• ETRI Journal
• /
• v.15 no.2
• /
• pp.35-51
• /
• 1993

This paper presents function approximation based on nonparametric estimation. As an estimation model of function approximation, a three layered network composed of input, hidden and output layers is considered. The input and output layers have linear activation units while the hidden layer has nonlinear activation units or kernel functions which have the characteristics of bounds and locality. Using this type of network, a many-to-one function is synthesized over the domain of the input space by a number of kernel functions. In this network, we have to estimate the necessary number of kernel functions as well as the parameters associated with kernel functions. For this purpose, a new method of parameter estimation in which linear learning rule is applied between hidden and output layers while nonlinear (piecewise-linear) learning rule is applied between input and hidden layers, is considered. The linear learning rule updates the output weights between hidden and output layers based on the Linear Minimization of Mean Square Error (LMMSE) sense in the space of kernel functions while the nonlinear learning rule updates the parameters of kernel functions based on the gradient of the actual output of network with respect to the parameters (especially, the shape) of kernel functions. This approach of parameter adaptation provides near optimal values of the parameters associated with kernel functions in the sense of minimizing mean square error. As a result, the suggested nonparametric estimation provides an efficient way of function approximation from the view point of the number of kernel functions as well as learning speed.

• Journal of Korean Association for Spatial Structures
• /
• v.18 no.4
• /
• pp.69-80
• /
• 2018

In this study, an algorithm applying deep learning to the truss structures was proposed. Deep learning is a method of raising the accuracy of machine learning by creating a neural networks in a computer. Neural networks consist of input layers, hidden layers and output layers. Numerous studies have focused on the introduction of neural networks and performed under limited examples and conditions, but this study focused on two- and three-dimensional truss structures to prove the effectiveness of algorithms. and the training phase was divided into training model based on the dataset size and epochs. At these case, a specific data value was selected and the error rate was shown by comparing the actual data value with the predicted value, and the error rate decreases as the data set and the number of hidden layers increases. In consequence, it showed that it is possible to predict the result quickly and accurately without using a numerical analysis program when applying the deep learning technique to the field of structural analysis.

• Journal of KSNVE
• /
• v.6 no.6
• /
• pp.749-754
• /
• 1996

In this paper, a design method of matching layers is presented for the sandwich type broad band underwater acoustic vibrators. The characteristic impedances of matching layers are determined to be matched to the characteristic impedance of head mass material. For the dynamic characteristic analysis of the sandwich type transducers, one dimensional FEM technique is also introduced. A test vibrator with the quarter wave matching layers has been designed to verify the proposed method. And the wide band characteristics of the input impedance and transmitting voltage response (TVR) are investigated.

• Journal of the Korean Society for Precision Engineering
• /
• v.13 no.12
• /
• pp.22-29
• /
• 1996

In this paper the nodes of the multilayer hidden layers have been modified for modeling the nonlinear systems. The structure of nodes in the hidden layers is built with the feedforward, the cross talk and the recurrent connections. The feedforward links are mapping the nonlinear function and the cross talks and the recurent links memorize the dynamics of the system. The cross talks are connected between the modes in the same hidden layers and the recurrent connection has self feedback, and these two connections receive one time delayed input signals. The simplified steam boiler and the analytic multi input multi output nonlinear system which contains process noise have been modeled using this neural networks.

• Journal of Information Display
• /
• v.13 no.3
• /
• pp.101-106
• /
• 2012

This study was conducted to enhance the cubic effect by representing an image with a sense of three-dimensional (3D) depth, using chromostereopsis, among the characteristics of human visual perception. An algorithm that enhances the cubic effect, based on the theory that the cubic effect of the chromostereoptic effect and the chromostereoptic reversal effect depends on the lightness of the background, classifies the layers of the 3D image input into the foreground, middle, and background layers according to the depth of the image input. It suits the characteristics of human visual perception because it controls the color factor that was adaptively detected through experiments on each layer; and it can achieve an enhanced cubic effect that is suitable for the characteristics of the image input.

• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10a
• /
• pp.1169-1174
• /
• 1991

In conventional geographic mapping system, it is needed to input many already existing geographic map into computer system for secure and efficient maintence. Because of large map data, it is required to construct layers from map image for easy display, fast retreval and efficient storage. Thus this paper represents a method of the extracting features from contour map and constructing three layers. The layers are symbol, building, contour line. Experimental results are presented confirming the method's high extraction.

• Journal of the Ergonomics Society of Korea
• /
• v.31 no.4
• /
• pp.493-498
• /
• 2012

Objective: The purpose of this study is to suggest the hierarchical structure with three layers of input task, input interaction, and input device. Background: Understanding the input interaction is very helpful to design an interface design. Method: We made a model of three layered input structure based on empirical approach and applied to a gesture interaction in TV. Result: We categorized the input tasks into six elementary tasks which are select, position, orient, text, and quantify. The five interactions described in this paper could accomplish the full range of input interaction, although the criteria for classification were not consistent. We analyzed the Microsoft kinect with this structure. Conclusion: The input interactions of command, 4 way, cursor, touch, and intelligence are basic interaction structure to understanding input system. Application: It is expected the model can be used to design a new input interaction and user interface.

• Journal of the Korea Society of Computer and Information
• /
• v.21 no.1
• /
• pp.1-8
• /
• 2016

Convolutional neural network (CNN) consists of a few pairs of both convolution layer and subsampling layer. Thus it has more hidden layers than multi-layer perceptron. With the increased layers, the size of convolution mask ultimately determines the total number of weights in CNN because the mask is shared among input images. It also is an important learning factor which makes or breaks CNN's learning. Therefore, this paper proposes the best method to choose the convolution size and the number of layers for learning CNN successfully. Through our face recognition with vast learning examples, we found that the best size of convolution mask is 5 by 5 and 7 by 7, regardless of the number of layers. In addition, the CNN with two pairs of both convolution and subsampling layer is found to make the best performance as if the multi-layer perceptron having two hidden layers does.

• Journal of Korean Society of Rural Planning
• /
• v.14 no.4
• /
• pp.11-19
• /
• 2008

The layer technique is to produce many memorable scenes by generating layers of new experiences on the existing ones as it is adding the cognitive layers on to the visually seen landscape. Its need is high for places whose landscape itself influences perception, value or expression and which determines the spatial and quality standards. The existing floor plan-based design methods have failed to be useful in generating complex visual experiences. In order to maximize the aesthetical landscape experiences in landscape planning, cognitive layers are needed which complement the input of adequate cognitive elements and the inter-element relationships. Here, layers are utilized to change the arrangement of the landscape elements and coordinate the cognitive flow so that the images could be connected and imagination could occur. A case in point is Anhyun Village in Gochang where physically distinctive layers are additionally set to make a visual experience enriching. The new landscape layers discover the fact that it provides diversity in experiencing the village landscape and forming the sense of beauty and that it is deeply immersed into the daily life of the village. Meanwhile, Guwau Village in Taebaek is an example showing the usefulness of various-layer setting in landscape planning in setting effective circulation planning. That is, the bottom line is the spacing-starting where and making it stay where for a few seconds, and the visual layers. It is also critical to encourage inducing circulation so that layers of the senses stimulating five senses could intervene. Lastly, Mulgeon-ri in Namhae is a case which directly made a parallel of the physical layers of the landscape composition and the cognitive layers of the landscape experience. Artificial landscape planning is mostly about manipulating of visual traits that people feel beautiful, but the layer technique is linked to how to make experiences enriching and renewed.

• Journal of the Korean Ceramic Society
• /
• v.44 no.3 s.298
• /
• pp.169-174
• /
• 2007

TRISO coated fuel particle is one of the most important materials for hydrogen production using HTGR (high temperature gas cooled reactors). It is composed of three isotropic layers: inner pyrolytic carbon (IPyC), silicon carbide (SiC), outer pyrolytic carbon (OPyC) layers. In this study, TRISO coated fuel particle layers were deposited through CVD process in a horizontal hot wall deposition system. Also the computational simulations of input gas velocity, temperature profile and pressure in the reaction chamber were conducted with varying process variable (i.e temperature and input gas ratios). As deposition temperature increased, microstructure, chemical composition and growth behavior changed and deposition rate increased. The simulation showed that the change of reactant states affected growth rate at each position of the susceptor. The experimental results showed a close correlation with the simulation results.