• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.7 no.4
• /
• pp.594-601
• /
• 2006

Coding of occupational and industrial codes is a major operation in census survey of Korean statistics bureau. The coding process has been done manually. Such manual work is very labor and cost intensive and it usually causes inconsistent results. This paper proposes an automatic coding system based on example-based learning. The system converts natural language input into corresponding numeric codes using code generation system trained by example-based teaming after applying manually built rules. As experimental results performed with training data consisted of 400,000 records and 260 manual rules, the proposed system showed about 76.69% and 99.68% accuracy for occupational code classification and industrial code classification, respectively.

• Tribology and Lubricants
• /
• v.29 no.6
• /
• pp.397-402
• /
• 2013

This study examined the tribological characteristics of the drive shaft and axle system in vehicles. The first drive shaft example contained end play for a CV joint that transferred part of the transmission power to the wheel. The joint part of the drive shaft was deformed because of reduced durability due to wear. Thus, vibrations caused the body to shake and become unbalanced when the drive shaft transferred the power. The second example was the cross-section of a shaft that connected the slip-connection of the propeller shaft on the input side to the yoke flange of the output side; the durability was reduced because of corrosion. End play caused by wear between the bearing and cross-section shaft appeared to cause shaking. In the third example, a grease leak reduced lubrication and thus caused damage to the hub bearing and inside the knuckle. The failure was produced by sticking. The fourth example had noise produced by the gear and gear transfer. This was due to the backlash of the pinion and few ring gears for the differential gear. Therefore, drive shaft and axle systems must be thoroughly checked and managed to minimize and reduce failure phenomena.

• Journal of Korean Association for Spatial Structures
• /
• v.14 no.3
• /
• pp.47-56
• /
• 2014

This paper investigates the characteristics of unstable behaviour and critical buckling load by joint rigidity of framed large spatial structures which are sensitive to initial conditions. To distinguish the stable from the unstable, a singular point on equilibrium path and a critical buckling level are computed by the eigenvalues and determinants of the tangential stiffness matrix. For the case study, a two-free node example and a folded plate typed long span example with 325 nodes are adopted, and these adopted examples' nonlinear analysis and unstable characteristics are analyzed. The numerical results in the case of the two-free node example indicate that as the influence of snap-through is bigger; that of bifurcation buckling is lower than that of the joint rigidity as the influence of snap-through is lower. Besides, when the rigidity decreases, the critical buckling load ratio increases. These results are similar to those of the folded-typed long span example. When the buckling load ratio is 0.6 or less, the rigidity greatly increases.

• Transactions of Materials Processing
• /
• v.22 no.6
• /
• pp.323-327
• /
• 2013

In this paper, the flow lines as a function of product design as well as the forging process design are explored using typical application examples. The prediction of flow lines using metal forming simulation technology is introduced along with their characterization. Experimental studies have shown that the metal flow lines have a strong influence on the structural rigidity of the final product. In this study we present several typical applications. One example is the case of severely cut metal flow lines during machining, especially in the region where periodic contacting forces are applied. Another example is the case of abnormal distortion of flow lines which can cause too much elongation or hot shortness due to viscous heating in the region of distortion. A third example is the case of a macrosegregation region which needs to be controlled so it is not adjacent to the region where the force is applied in the use of the final component. An example of weight reduction for an automobile component with improved flow lines is also introduced. These typical applications can provide process engineers with the insight in designing automobile or mechanical components as well as in designing the manufacturing methods to produce various parts.

• Current Optics and Photonics
• /
• v.1 no.5
• /
• pp.538-543
• /
• 2017

Progressive addition lenses (PAL) have very wide application in the modern glasses market. The unique progressive surface can make a lens have progressive refractive power, which can meet the human eye's different needs for distance-vision and near-vision. According to the national glasses fabrication standard, the difference between actual optical power after fabrication and nominal design value should be less than 0.1D over the lens effective area. The optical power distribution of PAL is determined directly by the surface. Consequently, the surface processing accuracy requirement is proposed. Beginning from the surface expressions of progressive addition lenses, the relationship equations between the surface sag and optical power distribution are derived. They are demonstrated through tolerance analysis and test of an example progressive addition lens with addition of 2.09D (5.46D-7.55D). The example addition surface is fabricated under given accuracy by a single-point diamond ultra-precision machine. The optical power of the PAL example is tested with a focal-meter after fabrication. The optical power addition difference between test result and design nominal value is 0.09D, which is less than 0.1D. The derived relationship between the surface error and optical power is verified from the PAL example simulation and test result. It can provide theoretical tolerance analysis proof for the PAL surface fabricating process.

• Journal of Educational Research in Mathematics
• /
• v.13 no.2
• /
• pp.159-178
• /
• 2003

This study analyzes on the reasoning in the process of justification and mathematical problem solving in our primary mathematics textbooks. In our analyses, we found that the inductive reasoning based on the paradima-tic example whose justification is founnded en a local deductive reasoning is the most important characteristics in our textbooks. We also found that some propositions on the properties of various quadrangles impose a deductive reasoning on primary students, which is very difficult to them. The inductive reasoning based on enumeration is used in a few cases, and analogies based on the similarity between the mathematical structures and the concrete materials are frequntly found. The exposition based en a paradigmatic example, which is the most important characteristics, have a problematic aspect that the level of reasoning is relatively low In Miyazaki's or Semadeni's respects. And some propositions on quadrangles is very difficult in Piagetian respects. As a result of our study, we propose that the level of reasoning in primary mathematics is leveled up by degrees, and the increasing levels are following: empirical justification on a paradigmatic example, construction of conjecture based on the example, examination on the various examples of the conjecture's validity, construction of schema on the generality, basic experiences for the relation of implication.

• Journal of Broadcast Engineering
• /
• v.9 no.2
• /
• pp.151-163
• /
• 2004

In this paper, we propose a content-based video indexing and retrieval system using MPEG-7 standard to retrieve and manage videos efficiently. The proposed system consists of video indexing module for a video DB and video retrieval module to allow various query methods on a web environment. Video indexing module stores metadata such as manually typed in keywords, automatically recognized character names, and MPEG-7 visual descriptors extracted by indexing module into a DB in a sever side. A user can access to retrieval module by a web and retrieve desired videos through various query methods like keywords, faces, example and sketch. For this retrieval system, we propose ATC(Adaptive Twin Comparison) as a cut detection method for efficient video indexing and QBME(Query By Modified Example) as an improved content-based query method for the convenience of users. Experimental results show that the proposed ATC method detects cuts well and the proposed QBME method provides the conveniences better than existing query methods such as QBE(Query By Example) and QBS(Query By Sketch).

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.7
• /
• pp.72-80
• /
• 2018

With the ever increasing advances in computers and their computing power, computational fluid dynamics(CFD) has become an essential engineering tool in the design and analysis of engineering applications. Accordingly, many universities have developed and implemented a course on CFD for undergraduate students. On the other hand, many professors have used industrial examples supplied by computational analysis software companies as CFD examples. This makes many students think of CFD as difficult and confusing. This paper presents a simple CFD example used in the department of mechanical design engineering of Kangwon National University and shows its effectiveness. Most students answered that a simple CFD example is more comprehensive than an industrial example. Therefore, it is necessary to develop simple computational analysis problems in the engineering education field.

• Convergence Security Journal
• /
• v.20 no.2
• /
• pp.111-121
• /
• 2020

Deep neural networks (DNNs) provide good performance for machine learning tasks such as image recognition and object recognition. However, DNNs are vulnerable to an adversarial example. An adversarial example is an attack sample that causes the neural network to recognize it incorrectly by adding minimal noise to the original sample. However, the disadvantage is that it takes a long time to generate such an adversarial example. Therefore, in some cases, an attack may be necessary that quickly causes the neural network to recognize it incorrectly. In this paper, we propose a fast misclassification sample that can rapidly attack neural networks. The proposed method does not consider the distortion of the original sample when adding noise. We used MNIST and CIFAR10 as experimental data and Tensorflow as a machine learning library. Experimental results show that the fast misclassification sample generated by the proposed method can be generated with 50% and 80% reduced number of iterations for MNIST and CIFAR10, respectively, compared to the conventional Carlini method, and has 100% attack rate.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.5
• /
• pp.195-204
• /
• 2018

In this paper, we propose both a new loss function and an online hard example mining scheme for improving the performance of single-stage object detectors which use deep convolutional neural networks. The proposed loss function and the online hard example mining scheme can not only overcome the problem of imbalance between the number of annotated objects and the number of background examples, but also improve the localization accuracy of each object. Therefore, the loss function and the mining scheme can provide intrinsically fast single-stage detectors with detection performance higher than or similar to that of two-stage detectors. In experiments conducted with the PASCAL VOC 2007 benchmark dataset, we show that the proposed loss function and the online hard example mining scheme can improve the performance of single-stage object detectors.