• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.507-510
• /
• 2004

본 논문에서는 다채널 음향 시스템에서 라우드 스피커 이외의 위치에 가상음상을 정위시킬 때 발생되는 음상정위 오차 및 음색왜곡을 최소화 시키는 패닝기법을 제안한다. 기존 패닝기법의 음상정위 성능을 평가하기 위해 방향심리인자를 도입하고[1], 음색왜곡을 평가하기 위해 머리전달함수를 이용하여 실제 음원과 가상 음원과의 주파수특성 차이를 분석하였다. 이러한 객관적 성능분석을 통해 음상정위 오차 및 음색왜곡을 줄일 수 있는 새로운 패닝기법을 제안한다. 향상된 다채널패닝기법은 스테레오 및 서라운드 믹싱콘솔, 음향 편집 프로그램, 음장 재생기, 가상 서라운드 디코더 등 녹음 및 재생 전 분야에 걸쳐 사용될 수 있다.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1997.10a
• /
• pp.287-290
• /
• 1997

신경회로망은 높은 정확도의 학습 결과를 제시하는 장점을 가지고 있어서 패턴 인식을 포함한 여러 분야에서 널리 사용되어지고 있다. 그러나 신경회로망의 설계에 있어 최적의 뉴런과 층의 개수, 그리고 그 연결 등의 기하학적 해답을 제시하기가 어렵고, 서은이 우수하다고 알려진 역전파 학습 알고리즘도 오차가 없는 완벽한 학습 결과를 제시하지 못하며, 상당히 많은 학습 시간이 걸린다는 단점들을 가지고 있다. 이러한 단점들을 극복하기 위해 선형 신경회로망을 합성하는 새로운 방법을 제안하는데, 이진 함수 최소화(binary function minimization)과정을 거친 minimal-sum-of-product(MSP)를 통해서 이진 클래스 패턴(binary class pattem)을 표현 함으로써 오차가 없는 학습 결과를 얻을 수 있으며, 학습에 필요한 패턴과 학습에 걸리는 시간도 대폭 줄일수 있다. 본 논문에서는 유전자 알고리즘을 이용하여 선형 신경회로망을 합성하는 방법을 제안하며, 여러 가지 예제를 통해 제안한 방법의 우수성을 보인다.

• The Journal of the Korea Contents Association
• /
• v.4 no.3
• /
• pp.99-105
• /
• 2004

This paper proposes optimal learning rates in the gradient descent training of multilayer perceptrons, which are a separate learning rate for weights associated with each neuron and a separate one for assigning virtual hidden targets associated with each training pattern Effectiveness of the proposed error function was demonstrated for a handwritten digit recognition and an isolated-word recognition tasks and very fast learning convergence was obtained.

• Journal of the Korean Society for Precision Engineering
• /
• v.22 no.5 s.170
• /
• pp.55-62
• /
• 2005

This paper presents mainly a procedure to get the mathematical form of the manufactured aspherical lens. Generally Schulz formula describes the aspherical lens profile. Therefore, the base curvature, conic constant. and high-order polynomial coefficient should be set to get the approximated design equation. To find the best-approximated aspherical form, lens profile is measured by a commercial stylus profiler, which has a sub-micrometer measurement resolution. The optimization tool is based on the minimization of the root mean square of error sum to get the estimated aspherical surface equation from the scanned aspherical profile. Error minimization step uses the Nelder-Mead simplex (direct search) method. The result of the lens refractive power measurement shows the experimental consistency with the curvature distribution of the best-approximated aspherical surface equation

• Geophysics and Geophysical Exploration
• /
• v.25 no.3
• /
• pp.99-108
• /
• 2022

Among induced polarization (IP) methods, spectral IP (SIP) uses alternating current as a transmission source to measure amplitudes and phase of complex electrical resistivity at each source frequency, which disperse with respect to source frequencies. The frequency dependence, which can be explained by a relaxation model such as Cole-Cole model or equivalent models, is analyzed to estimate SIP parameters from dispersion curves of complex resistivity employing multi-objective optimization (MOO). The estimation uses a generic algorithm to optimize two objective functions minimizing data misfits of amplitude and phase based on Cole-Cole model, which is most widely used to explain IP relaxation effects. The MOO-based estimation properly recovered Cole-Cole model parameters for synthetic examples but hardly fitted for the real laboratory measures ones, which have relatively smaller values of phases (less than about 10 mrad). Discrepancies between scales for data misfits of amplitude and phase, used as parameters of MOO method, and it is in necessity to employ other methods such as machine learning, which can deal with the discrepancies, to estimate SIP parameters from dispersion curves of complex resistivity.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2007.05a
• /
• pp.1503-1506
• /
• 2007

수공구조물의 설계를 위해서 주로 사용되는 강우강도식은 연최대치 강우자료를 이용하여 빈도별 혹은 지속기간별 확률강우량을 구한 후 이 값들을 선형 혹은 비선형식의 형태로 회귀분석하여 구하게 된다. 그러나, 이와 같이 회귀분석을 이용하여 추정된 강우강도식은 원래의 강우자료가 가지고 있는 확률적인 특성을 재현한다고 하기는 어렵기 때문에, 본 연구에서는 연최대치 강우자료에 대한 적정 확률분포형으로부터 직접 강우 강도식을 유도하는 방법을 적용하여 대상지역 강우강도식의 매개변수를 산정하였다. 선정된 적정 확률분포형을 이용하여 강우강도식의 매개변수를 추정하는데 있어서, 평균제곱오차의 제곱근을 최소화하는 형태의 목적함수를 구성한 후 유전자알고리즘을 이용하여 적절한 매개변수를 산정하였다. 산정된 매개변수를 사용한 강우강도식으로 구한 결과값과 기존의 강우강도식에 의한 결과값 그리고 지점빈도해석에 의한 결과값을 비교하여 본 연구에서 산정된 강우강도식의 적용성을 평가해 보았다.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.15 no.2
• /
• pp.70-78
• /
• 2011

The paper aims at developing a more reliable and accurate BWIM(Bridge Weigh-In-Motion) algorithm using measured strain data and examining its efficiency with various tests on bridges. It proposes a BWIM algorithm using density estimation function and average modification factor for moment-strain relationship. Density estimation function has been proved to be reliably applied when multiple axle loads are estimated. An average modification factor is applied to minimize overall error that can be encountered between theoretically computed moments and measured strains at multiple locations in a bridge. The developed algorithm has been successfully examined through numerical simulations, laboratory tests, and also by field tests on a multi-girder composite bridge.

• Journal of Korea Water Resources Association
• /
• v.45 no.11
• /
• pp.1169-1186
• /
• 2012

Storage function method has been used for flood forecasting in the major rivers in Korea, however, the researches on the relationship between the parameters and runoff characteristics was not sufficient. In addition, there has been a controversy about the optimized parameters without the consideration of the physical characteristics of the basin. Therefore, in this study, the SCE-UA method is used to optimize the parameters and the proposed method was applied with two stage optimization in the Jeongseon and Yeongwol watersheds located in the most upstream in the South Han river. The contour map was developed to investigate parameters and the error surface calculated from the runoff. The proposed parameters is to provide a range of the possible parameter set in a watershed, rather than a specific value. However, the applicability is examined using the average value of the proposed ranged parameters. In this study, the criticism about the optimization technique to find an optimal value having no physical meaning on a watershed is tried to avoid. The objective of this study is to provide a range of parameters for the flood forecasting model and the intuition about the behavior of the parameters, so the efficiency of flood forecasting is increased.

• The Korean Journal of Applied Statistics
• /
• v.27 no.3
• /
• pp.501-512
• /
• 2014

Modified cut-off sampling is widely used for highly skewed data. A serious drawback of modified cut-off sampling is the difficulty of adjustment of non-response in take-all stratum. Therefore, solutions of the problems of non-response in take-all stratum have been studied in various ways such as substitute of samples, imputation or re-weight method. In this paper, a new cut-off point based on minimizing MSE being used in exponential and power functions is suggested and it can be reduced the number of take-all stratum. We also investigate another cut-off point determination method with underlying distributions such as truncated log-normal and truncated gamma distributions. Finally we suggest the optimal cut-off point which has a minimum of take-all stratum size among suggested methods. Simulation studies are performed and Labor Survey data and simulated data are used for the case study.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.12
• /
• pp.1682-1687
• /
• 2020

We study the problem of sampling a subset of nodes of graphs for bandlimited graph signals such that the signal values on the sampled nodes provide the most information in order to reconstruct the original graph signal. Instead of directly minimizing the reconstruction error, we focus on minimizing the upper bound of the reconstruction error to reduce the complexity of the selection process. We further simplify the upper bound by applying useful approximations to propose a low-weight greedy selection process that is iteratively conducted to find a suboptimal sampling set. Through the extensive experiments for various graphs, we inspect the performance of the proposed algorithm by comparing with different sampling set selection methods and show that the proposed technique runs fast while preserving a competitive reconstruction performance, yielding a practical solution to real-time applications.