• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.10
• /
• pp.75-80
• /
• 2017

In this paper, we describe the effect of the regularization method and the network with identity mapping on the performance of the activation functions in deep convolutional neural networks. The activation functions act as nonlinear transformation. In early convolutional neural networks, a sigmoid function was used. To overcome the problem of the existing activation functions such as gradient vanishing, various activation functions were developed such as ReLU, Leaky ReLU, parametric ReLU, and ELU. To solve the overfitting problem, regularization methods such as dropout and batch normalization were developed on the sidelines of the activation functions. Additionally, data augmentation is usually applied to deep learning to avoid overfitting. The activation functions mentioned above have different characteristics, but the new regularization method and the network with identity mapping were validated only using ReLU. Therefore, we have experimentally shown the effect of the regularization method and the network with identity mapping on the performance of the activation functions. Through this analysis, we have presented the tendency of the performance of activation functions according to regularization and identity mapping. These results will reduce the number of training trials to find the best activation function.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.41 no.6
• /
• pp.672-677
• /
• 2016

Development of Internet has been successfully inspired with extensive deployment of the network technology and application. However, increases in Internet usage had caused a lot of traffic overload in these days. Thus, we need a continuous research and development on the network virtualization for effective resource allocation. In this paper, we propose a minimal cost virtual network mapping algorithm using Piecewise Linear Cost Function. We exploited an algorithm with Linear Programming and D-VINE for node mapping, and Shortest Path Algorithm based on linear programming solution is used for link mapping. In this way, we compared and analyzed the average cost for arrival rate of VN request with linear and tree structure. Simulation results show that the average cost of our algorithm shows better efficiency than ViNEyard.

• Journal of Broadcast Engineering
• /
• v.23 no.2
• /
• pp.283-292
• /
• 2018

Devices capable of capturing and displaying high dynamic range (HDR) images with significantly increased brightness range compared to low dynamic range (LDR) images have been developed and various methods for efficiently converting the brightness range of an image have been developed. In this paper, we propose a reverse tone mapping method using a guided filter to efficiently convert LDR images into HDR images. After obtaining brightness enhancement function (BEF) using a guided filter, we can reconstruct HDR image from one LDR image. In addition, when the image is too bright or dark, the proposed method maximizes the image quality of the reconstructed HDR image by estimating and adjusting the exposure value before expanding the brightness range of images. Computer simulations show that the proposed method produces HDR images of superior quality compared with the conventional methods.

• Bulletin of the Korean Mathematical Society
• /
• v.46 no.5
• /
• pp.1013-1018
• /
• 2009

Let * be an e.a.b. star operation on an integrally closed domain D, and let $K\gamma$(D, *) be the Kronecker function ring of D. We show that if D is a P*MD, then the mapping $D_{\alpha}{\mapsto}K{\gamma}(D_{\alpha},\;{\upsilon})$ is a bijection from the set {$D_{\alpha}$} of *-linked overrings of D into the set of overrings of $K{\gamma}(D,\;{\upsilon})$. This is a generalization of [5, Proposition 32.19] that if D is a Pr$\ddot{u}$fer domain, then the mapping $D_{\alpha}{\mapsto}K_{\gamma}(D_{\alpha},\;b)$ is a one-to-one mapping from the set {$D_{\alpha}$} of overrings of D onto the set of overrings of $K_{\gamma}$(D, b).

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.21 no.12
• /
• pp.1973-1983
• /
• 1997

Higher order isoparametric elements are usually used in the finite element analysis because they can represent easily the geometric shape of a complex structure ad can improve the solution quality. When these elements are used, the position of internal nodes affects greatly on the solution accuracy. Decreasing of the accuracy related to the position of internal nodes is due to non-conformal mapping often results in an unstable Jacobian value. This paper, in order to remove this difficulty, suggests a modified shape function which can establish conformal mapping between two coordinate systems. Numerical experiments with the proposed shape function show that a stable solution can be obtained without respect to the position of internal nodes, and offer convenience that one can take arbitrarily the position of internal nodes considering only the geometric shape of element boundaries.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.4
• /
• pp.1846-1863
• /
• 2016

High dynamic range (HDR) images can present the perfect real scene and rich color information. A commonly encountered problem in practical applications is how to well visualize HDR images on standard display devices. In this paper, we propose a multi-scale decomposition method using guided filtering for HDR image tone mapping. In our algorithm, HDR images are directly decomposed into three layers：base layer, coarse scale detail layer and fine detail layer. We propose an effective function to compress the base layer and the coarse scale detail layer. An adaptive function is also proposed for detail adjustment. Experimental results show that the proposed algorithm effectively accomplishes dynamic range compression and maintains good global contrast as well as local contrast. It also presents more image details and keeps high color saturation.

• Journal of Positioning, Navigation, and Timing
• /
• v.10 no.4
• /
• pp.371-377
• /
• 2021

This research proposes the best combination of tropospheric delay models for Korean Positioning System (KPS). The overall results are based on real observation data of Japanese Quasi-Zenith satellite system (QZSS), whose constellation is similar to the proposed constellation of KPS. The tropospheric delay models are constructed as the combinations of three types of zenith path delay (ZPD) models and four types of mapping functions (MFs). Two sets of International GNSS Service (IGS) stations with the same receiver are considered. Comparison of observation residuals reveals that the ZPD models are more influential to the measurement model rather than MFs, and that the best tropospheric delay model is the combination of GPT3 with 5 degrees grid and Vienna Mapping Function 1 (VMF1). While the bias of observation residual depends on the receivers, it still remains to be further analyzed.

• Nonlinear Functional Analysis and Applications
• /
• v.27 no.4
• /
• pp.757-771
• /
• 2022

In this paper, we present some fixed point theorems for rational type contractive conditions in the setting of a complete metric space via a cyclic (𝛼, 𝛽)-admissible mapping imbedded in simulation function. Our results extend and generalize some previous works from the existing literature. We also give some examples to illustrate the obtained results.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.38A no.8
• /
• pp.669-679
• /
• 2013

When an image is converted from HDR (high dynamic range) to LDR (low dynamic range), a tone mapping process is the essential component. Many TMOs (tone mapping operators) have been motivated by human vision which has lower physical luminance range than that in real scene. The representative of human vision properties which motivate TMOs is the local adaptation. However, TMOs are ultimately compressing image information such as contrast, saturation, etc. and the compression causes defects in image quality. In this paper, in order to compensate the degradation of the image which is caused by TMOs, the visual acuity-based edge stop function is proposed for applying the property of human vision to base-detail separation. In addition, using CSF (contrast sensitivity function) which represents the relationship among spatial frequency, contrast sensitivity, and luminance, the sharpness filter is designed and adaptively applied to the detail layer in regard to surround luminance.

• Journal of the Korean Mathematical Society
• /
• v.48 no.3
• /
• pp.641-653
• /
• 2011

The intermediate value theorem for a continuous real valued function is a kind of Bolzano's theorem. Similar results also hold for compact, monotone or accretive mappings in Banach spaces. In this paper we give multivalued versions of Bolzano's theorem.