• The Transactions of the Korea Information Processing Society
• /
• v.4 no.9
• /
• pp.2412-2420
• /
• 1997

An algorithm for detecting dominant points on a digital contour is proposed. The algorithm detects the dominant points from the given contour with the given number of points as a constraint condition. on the basis of the principle of the top-down approach. The dominant points are detected by minimizing the object function that presents the similarity between the given contour and the approximated polygon drawn by connecting the dominant points of candicate. The penalty multiplier method is applied to minimize the augmented Lagrangean function which is made by adding the penalty of the constraint condition to the object function. On the minimization, a local searching method by the partial problem division is considered, and it is clarified that the reasonable solution is obtained by the method. The proposed algorithm has a merit that the dominant points can be detected exactly and stably even for the digital contour composed of multiple-scale features and the similar contours, because it detects them on considering the property of a whole figure of the given contour. It is confirmed that the proposed algorithm is more excellent than other previously proposed algorithms by the comparison and the evaluation through the experiment on suing typical digital curves.

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.6
• /
• pp.670-677
• /
• 2019

As the presence of background noise in acoustic signal degrades the performance of speech or acoustic event recognition, it is still challenging to extract noise-robust acoustic features from noisy signal. In this paper, we propose a combined structure of Wasserstein Generative Adversarial Network (WGAN) and MultiTask AutoEncoder (MTAE) as deep learning architecture that integrates the strength of MTAE and WGAN respectively such that it estimates not only noise but also speech features from noisy acoustic source. The proposed MTAE-WGAN structure is used to estimate speech signal and the residual noise by employing a gradient penalty and a weight initialization method for Leaky Rectified Linear Unit (LReLU) and Parametric ReLU (PReLU). The proposed MTAE-WGAN structure with the adopted gradient penalty loss function enhances the speech features and subsequently achieve substantial Phoneme Error Rate (PER) improvements over the stand-alone Deep Denoising Autoencoder (DDAE), MTAE, Redundant Convolutional Encoder-Decoder (R-CED) and Recurrent MTAE (RMTAE) models for robust speech recognition.

• Proceedings of the KIEE Conference
• /
• 2007.11b
• /
• pp.226-228
• /
• 2007

This paper proposes Nonlinear Interior Point Method (NIPM) including discrete control variables optimal power flow formulations. The algorithm utilizes the robustness in terms of starting point and fast convergence for large scale power system of NIPM and an introduction of rounding penalty function which is augmented in the Lagrangian function to handle discrete control variables. The derived formulation shows a simplified approach to deal with discrete control problems which is implementable in real large scale systems.

• Journal of applied mathematics & informatics
• /
• v.27 no.5_6
• /
• pp.981-1000
• /
• 2009

This paper considers asset liability management problems when their deterministic equivalent formulations are general nonlinear optimization problems. The presented approach uses a nonmonotone trust region strategy for solving a sequence of unconstrained subproblems parameterized by a scalar parameter. The objective function of each unconstrained subproblem is an augmented penalty-barrier function that involves both primal and dual variables. Each subproblem is solved approximately. The algorithm does not restrict a monotonic decrease of the objective function value at each iteration. If a trial step is not accepted, the algorithm performs a non monotone line search to find a new acceptable point instead of resolving the subproblem. We prove that the algorithm globally converges to a point satisfying the second-order necessary optimality conditions.

• Proceedings of the KIEE Conference
• /
• 1987.11a
• /
• pp.425-429
• /
• 1987

This paper describes a collision avoidance method for planning safe trajectories for multi-robot system in common work space. Usually objects have been approximated to convex polyhedra in most previous researches, but in case using such the approximation method it is difficult to represent objects analytically in terms of functions and also to describe tile relationship between the objects. In this paper, in order to solve such problems a modeling method which approximates objects to cylinder ended by hemispheres and or sphere is used and the maximum distance functions is defined which call be calculated simply. Using an objective function with inequality constraints which are related to minimum distance functions, work range and maximum allowable angular velocities of the robots, tile collision avoidance for two robots is formulated to a constrained function optimization problem. With a view to solve tile problem a penalty function having simple form is defined and used. A simple numerical example involving two PUMA-type robots is described.

• Journal of Information Processing Systems
• /
• v.15 no.6
• /
• pp.1503-1516
• /
• 2019

In this paper, to improve the accuracy of long sentence similarity calculation, we proposed a sentence similarity calculation method based on a system similarity function. The algorithm uses word2vector as the system elements to calculate the sentence similarity. The higher accuracy of our algorithm is derived from two characteristics: one is the negative effect of penalty item, and the other is that sentence similar function (SSF) based on word2vector similar elements doesn't satisfy the exchange rule. In later studies, we found the time complexity of our algorithm depends on the process of calculating similar elements, so we build an index of potentially similar elements when training the word vector process. Finally, the experimental results show that our algorithm has higher accuracy than the word mover's distance (WMD), and has the least query time of three calculation methods of SSF.

• Bulletin of the Society of Naval Architects of Korea
• /
• v.24 no.4
• /
• pp.45-52
• /
• 1987

In this paper, merits and demerits of Nelder and Mead Penalty Function Method(SUMTNM) and Flexible Tolerance Method(FTM) are investigated from the standpoint of generality, accuracy and efficiency. SUMTNM is combined with Nelder and Method and SUMT, but FTM improves the values of the objective function by using information provided by feasible points as well as certain nonfeasible points termed near-feasible points. Therefore, FTM uses more information than SUMTNM for minimizing object function. The structural analysis of a vertically corrugated bulkhead is performed by collapse mechanism and plate buckling analysis. Based on the results of this analysis, minimum structural weight design of a corrugated bulkhead by use of above two optimization techniques is carried out by investigating the effects of sizes of bulkhead on the structural weight.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.33 no.10
• /
• pp.410-422
• /
• 1984

This paper presents a new practical method for optimal active and reactive power control for the economic operation in electrical power system, and the programs are developed for digital computer solution. The major features and techniques of this paper are as follows: 1) The method is presented for finding the equivalent active power balance equation applying the sparse Jacobian matrix of power flow equation instead of using B constant as active power balance equation considering transmission loss, and thus for determining directly optimal active power allocation berween generator unitw satisfying the equality and inequality constraints. 2) The method is proposed for solving directly the optimum economim dispatch problem without using gradient method and penalty function for both active and reactive power control. As a result, the computing time are reduced and convergence characteristic is remarkably improved. 3) Unlike most of conventional methods which adopt the transmission loss as a objective function for reactive power control, the total fuel cost of themal power plant is adopted as objective function for both active and reactive power control. consequently, more reasonable and economic profit can be achieved.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.20 no.44
• /
• pp.93-101
• /
• 1997

This paper describes the application of Genetic Algorithms(GAs) to nonlinear constrained mixed optimization problems. Genetic Algorithms are combinatorial in nature, and therefore are computationally suitable for treating discrete and integer design variables. But, several problems that conventional GAs are ill defined are application of penalty function that can be adapted to transform a constrained optimization problem into an unconstrained one and premature convergence of solution. Thus, we developed an improved GAs to solve this problems, and two examples are given to demonstrate the effectiveness of the methodology developed in this paper.

• Journal of Korean Institute of Industrial Engineers
• /
• v.20 no.3
• /
• pp.105-116
• /
• 1994

This paper presents an inventory model with partial backorders for the situation in which demand is deterministic, lead time follows normal distribution and back order ratio during the stockout period varies in proportion to the length of backorder period In this situations, an objective function is formulated to minimize a time-proportional backorder cast and a fixed penalty cost per unit lost. And then the procedure of iterative solution method for the model is developed to find optimal reorder paint and order quantity and a numerical example to illustrate the proposed method is presented.