• 대한수학회보
• /
• 제42권4호
• /
• pp.837-849
• /
• 2005

In this paper, we obtain a successive quadratic programming algorithm for solving the semidefinite programming (SDP) relaxation of the binary quadratic programming. Combining with a randomized method of Goemans and Williamson, it provides an efficient approximation for the binary quadratic programming. Furthermore, its convergence result is given. At last, We report some numerical examples to compare our method with the interior-point method on Maxcut problem.

• 응용통계연구
• /
• 제20권1호
• /
• pp.103-115
• /
• 2007

본 논문에서는 1차 자기회귀모형에서 자기회귀계수에 대한 여러 가지 추정량들의 분포함수에 대한 근사 방법에 대해 연구하였다. 자기회귀계수의 여러 추정량들을 이차형식의 관점에서 이해하고, Na와 Kim(2005)에 의한 안장점근사의 결과를 이용한 새로운 근사법을 제시하였다. 이 방법은 정규근사를 비롯한 기존의 근사법과는 달리 추정량에 대한 근사분포의 유도과정이 불필요하며, 소표본은 물론 통계적 추론의 주요 관심영역에서의 근사정도가 매우 뛰어난 장점을 가지고 있다. 모의실험을 통해 Edgeworth 근사를 비롯한 기존의 여러 근사법보다 효율이 뛰어남을 확인하였다.

• International Journal of Control, Automation, and Systems
• /
• 제6권4호
• /
• pp.506-514
• /
• 2008

This paper presents an intelligent model; named as free model, approach for a closed-loop system identification using input and output data and its application to design a power system stabilizer (PSS). The free model concept is introduced as an alternative intelligent system technique to design a controller for such dynamic system, which is complex, difficult to know, or unknown, with input and output data only, and it does not require the detail knowledge of mathematical model for the system. In the free model, the data used has incremental forms using backward difference operators. The parameters of the free model can be obtained by simultaneous perturbation stochastic approximation (SPSA) method. A linear transformation is introduced to convert the free model into a linear model so that a conventional linear controller design method can be applied. In this paper, the feasibility of the proposed method is demonstrated in a one-machine infinite bus power system. The linear quadratic regulator (LQR) method is applied to the free model to design a PSS for the system, and compared with the conventional PSS. The proposed SPSA-based LQR controller is robust in different loading conditions and system failures such as the outage of a major transmission line or a three phase to ground fault which causes the change of the system structure.

• 응용통계연구
• /
• 제18권1호
• /
• pp.183-196
• /
• 2005

본 논문에서는 다변량 정규분포하에서 비동차(non-homogeneous) 이차형식의 분포 함수에 대한 안장점근사법을 다루었다. 이는 Kuonen (1999)의 동차(homogeneous) 이차형식에 대한 안장점근사를 비동차의 경우로 확장한 것이다. 안장점근사의 적용을 위해 비동차 이차형식의 누율생성함수 및 관련 성질들을 유도하였다. 모의실험을 통해 안장점근사의 정도가 매우 뛰어남을 확인하였다.

• Communications for Statistical Applications and Methods
• /
• 제19권1호
• /
• pp.135-145
• /
• 2012

It is well known that the maximum likelihood estimator(MLE) in normal mixture models with unequal variances does not fall in the interior of the parameter space. Recently, a doubly smoothed maximum likelihood estimator(DS-MLE) (Seo and Lindsay, 2010) was proposed as a general alternative to the ordinary maximum likelihood estimator. Although this method gives a natural modification to the ordinary MLE, its computation is cumbersome due to intractable integrations. In this paper, we derive an EM algorithm for the DS-MLE under normal mixture models and propose a fast computational tool using a local quadratic approximation. The accuracy and speed of the proposed method is then presented via some numerical studies.

• Journal of the Korean Data and Information Science Society
• /
• 제21권3호
• /
• pp.561-567
• /
• 2010

Multiclassification is typically performed using voting scheme methods based on combining a set of binary classifications. In this paper we use multiclassification method with a hat matrix of least squares support vector machine (LS-SVM), which can be regarded as the revised one-against-all method. To tackle multiclass problems for large data, we use the $Nystr\ddot{o}m$ approximation and the quadratic Renyi entropy with estimation in the primal space such as used in xed size LS-SVM. For the selection of hyperparameters, generalized cross validation techniques are employed. Experimental results are then presented to indicate the performance of the proposed procedure.

• Communications for Statistical Applications and Methods
• /
• 제20권6호
• /
• pp.481-490
• /
• 2013

Classification is an important research area as data can be easily obtained even if the number of predictors becomes huge. The support vector machine(SVM) is widely used to classify a subject into a predetermined group because it gives sound theoretical background and better performance than other methods in many applications. The SVM can be viewed as a penalized method with the hinge loss function and penalty functions. Instead of $L_2$ penalty function Fan and Li (2001) proposed the smoothly clipped absolute deviation(SCAD) satisfying good statistical properties. Despite the ability of SVMs, they have drawbacks of non-robustness when there are outliers in the data. We develop a robust SVM method using a weight function with the SCAD penalty function based on the local quadratic approximation. We compare the performance of the proposed SVM with the SVM using the $L_1$ and $L_2$ penalty functions.

• 전자공학회논문지B
• /
• 제30B권6호
• /
• pp.67-75
• /
• 1993

ECG pattern was classified using a back-propagation neural network. An improved feature extractor of ECG is proposed for better classification capability. It is consisted of preprocessing ECG signal by an FIR filter faster than conventional one by a factor of 5. QRS complex recognition by moving-window integration, and peak extraction by quadratic approximation. Since the FIR filter had a periodic frequency spectrum, only one-fifth of usual processing time was required. Also, segmentation of ECG signal followed by quadratic approximation of each segment enabled accurate detection of both P and T waves. When improtant features were extracted and fed into back-propagation neural network for pattern classification, the required number of nodes in hidden and input layers was reduced compared to using raw data as an input, also reducing the necessary time for study. Accurate pattern classification was possible by an appropriate feature selection.

• 대한전기학회논문지:전력기술부문A
• /
• 제53권5호
• /
• pp.296-301
• /
• 2004

The Available Transfer Capability (ATC) is defined as the measure of the transfer capability remaining in the physical transmission network for further commercial activity above already committed uses. Available Transfer Capability (ATC) calculation is a complicated task, which involves the determination I of total transfer capability (TTC), transmission reliability margin (TRM) and capability benefit margin (CBM). As the electrical power industry is restructured and the electrical power exchange is updated per hour, it is important to accurately and rapidly quantify the available transfer capability (ATC) of the transmission system. In ATC calculation,. the existing CPF method is accurate but it has long calculation time. On the contrary, the method using PTDF is fast but it has relatively a considerable error. This paper proposed QFA method, which can reduce calculation time comparing with CPF method and has few errors in ATC calculation. It proved that the method can calculate ATC more fast and accurately in case study using IEEE 24 bus RTS.

• 대한기계학회논문집A
• /
• 제20권12호
• /
• pp.3804-3821
• /
• 1996

The configuration optimization is a structural optimization method which includes the coordinates of a structure as well as the sectional properties in the design variable set. Effective reduction of the weight of discrete structures can be obrained by changing the geometry while satisfying stress, Ei;er bickling, displacement, and frequency constraints, etc. However, the nonlinearity due to the configuration variables may cause the difficulties of the convergence and expensive computational cost. An efficient approximation method for the configuration optimization has been developed to overcome the difficulties. The method approximates the constraint functions based onthe second-order Taylor series expansion with reciprocal design variables. The Hessian matrix is approzimated from the information on previous design points. The developed algotithms are coded and the examples are solved.