• Korean Journal of Mathematics
• /
• 제13권2호
• /
• pp.127-137
• /
• 2005

The objective of numerical analysis is to devise and analyze efficient algorithms or numerical methods for equations arising in mathematical modeling for science and engineering. In this article, we present some recent topics in computational mathematics, specially in the finite element method and overview the development of the mixed finite element method in the context of second order elliptic and parabolic problems. Multiscale methods such as MsFEM, HMM, and VMsM are included.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제13권3호
• /
• pp.169-180
• /
• 2009

A discontinuous Galerkin method with interior penalty terms is presented for linear Sobolev equation. On appropriate finite element spaces, we apply a symmetric interior penalty Galerkin method to formulate semidiscrete approximate solutions. To deal with a damping term $\nabla{\cdot}({\nabla}u_t)$ included in Sobolev equations, which is the distinct character compared to parabolic differential equations, we choose special test functions. A priori error estimate for the semidiscrete time scheme is analyzed and an optimal $L^\infty(L^2)$ error estimation is derived.

• 대한수학회지
• /
• 제44권6호
• /
• pp.1281-1290
• /
• 2007

We introduce three spectral regularization methods for solving a backward heat conduction problem (BHCP). For the three spectral regularization methods, we give the stability error estimates with optimal order under an a-priori and an a-posteriori regularization parameter choice rule. Numerical results show that our theoretical results are effective.

• The Journal of the Acoustical Society of Korea
• /
• 제29권2E호
• /
• pp.86-99
• /
• 2010

This paper analyzes the performance of various single channel speech enhancement algorithms when they are applied to automatic speech recognition (ASR) systems as a preprocessor. The functional modules of speech enhancement systems are first divided into four major modules such as a gain estimator, a noise power spectrum estimator, a priori signal to noise ratio (SNR) estimator, and a speech absence probability (SAP) estimator. We investigate the relationship between speech recognition accuracy and the roles of each module. Simulation results show that the Wiener filter outperforms other gain functions such as minimum mean square error-short time spectral amplitude (MMSE-STSA) and minimum mean square error-log spectral amplitude (MMSE-LSA) estimators when a perfect noise estimator is applied. When the performance of the noise estimator degrades, however, MMSE methods including the decision directed module to estimate a priori SNR and the SAP estimation module helps to improve the performance of the enhancement algorithm for speech recognition systems.

• 대한수학회논문집
• /
• 제33권3호
• /
• pp.1025-1037
• /
• 2018

We consider an ill-posed problem for the heat equation $u_{xx}=u_t$ in the quarter plane {x > 0, t > 0}. We propose a new method to compute the heat flux $h(t)=u_x(1,t)$ from the boundary temperature g(t) = u(1, t). The operator $g{\mapsto}h=Hg$ is unbounded in $L^2({\mathbb{R}})$, so we approximate h(t) by $h_{\delta}(t)=u_x(1+{\delta},\;t)$, ${\delta}{\rightarrow}0$. When noise is present, the data is $g_{\epsilon}$ leading to a corresponding heat $h_{{\delta},{\epsilon}}$. We obtain an estimate of the error ${\parallel}h-h_{{\delta},{\epsilon}}{\parallel}$, as well as the error when $h_{{\delta},{\epsilon}}$ is approximated by the trapezoidal rule. With an a priori choice rule ${\delta}={\delta}({\epsilon})$ and ${\tau}={\tau}({\epsilon})$, the step size of the trapezoidal rule, the main theorem gives the error of the heat flux as a function of noise level ${\epsilon}$. Numerical examples show that the proposed method is effective and stable.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1991년도 춘계학술대회
• /
• pp.65-70
• /
• 1991

A hybrid model which uses a probabilistic model and a MLP(multi layer perceptron) model for pattern recognition of EMG(electromyogram) signals is proposed in this paper. MLP model has problems which do not guarantee global minima of error due to learning method and have different approximation grade to bayesian probabilities due to different amounts and quality of training data, the number of hidden layers and hidden nodes, etc. Especially in the case of new test data which exclude design samples, the latter problem produces quite different results. The error probability of probabilistic model is closely related to the estimation error of the parameters used in the model and fidelity of assumtion. Generally, it is impossible to introduce the bayesian classifier to the probabilistic model of EMG signals because of unknown priori probabilities and is estimated by MLE(maximum likelihood estimate). In this paper we propose the method which get the MAP(maximum a posteriori probability) in the probabilistic model by estimating the priori probability distribution which minimize the error probability using the MLP. This method minimize the error probability of the probabilistic model as long as the realization of the MLP is optimal and approximate the minimum of error probability of each class of both models selectively. Alocating the reference coordinate of EMG signal to the outside of the body make it easy to suit to the applications which it is difficult to define and seperate using internal body coordinate. Simulation results show the benefit of the proposed model compared to use the MLP and the probabilistic model seperately.

• 제어로봇시스템학회논문지
• /
• 제13권4호
• /
• pp.335-343
• /
• 2007

This paper deals with the method to estimate the friction in a system. We study a nonlinear friction model to estimate the friction in an inverted pendulum and approximate the friction model using fuzzy basis functions expansion. To demonstrate the friction observer using FBFs, we derive a update rule based on the error term that is formed by the output from a real system and observer output with a friction estimate. And two compensation algorithms to improve the response of an inverted pendulum are proposed. The first method that a observer parameter is updated in on-line and the friction is compensated at the same time. The second method is to compensate the friction with observer parameter estimated priori. The two methods is compared through the experimental results.

• 한국멀티미디어학회논문지
• /
• 제18권7호
• /
• pp.886-894
• /
• 2015

In this paper, we conduct sensitivity analysis of parameters used for inverse modeling in order to estimate the PM₁₀ emissions from the 16 areas in East Asia accurately. Parameters used in sensitivity analysis are R, the observational error covariance matrix, and B, a priori (background) error covariance matrix. In previous studies, it was used with the predetermined parameter empirically. Such a method, however, has difficulties in estimating an accurate emissions. Therefore, an automatically determining method for the most suitable value of R and B with an error measurement criteria and posteriori emissions accuracy is required. We determined the parameters through a sensitivity analysis, and improved the accuracy of posteriori emissions estimation. Inverse modeling methods used in the emissions estimation are pseudo inverse, NNLS (Nonnegative Least Square), and BA(Bayesian Approach). Pseudo inverse has a small error, but has negative values of emissions. In order to resolve the problem, NNLS is used. It has a unrealistic emissions, too. The problems are resolved with BA(Bayesian Approach). We showed the effectiveness and the accuracy of three methods through case studies.

• 전기학회논문지
• /
• 제59권10호
• /
• pp.1874-1881
• /
• 2010

A recursive linear stochastic error compensation algorithm is newly proposed for target localization in sensor network which provides range difference of arrival(RDOA) measurements. Target localization with RDOA is a well-known nonlinear estimation problem. Since it can not solve with a closed-form solution, the numerical methods sensitive to initial guess are often used before. As an alternative solution, a pseudo-linear estimation scheme has been used but the auto-correlation of measurement noise still causes unacceptable estimation errors under low SNR conditions. To overcome these problems, a stochastic error compensation method is applied for the target localization problem under the assumption that a priori stochastic information of RDOA measurement noise is available. Apart from the existing methods, the proposed linear target localization scheme can recursively compute the target position estimate which converges to true position in probability. In addition, it is remarked that the suggested algorithm has a structural reconciliation with the existing one such as linear correction least squares(LCLS) estimator. Through the computer simulations, it is demonstrated that the proposed method shows better performance than the LCLS method and guarantees fast and reliable convergence characteristic compared to the nonlinear method.

• 전자공학회논문지S
• /
• 제35S권3호
• /
• pp.166-175
• /
• 1998

When we make a speaker adapted model using MAPE (maximum a posteriori estimation), the adapted model has one mixture in each state. This is because we cannot estimate a number of a priori distribution from a speaker-independent model in each state. If the model is represented by one mixture in each state, it is not well adadpted to specific speaker because it is difficult to represent various speech informationof the speaker with one mixture. In this paper, we suggest the method using several mixtures to well represent various speech information of the speaker in each state. But, because speaker-specific training dat is not sufficient, this method can't be used in every state. So, we make the number of mixtures in each state variable in proportion to the number of frames and to the determinant ofthe variance matrix in the state. Using the proposed method, we reduced the error rate than methods using one branch in each state.