• Journal of Electrical Engineering and Technology
• /
• v.13 no.5
• /
• pp.2099-2106
• /
• 2018

This study presents a novel approach of discriminative feature vectors based on manifold learning using nonlinear dimension reduction (DR) technique to improve loss function, and combine with the Adversarial examples to regularize the object function for image classification. The traditional convolutional neural networks (CNN) with many new regularization approach has been successfully used for image classification tasks, and it achieved good results, hence it costs a lot of Calculated spacing and timing. Significantly, distrinct from traditional CNN, we discriminate the feature vectors for objects without empirically-tuned parameter, these Discriminative features intend to remain the lower-dimensional relationship corresponding high-dimension manifold after projecting the image feature vectors from high-dimension to lower-dimension, and we optimize the constrains of the preserving local features based on manifold, which narrow the mapped feature information from the same class and push different class away. Using Adversarial examples, improved loss function with additional regularization term intends to boost the Robustness and generalization of neural network. experimental results indicate that the approach based on discriminative feature of manifold learning is not only valid, but also more efficient in image classification tasks. Furthermore, the proposed approach achieves competitive classification performances for three benchmark datasets : MNIST, CIFAR-10, SVHN.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 1999.11a
• /
• pp.53-59
• /
• 1999

A Patent Pool tends to be activated to establish the harmonization in recently standardization activities related to information and telecommunications with patent rights. Recent main examples of patent pool related to telecommunications standardization are concentrated in this paper. We seem to examine the practical meanings of patent pool from review the examples of utilizing patent pool. Those examples are analyzed and presented in this paper as typical construction cases of patent pool by characteristic of formation : The MPEG-2 patent pool which has become a pioneering precedence being managed by establishing MPEG LA as professional and administrative entity having organizied scalef; DVD patent pool which is a type of dividied case of a patent pool; the patent pool of IEEE1394 which was formed by surronding technology market effect outside the entrepreneur; finally the UMTS case in mobile telecommunications area.

• Journal of the Korean School Mathematics Society
• /
• v.13 no.2
• /
• pp.263-287
• /
• 2010

The purposes of the study were to investigate how children define a triangle, their reasoning types in identifying examples and non-examples of a triangle, and the relationship between their reasoning types and geometrical levels. Twenty-nine students consisted of 3th to 6th grades were involved in the study. Using the van Hiele levels of geometrical thought, children's reasoning types for identifying a figure as a triangle or non-triangle were categorized into visual reasoning, reasoning based on the figure's attributes and formal reasoning. The figure's attributes were further divided into critical and non-critical attributes. Most children identified a figure as a triangle or non-triangle based on critical attributes of the figure(e.g. closed figure, three, vertices, straight sides etc.) Some children identified a figure based on non-critical attributes of the figure(e.g. the length of the sides, the measurement of the angles, or the orientation of the figure). Particularly, some children who had lower levels of geometry identified a figure using visual reasoning, taking in the whole shape without considering that the shape is made up of separate components.

• Korean Institute of Interior Design Journal
• /
• v.22 no.2
• /
• pp.203-210
• /
• 2013

The purpose of this study was to investigate and analyze the function and scale on aspect of construction planning with examples of exhibition and convention facilities constructed recently in domestic and foreign area. Also, it was to provide the result for basic data for construction planning and architecture of similar facilities constructed later. The study methodles, but they had characteristics to use the space with compound purpose for exhibition, conference, etc through installing multi-purpose hall. Finally, for effective construction planning of exhibition and convention facilities, it was required to decrease the ratio of common area gradually and derive rationalization of whole space composition by deriving complexity of function rather than planning of single purpose of each facility for exhibition and conference function. was to analyze the characteristics of construction planning for composition and scale of exhibition, conference, public use, support, maintenance functions, etc and placement composition of unit rooms by each floor on the basis of internal area of exhibition and convention facility in Korea and Japan. As the result of analyzing 10 examples(K1~J5), the placement of auditorium and multi-purpose hall was affected according to location of exhibition room and conference room in the composition of unit rooms by each floor and the installment ratio of support facility in Korean examples was higher than in Japanese examples. In case of facility scale, the Japanese examples were overall small in the whole scale relatively to Korean examp.

• Cross-Cultural Studies
• /
• v.25
• /
• pp.237-265
• /
• 2011

This thesis is to argue what is the meaning of the expression of cherry blossoms appearing in "Genjimonogatari(源氏物語)" in relation with the structure of Monogatari after checking each example. First we reviewed some examples of the works before "Genjimonogatari", and clarified that cherry blossoms were used as metaphorical expressions of a beautiful woman in waka(和歌), and that at the same time cherry blossoms functioned as the symbol of futility, anxiety together with beauty. Especially in "Kokinshu(古今集)" we identified that there increased the examples where the combination of the expression as falling cherry blossoms, and fog and cherry blossoms was done. In "Genjimonogatari" we examined the examples of cherry blossoms of Chapter 1, Chapter 2, and clarified that Murasakinoue(紫の上) was a woman with an ideal character in looks and inner personality through the diverse viewpoints of characters. In Monogatari of Kashiwagi(柏木) of Chapter 2, the scene where he peeped Onnasannomiya(女三宮) was composed by the combination of cherry blossoms and fog, and we pointed out that we could feel the anxiety, ominousness of the love of the couple through this scene. And the composition of these scenes was also repeated in Monogatari of Chapter 3, and we discussed through these examples that the figurative expressions of cherry blossoms were used about the object of the love not to be realized. Next, we checked that Yugiri(夕霧) who had observed Kashiwagi's love toward Onnasannomiya gave a figurative expression of Kashiwagi's wife Ochibanomiya(落葉の宮) as cherry blossoms. Although Genji(源氏) realized Onnasannomiya's immaturity, Kashiwagi compared her to "cherry blossoms", and Yoogiri compared the woman whom Kashiwagi recognized as "Ochiba(落葉)" to "cherry blossoms", through which we have read the twisted human relation and ironical structure of Chaper 2. Finally, we checked the examples of cherry blossoms in Ooigimi(大君), Nakanokimi(中君), Ukihune(浮舟) who were Ujijujo's heroines. Ooigimi and Nakanokimi were expressed as cherry blossoms by Kaoru(?) and Nioumiya(?宮) and were disclosed as their ideal women. In contrast, Ukihune had something to do with two men with no such figurative expression, so it was pointed out that the absence of such metaphorical expression clarified that the meaning of her existence was nothing but the meaning as a mesiudo(召人).

• Journal of the Korean School Mathematics Society
• /
• v.4 no.2
• /
• pp.1-9
• /
• 2001

Since the computer is becoming more and more indispensible tool in every fields of the modern society, it is needed and desirable to utilize the computer as a basic tool from the very early stage of the education process. Recently Maple is gaining its popularity as a comprehensive mathematical software with its power of symbolic calculations and graphics as well as its great numerical computational ability. We demonstrate the suitability of this software as a tool for the mathematical education and presents several examples of the applications of Maple. For the middle and the high school mathematics courses, we give the application examples for the quadratic functions and their graphs, statistics, the three dimensional shapes, algebraic problems. Through the examples, we confirm that mathematical education can be much more effective and simple by using Maple. If we establish computer-assisted mathematical classes, we can draw more attention and excitement from the attendants than traditional classes and eventually improve more rapidly their problem-solving ability On the other hand, the excess of the computer-aided education give to obstacle of psychological, not to be passing over the fact.

• Journal of Educational Research in Mathematics
• /
• v.8 no.1
• /
• pp.269-278
• /
• 1998

This paper is composed of computer programs and practical examples for the education of mathematics. The Animation package in the Mathematica standard packages is the Graphics Animation. Animations are the ways which produce a sequence of different pictures by rapid succession. The command Animate has a number of options such as Animate, SpinShow and ShowAnimation. Mathematica is available for a large variety of practical education of mathematics. The method in this report are familiar with the general principles of operating the computer in Mathematica 3.0. In this paper there a set of useful examples of mathematical education and we discusses some of the advanced technique for the Input-Output systems.

• 한국전산유체공학회:학술대회논문집
• /
• 2004.03a
• /
• pp.71-76
• /
• 2004

An extension of our recently developed locally linear reconstruction scheme to 2 dimensional incompressible flow solver is presented. The solver is based on a semi-implicit fractional step method in which the convective term is discretized by Adams-Bashforth method and the diffusion term by Crank-Nicolson method. Several numerical examples are tested to demonstrate the mesh type independent accuracy of the solver, which include decaying vortex flow, square cavity flow, and flow around a circular cylinder. The above examples are solved on quadrilateral or hybrid meshes. For all numerical examples, we obtained reasonable results.

• Journal of the Korean Mathematical Society
• /
• v.57 no.5
• /
• pp.1187-1237
• /
• 2020

Simply reducible groups are important in physics and chemistry, which contain some of the important groups in condensed matter physics and crystal symmetry. By studying the group structures and irreducible representations, we find some new examples of simply reducible groups, namely, dihedral groups, some point groups, some dicyclic groups, generalized quaternion groups, Heisenberg groups over prime field of characteristic 2, some Clifford groups, and some Coxeter groups. We give the precise decompositions of product of irreducible characters of dihedral groups, Heisenberg groups, and some Coxeter groups, giving the Clebsch-Gordan coefficients for these groups. To verify some of our results, we use the computer algebra systems GAP and SAGE to construct and get the character tables of some examples.

• Journal of information and communication convergence engineering
• /
• v.16 no.3
• /
• pp.148-152
• /
• 2018

Adversarial attacks on artificial intelligence (AI) systems use adversarial examples to achieve the attack objective. Adversarial examples consist of slightly changed test data, causing AI systems to make false decisions on these examples. When used as a tool for attacking AI systems, this can lead to disastrous results. In this paper, we propose an ensemble of degraded convolutional neural network (CNN) modules, which is more robust to adversarial attacks than conventional CNNs. Each module is trained on degraded images. During testing, images are degraded using various degradation methods, and a final decision is made utilizing a one-hot encoding vector that is obtained by summing up all the output vectors of the modules. Experimental results show that the proposed ensemble network is more resilient to adversarial attacks than conventional networks, while the accuracies for normal images are similar.