• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.58 no.12
• /
• pp.2511-2519
• /
• 2009

In this paper, we introduce the methodological system design via feature selection using Principal Component Analysis and Particle Swarm Optimization algorithms. The overall methodological system design comes from three kinds of modules such as preprocessing module, feature extraction module, and recognition module. First, Histogram equalization enhance the quality of image by exploiting contrast effect based on the normalized function generated from histogram distribution values of 2D face image. Secondly, PCA extracts feature vectors to be used for face recognition by using eigenvalues and eigenvectors obtained from covariance matrix. Finally the feature selection for face recognition among the entire feature vectors is considered by means of the Particle Swarm Optimization. The optimized Polynomial-based Radial Basis Function Neural Networks are used to evaluate the face recognition performance. This study shows that the proposed methodological system design is effective to the analysis of preferred face recognition.

• Proceedings of the KIEE Conference
• /
• 1991.07a
• /
• pp.329-332
• /
• 1991

This paper presents an algorithm to reduce the time to solve Power Equations using a Neural Net. The Neural Net is trained with samples obtained through the conventional AC Load Flow. With these samples, the Neural Net is constructed and has the function of a linear interpolation network. Given arbitrary load level, this Neural Net generates voltage magnitudes and angles which are linear interpolation of real and reactive powers. Obtained voltage magnitudes and angles are substituted to Power Equations, Real and reactive powers are found. Thus, a new sample is generated. This new experience modifies weight matrix. Continuing to modify the weight matrix, the correct solution is achieved. comparing this method with AC Load flow, this method is faster. If we consider parallel processing, this method is far faster than conventional ones.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1314-1320
• /
• 2019

In this paper, we propose a new method for optimizing bit operations and applying them to DNN(Deep Neural Network) in software environment. As a method for this, we propose a packing function for bitwise optimization and a masking matrix multiplication operation for application to DNN. The packing function converts 32-bit real value to 2-bit quantization value through threshold comparison operation. When this sequence is over, four 32-bit real values are changed to one 8-bit value. The masking matrix multiplication operation consists of a special operation for multiplying the packed weight value with the normal input value. And each operation was then processed in parallel using a GPU accelerator. As a result of this experiment, memory saved about 16 times than 32-bit DNN Model. Nevertheless, the accuracy was within 1%, similar to the 32-bit model.

• Journal of computational fluids engineering
• /
• v.19 no.2
• /
• pp.86-92
• /
• 2014

In this paper, prediction of separation trajectory for Two-stage-To-Orbit space launch vehicle has been numerically simulated by using an aerodynamic database based on steady state analysis. Aerodynamic database were obtained for matrix of longitudinal and vertical positions. The steady flow simulations around the launch vehicle have been made by using a 3-D RANS flow solver based on unstructured meshes. For this purpose, a vertex-centered finite-volume method was adopted to discretize inviscid and viscous fluxes. Roe's finite difference splitting was utilized to discretize the inviscid fluxes, and the viscous fluxes were computed based on central differencing. To validate this flow solver, calculations were made for the wind-tunnel experiment model of the LGBB TSTO vehicle configuration on steady state conditions. Aerodynamic database was constructed by using flow simulations based on test matrix from the wind-tunnel experiment. ANN(Artificial Neural Network) was applied to construct interpolation function among aerodynamic variables. Separation trajectory for TSTO launch vehicle was predicted from 6-DOF equation of motion based on the interpolated function. The result of present separation trajectory calculation was compared with the trajectory using experimental database. The predicted results for the separation trajectory shows fair agreement with reference[4] solution.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.12
• /
• pp.5116-5134
• /
• 2015

In this paper, due to the network security defense is mainly static defense, a dynamic classification network security defense strategy model is proposed by analyzing the security situation of complex computer network. According to the network security impact parameters, eight security elements and classification standard are obtained. At the same time, the dynamic classification algorithm based on fuzzy theory is also presented. The experimental analysis results show that the proposed model and algorithm are feasible and effective. The model is a good way to solve a safety problem that the static defense cannot cope with tactics and lack of dynamic change.

• Korean Journal of Agricultural Science
• /
• v.47 no.4
• /
• pp.1109-1122
• /
• 2020

The aim of this study was to develop a marbling classification and prediction model using small parts of sirloin images based on a deep learning algorithm, namely, a convolutional neural network (CNN). Samples were purchased from a commercial slaughterhouse in Korea, images for each grade were acquired, and the total images (n = 500) were assigned according to their grade number: 1++, 1+, 1, and both 2 & 3. The image acquisition system consists of a DSLR camera with a polarization filter to remove diffusive reflectance and two light sources (55 W). To correct the distorted original images, a radial correction algorithm was implemented. Color images of sirloins of Hanwoo (mixed with feeder cattle, steer, and calf) were divided and sub-images with image sizes of 161 × 161 were made to train the marbling prediction model. In this study, the convolutional neural network (CNN) has four convolution layers and yields prediction results in accordance with marbling grades (1++, 1+, 1, and 2&3). Every single layer uses a rectified linear unit (ReLU) function as an activation function and max-pooling is used for extracting the edge between fat and muscle and reducing the variance of the data. Prediction accuracy was measured using an accuracy and kappa coefficient from a confusion matrix. We summed the prediction of sub-images and determined the total average prediction accuracy. Training accuracy was 100% and the test accuracy was 86%, indicating comparably good performance using the CNN. This study provides classification potential for predicting the marbling grade using color images and a convolutional neural network algorithm.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.53 no.2
• /
• pp.65-75
• /
• 2004

This paper presents a generalized predictive control method based on a fuzzy neural network(FNN) model, which uses the on-line multi-step prediction, fur the intelligent control of chaotic nonlinear systems whose mathematical models are unknown. In our design method, the parameters of both predictor and controller are tuned by a simple gradient descent scheme, and the weight parameters of FNN are determined adaptively during the operation of the system. In order to design a generalized predictive controller effectively, this paper describes computing procedure for each of the two important parameters. Also, we introduce a projection matrix to determine the control input, which deceases the control performance function very rapidly. Finally, in order to evaluate the performance of our controller, the proposed method is applied to the Doffing and Henon systems, which are two representative continuous-time and discrete-time chaotic nonlinear systems, res reactively.

• Journal of Korean Institute of Industrial Engineers
• /
• v.1 no.2
• /
• pp.39-49
• /
• 1975

The problem treated is that of locating distribution centers(depot) in a network, so as to minimize the total cost which is the sum of transportation cost, (from factory to centers and from centers to demand points), construction cost, inventory cost and traffic increasing cost. This problem is mathematically an integer program and a non-linear model. This study avoids various inefficient aspects, which many studies have shown, by introducing a matrix notation, node and link function. An algorithm, for determining the optimal location of distribution center which has zone in which demand points are located at some node of a network, is presented. Finally this paper describes a numerical example and discusses its results.

• Journal of Korea Society of Industrial Information Systems
• /
• v.11 no.1
• /
• pp.29-37
• /
• 2006

Due to the spread of mobile communication with the development of computer network, nowadays various types of multimedia data play an important role in many areas such as entertainments, culture contents, e-commerce or medical science. But for the real application of these data, the security in the course of saving or transferring them through the public network should be assured. In this sense, many encryption algorithm have been developed and utilized. Nonetheless, most of them have focused on the text data. So they may not be suitable to the multimedia application because of their large size and real time constraint. In this paper, a chaotic map has been employed to create a symmetric stream type of encryption scheme which may be applied to the digital images with a large amounts of data. Then an efficient algebraic encryption algorithm based on the elementary operations of the Boolean matrix and image data characteristics.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.12 no.6
• /
• pp.577-582
• /
• 2002

Neural network can be trained to approximate an arbitrary nonlinear function of multivariate data like the mini-mill crown values in Automatic Shape Control. The trained weights of neural network can evaluate or generalize the process data outside the training vectors. Sometimes, the blind modeling of the process data is necessary to compare with the scattered analytical model of mini-mill process in isolated electro-mechanical forms. To come up with a viable model, we propose the blind neural-based range-division domain-clustering piecewise-linear modeling scheme. The basic ideas are: 1) dividing the range of target data, 2) clustering the corresponding input space vectors, 3)training the neural network with clustered prototypes to smooth out the convergence and 4) solving the resulting matrix equations with a pseudo-inverse to alleviate the ill-conditioning problem. The simulation results support the effectiveness of the proposed scheme and it opens a new way to the data analysis technique. By the comparison with the statistical regression, it is evident that the proposed scheme obtains better modeling error uniformity and reduces the magnitudes of errors considerably. Approximatly 10-fold better performance results.