• 한국음향학회지
• /
• 제20권3호
• /
• pp.34-41
• /
• 2001

기존의 IMBE (개량형 다중대역 여기 : Improved Multi-band Excitation) 음성 부호기의 초기 피치 추정 과정은 전체 부호기 연산 시간의 대부분을 차지하며 또한 미래의 음성 프레임들이 초기 피치 추정시 사용되므로 시간 지연이 유발되어 실시간 구현에 장애 요소로 작용되었다. 또한 무성음에 해당되는 프레임에 대해서도 유성음과 동일한 피치 추정을 수행하므로 알고리즘의 효율성을 떨어뜨린다 본 논문에서는 초기 피치 추정 전에 다이애딕 웨이브렛 변환 (Dyadic Wavelet Transform)을 이용하여 이를 바탕으로 유/무성음을 판별한 후 유성음으로 결정된 프레임에 대해서만 피치 추정을 행하고 무성음으로 결정된 프레임은 랜덤 잡음을 주어서 부호화시의 처리 시간을 단축하였다. 또한, 초기 피치 추정 전에 판별된 유/무성음을 판별하여 유성음과 무성음에 각기 다른 초기 피치 추정 알고리즘을 사용하고 미래의 두 프레임을 사용하지 않음으로써 송, 수신단에 유발되는 시간 지연을 제거하였다. 그 결과 초기 피치 추정 과정의 상대적인 복잡도가 23％ 감소되었고 프레임당 처리 시간이 1/10∼ 1/11로 감소되었고 기존의 부호기와 거의 같은 음질을 얻을 수 있었다.

• 한국데이터정보과학회:학술대회논문집
• /
• 한국데이터정보과학회 2002년도 춘계학술대회
• /
• pp.75-84
• /
• 2002

This paper compares long-run returns of privatization initial public offerings to those of domestic stock markets of respective countries using a sample of 196 privatization initial public offerings from 39 countries. The evidence indicates that the privatization initial public offerings (IPOs) significantly outperform their domestic stock markets. There are substantial differences in the long-run performance of privatization IPOs depending on the return estimation techniques, however. Evidence indicates that the inference based either on conventional t or on skewness-adjusted t statistics may yield misspecified test statistics. The quality of estimation tends to be improved by simply eliminating the outliers from the sample, especially for the buy-and-hold abnormal return technique.

• 대한의용생체공학회:의공학회지
• /
• 제10권2호
• /
• pp.114-116
• /
• 1989

A method has been proposed for the fully automatic detection of left ventricular endocardial boundary in 2D short axis echocardiogram using geometric model. The procedure has the following three distinct stages. First, the initial center is estimated by the initial center estimation algorithm which is applied to decimated image. Second, the center estimation algorithm is applied to original image and then best-fit elliptic model estimation is processed. Third, best-fit boundary is detected by the cost function which is based on the best-fit elliptic model. The proposed method shows effective result without manual intervention by a human operator.

• 한국정보과학회논문지:컴퓨팅의 실제 및 레터
• /
• 제16권12호
• /
• pp.1219-1223
• /
• 2010

본 논문은 모션 속도와 다수의 초기 중심점에 기반한 빠른 모션 추정 알고리즘을 제안한다. 제안한 방법은 시공간적 이웃 모션 벡터들에 의해 초기 탐색점을 예측한다. 모션벡터를 빠르게 구하기 위하여 모션 속도와 예측된 초기 중심점들에 기반한 동적인 탐색 패턴이 이용된다. 제안한 방법은 시공간 정보와 동적 탐색 패턴을 이용하여 탐색 속도를 매우 빠르게 할 뿐만 아니라 양질의 영상화질을 유지할 수 있다. 실험결과를 통해 제안한 방법이 완전 탐색, 새로운 삼단계 탐색, 사단계 탐색 방법들과 비교하여 검색시간을 줄이면서 PSNR 관점에서 양질의 영상화질을 제공함을 알 수 있다.

• 전력전자학회논문지
• /
• 제10권2호
• /
• pp.124-129
• /
• 2005

본 논문에서는 서보분야에 적용되는 고분해능 엔코더를 장착하고 자극위치 센서인 홀센서를 부착하지 않은 선형영구자석 동기전동기의 개선된 초기 자극위치 추정방법을 제안하였다. 제안한 방법은 수치해석법인 할선법을 적용하여 추력이 영인 두 곳 중에서 한 곳을 추정하고, 추정 q축에 시험전류를 인가하여 실제 전동기의 d축을 추정한다. 그리고 전류제어기와 위치정보만을 사용하는 제안한 추정방법은 적용이 용이한 장점을 가지며, 회전형 동기전 동기에도 쉽게 적용할 수 있다. PMLSM에 대한 실험을 통하여 매우 짧은 이동거리(약 평균 85㎛)내에서 빠른 시간 안에 높은 초기각 추정 정밀도를 보이는 우수한 실험결과를 제시하여 제안한 방법의 타당성을 검증하였다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1996년도 추계학술대회 논문집 학회본부
• /
• pp.182-185
• /
• 1996

The purpose of slate estimation in power system is to estimate the best-fit Slate variables from the measurements contaminated by various kind of noise. But because the majority of state estimation modules in EMS lack the convergence characteristics, sometimes the desirable outputs can't be obtained. So, in this paper, the new algorithm using the load now output as initial values in the state estimation calculation is proposed to guarantee the convergence. And if the load now outputs were used as the initial values in the calculation, the change in each step would be small compared to the original method using the flat start point. And the Inverse Lemma is used in the algorithm to calculate the new stale in each iteration step for reducing the calculation time. The proposed algorithm was tested on the IEEE 14, 30, 118 bus systems. Eventually, we were able to verity that the differences between the results obtained by the original method and proposed method were relatively small, and the effectiveness of the proposed algorithm increased when applied to the bigger systems.

• 한국해양공학회지
• /
• 제19권6호통권67호
• /
• pp.78-85
• /
• 2005

Initial alignment and localization are important topics in inertial navigation systems, since misalignment and initial position error wholly propagate into the navigation systems and deteriorate the performance of the systems. This paper presents the error convergence characteristics of the hybrid navigation system for underwater vehicles initial position, which is based on an inertial measurement unit (IMU) accompanying a range sensor. This paper demonstrates the improvement on the navigational performance oj the hybrid system with the range information, especially focused on the convergence of the estimation of underwater vehicles initial position error. Simulations are performed with experimental data obtained from a rotating ann test with a fish model. The convergence speed and condition of the initial error removal for random initial position errors are examined with Monte Carlo simulation. In addition, numerical simulation is conducted with an AUV model in lawn-mowing survey mode to illustrate the error convergence of the hybrid navigation System for initial position error.

• Journal of information and communication convergence engineering
• /
• 제8권5호
• /
• pp.519-523
• /
• 2010

In Gen-2 RFID system, the initial value of $Q_{fp}$, which is the slot-count parameter of Q-algorithm, is not defined in the standard. In this case, if the number of tags within the reader's identification range is small and we let the initial $Q_{fp}$ be large, the number of empty slot will be large. On the other hand, if we let the initial $Q_{fp}$ be small in spite of many tags, almost all the slots will be collided. As a result, the performance will be declined because the frame size does not converge to the optimal point quickly during the query round. In this paper, we propose a scheme to allocate the optimal initial $Q_{fp}$ through the tag number estimation before the query round begins. Through computer simulations, it is demonstrated that the proposed scheme achieves more stable performance than Gen-2 Q-algorithm.

• 전자공학회논문지B
• /
• 제30B권8호
• /
• pp.54-63
• /
• 1993

This paper proposes a method for improving the performances of the neural network for optimization by an optimal estimation of initial states. The optimal initial state that leads to the global minimum is estimated by using the stochastic approximation. And then the update rule of Hopfield model, which is the high speed deterministic algorithm using the steepest descent rule, is applied to speed up the optimization. The proposed method has been applied to the tavelling salesman problems and an optimal task partition problems to evaluate the performances. The simulation results show that the convergence speed of the proposed method is higher than conventinal Hopfield model. Abe's method and Boltzmann machine with random initial neuron output setting, and the convergence rate to the global minimum is guaranteed with probability of 1. The proposed method gives better result as the problem size increases where it is more difficult for the randomized initial setting to give a good convergence.

• Communications for Statistical Applications and Methods
• /
• 제7권1호
• /
• pp.65-86
• /
• 2000

The estimates from an EM when it is applied to a large causal model of 10 or more categorical variables are often subject to the initial values for the estimates. This phenomenon becomes more serious as the model structure becomes more serious as the model structure becomes more complicated involving more variables. In this regard Wu(1983) recommends among others that EMs are implemented several times with different sets of initial values to obtain more appropriate estimates. in this paper a new approach for initial values is proposed. The main idea is that we use initials that are calibrated to data. A simulation result strongly indicates that the calibrated initials give rise to the estimates that are far closer to the true values than the initials that are not calibrated.