdensome analyses, 10-fold CV may be preferable to LOOCV.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.3
• /
• pp.273-278
• /
• 2004

In this paper, the recurrent neural networks (RNN) is applied to compensate for HMM recognition algorithm, which is commonly used as main recognizer. Among these recurrent neural networks, the multi-layer recurrent neural prediction model (MRNPM), which allows operating in real-time, is used to implement learning and recognition, and HMM and MRNPM are used to design a hybrid-type main recognizer. After testing the designed speech recognition algorithm with Korean number pronunciations (13 words), which are hardly distinct, for its speech-independent recognition ratio, about 5% improvement was obtained comparing with existing HMM recognizers. Based on this result, only optimal (recognition) codes were extracted in the actual DSP (TMS320C6711) environment, and the embedded speech recognition system was implemented. Similarly, the implementation result of the embedded system showed more improved recognition system implementation than existing solid HMM recognition systems.

• Journal of the Korean Geotechnical Society
• /
• v.38 no.8
• /
• pp.39-52
• /
• 2022

Multilayer perceptron neural network was trained to determine the factor of safety and slip surface of the slope. Slope geometry is a simple slope based on Korean design standards, and the case of dry and existing groundwater levels are both considered, and the properties of the soil composing the slope are considered to be sandy soil including fine particles. When curating the data required for model training, slope stability analysis was performed in 42,000 cases using the limit equilibrium method. Steady-state seepage analysis of groundwater was also performed, and the results generated were applied to slope stability analysis. Results show that the multilayer perceptron model can predict the factor of safety and failure arc with high performance when the slope's physical properties data are input. A method for quantitative validation of the model performance is presented.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.9 no.4
• /
• pp.103-112
• /
• 1989

A numerical model for Steady state groundwater flow has been established to understand the groundwater flow phenomena near a radioactive waste repository. The integrated finite difference method based on a network composed of nodes and members was applied to investigate groundwater flow in homogeneous, heterogeneous and layered media. Its numerical solution was in good agreement with analytic solution. Physical phenomena associated in the groundwater flow depending on both hydraulic characteristics and effects of fractured zone were also investigated. A method by which feasible groundwater flow paths can be identified was developed. This method used the composite network for the geologic media near a repository and the direction of computed groudwater velocity. Groundwater velocity and travel time were predicted for the possible pathway form a repository to a biosphere.

• Journal of Welding and Joining
• /
• v.11 no.3
• /
• pp.34-47
• /
• 1993

Finite element models were developed for thermal and residual stress analysis for the specific welding problems. They were used to evaluate the effectiveness of the various welding heat input models, such as ramp heat input function and lumped pass models. Through the parametric studies, thermal-mechanical modeling sensitivity to the ramp function and lumping techniques was determined by comparing the predicted results with experimental data. The kinetics for residual stress formation during welding can be developed by iteration of various proposed mechanisms in the parametric study. A ramp heat input function was developed to gradually apply the heat flux with variable amplitude to the model. This model was used to avoid numerical convergence problems due to an instantaneous increase in temperature near the fusion zone. Additionally, it enables the model to include the effect of a moving arc in a two-dimensional plane. The ramp function takes into account the variation in the out of plane energy flow in a 2-D model as the arc approaches, travels across, and departs from each plane under investigation. A lumped pass model was developed to reduce the computation cost in the analysis of multipass welds. Several weld passes were assumed as one lumped pass in this model. Recommendations were provided about ramp lumping techniques and the optimum number of weld passes that can be combined into a single thermal input.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.12 no.3
• /
• pp.281-292
• /
• 1999

지반-기초 상호작용계를 해석할 때 실제로 지반은 다양한 지반종류와 다층으로 형성되어 있으므로 지반 특성의 변화를 고려해야 한다. 초기의 대부분의 상호작용계의 정·동적 해석은 지반의 복잡한 성질을 역학적으로 탄성거동을 한다고 가정한 Winkler 지반모델 혹은 지반을 등방성이고 균질한 반무한 탄성체로 가정한 반무한 탄성지반 모델로 보아 수행되었다. 본 연구는 유한 요소법을 이용하여 지반-기초 상호작용계의 동적 거동을 해석하기 위해 기초는 4절점 후판요소를 사용하고 지반은 지반특성을 고려할 수 있도록 8절점 6면체 요소를 사용하였고, 지반의 감쇠효과 및 지반특성을 고려한 지반-기초 상호작용계의 동적 거동을 유한요소법으로 해석하고 지반의 영향범위를 결정하는 것이다.

• IE interfaces
• /
• v.10 no.1
• /
• pp.37-45
• /
• 1997

본 연구에서는 웨이퍼 팹공정에서 웨이퍼 로트들이 뱃치공정을 위해 확률적으로 도착되는 상황에서 최적 뱃치크기를 결정하는 뱃칭문제를 다루고 있다. 뱃치공정이란 여러 개의 웨이퍼 로트들을 기계의 용량을 넘지 않는 한도 내에서 하나의 뱃치로 구성하여 한꺼번에 가공하는 공정을 말한다. 목적함수는 생산율을 높이고 재공재고 및 사이클타임을 줄이기 위해 웨이퍼 로트들의 평균 대기시간의 최소화를 채택하였다. 문제의 해결을 위해서, 확률적인 상황변동 하에서 실시간 제어를 위해 많이 활용되고 있는 신경회로망 중 다층 퍼셉트론을 이용한 뱃치크기 결정 모델을 제시하였다. 제시한 모델의 효율성을 확인하기 위해 기존에 잘 알려져 있는 최저뱃치크기(MBS) 규칙과 실험, 비교하였다.

• The Journal of the Acoustical Society of Korea
• /
• v.9 no.3
• /
• pp.42-47
• /
• 1990

A neural auditory system is studied for the aim of making better speech recognition systems. The cochlear mechanics is described. A IIR digital filter modeling of basilar membrane is discussed for the speech recognition. A multi-layer model of consonant recognition using phoneme detection filters and discriminant functions for feature estimation is constructed. This model shows more then 90% recognition rate in consonants.

• Journal of the Korean Geotechnical Society
• /
• v.16 no.6
• /
• pp.105-115
• /
• 2000

본 연구는 현실에 부합하는 연약지반의 압밀거동을 예측하기 위한 연구로서, 일단 3차원 배수 조건하에서 지반의 자중 및 압축성과 투수성의 비선형적 성질이 고려된 비선형 압밀모델을 구성하였다. 또한 연직 배수재의 시공과정에서 발생할수 있는 지반의 교란현상 및 다양한 이질층의 구성, 점증적인 하중재하 조건, 연직배수재의 부분관입 조건에 대한 고려가 가능하도록 비선형 압밀모델을 수정, 보완하였다. 이상의 연구결과를 바탕으로 유한차분방법에 의한 수치해석을 실시하였고 최종적으로 각종 희귀분석과정을 도입한 3차원 비선형 압밀해석 프로그램을 개발하였다. Ska-Edeby의 시험시공 사례를 통한 개발 프로그램의 검증을 실시하였는데, 시험시공 사례의 경우, 현장에서 측정한 깊이별 침하량 및 간극수압 결과를 개발 프로그램에 의한 예측결과와 비교, 분석하였다. 또한 개발 프로그램을 이용하여 다층지반 해석과 관련된 기존 해석방법의 문제점 및 지반의 교란효과와 연직배수재의 부분관입조건, 점증적인 하중재하 조건등이 지반의 압밀거동에 미치는 영향에 대해 살펴보았다.

• Annual Conference of KIPS
• /
• 2001.10a
• /
• pp.515-518
• /
• 2001

본 논문에서는 전처리단계로 영평균 정규화 기법과 주요성분분석 기법을 도입하여 다층신경망을 이용한 고신뢰성의 회귀분석 모델을 제안한다. 영평균 정규화 기법은 데이터의 1차적 통계성을 고려하여 알고리즘을 간략화시키며, 주요성분분석 기법은 입력 데이터의 2차적 통계성을 고려하여 독립인 특징들의 집합으로 변환시켜 학습데이터의 차원을 감소시킬 수 있어 고차원의 학습데이터에 따른 회귀분석 모델의 제약을 해결할 수 있었다. 제안된 기법의 신경망을 3개의 독립변수를 가진 암모니아 제조공정문제와 10개의 독립변수를 가진 자동차 연비문제에 각각 적용하여 시뮬레이션한 결과, 단순정규화나 PCA를 적용하지 않는 경우보다 제안된 기법의 학습속도와 회귀성능이 더욱 더 우수함을 확인할 수 있었다.