• International Journal of Contents
• /
• v.10 no.1
• /
• pp.23-28
• /
• 2014

When developing a classifier using various objective functions, it is important to compare the performances of the classifiers. Although there are statistical analyses of objective functions for classifiers, simulation results can provide us with direct comparison results and in this case, a comparison criterion is considerably critical. A Receiver Operating Characteristics (ROC) graph is a simulation technique for comparing classifiers and selecting a better one based on a performance. In this paper, we adopt the ROC graph to compare classifiers trained by mean-squared error, cross-entropy error, classification figure of merit, and the n-th order extension of cross-entropy error functions. After the training of feed-forward neural networks using the CEDAR database, the ROC graphs are plotted to help us identify which objective function is better.

• Journal of Industrial Technology
• /
• v.16
• /
• pp.123-129
• /
• 1996

A method of calculating the natural frequency corresponding to the first mode of vibration of beams and tower structures with irregular cross-sections and with arbitrary boundary conditions was developed and reported by D. H. Kim in 1974. In order to obtain the natural frequency by the presented method, the first step to take is to obtain the deflection influence surfaces. In design and analysis of any structure, the first step to be taken is to obtain this deflection influence surfaces. Any method can be used for this purpose. Then using this surfaces, deflections, slopes, moments, shears, and natural frequencies can be obtained by differentiating either the continuous function or discrete function defined at certain points. The merit of the presented method is that the natural frequency can be obtained by the deflection influence surfaces obtaining which is the first step in structural analysis.

• Clinics in Shoulder and Elbow
• /
• v.5 no.1
• /
• pp.23-28
• /
• 2002

Purpose Displaced lateral end fracture of clavicle and acromioclavicular dislocation type Ⅲ∼Ⅵ may be required surgical treatment. Material and Methods : From May 1998 to March 2001, we operated with Welter plate with minimal incision by one surgeon. Immediately, pendulum and passive exercise was initiated after surgery. The shoulder function was evaluated using UCLA score. Average follow up was on 28(12∼45) months. Results : All 11 patients were regained satisfactory function. Average UCLA score was 31.9(29~35) at last follow up. Conclusion . The merit of Welter plate fixation with minimal incision is simple technique, reduced surgical time, smaller scar than large plate, strong fixation, early exercise, reduced implant failure. The disadvantage is expensive, skin irritatatation by long hook. But Welter plate fixation with minimal incision is a good method of internal fixation and excellent clinical result in surgical treatment of type H displaced lateral end fracture of the clavicle and type Ⅲ∼Ⅵ acromioclavicular dislocation.

• International Journal of Contents
• /
• v.8 no.4
• /
• pp.30-35
• /
• 2012

Error surfaces provide us with very important information for training of feed-forward neural networks (FNNs). In this paper, we draw the contour plots of various error or objective functions for training of FNNs. Firstly, when applying FNNs to classifications, the weakness of mean-squared error is explained with the viewpoint of error contour plot. And the classification figure of merit, mean log-square error, cross-entropy error, and n-th order extension of cross-entropy error objective functions are considered for the contour plots. Also, the recently proposed target node method is explained with the viewpoint of contour plot. Based on the contour plots, we can explain characteristics of various error or objective functions when training of FNNs proceeds.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.32 no.5C
• /
• pp.530-535
• /
• 2007

Disparity space image (DSI) technique is a method of establishing correspondence between a pair of images. It has a merit of generating a dense disparity map for each pixel. DSI has a cost function to be minimized, and it needs empirical weighting factors for occlusion penalty and match reward. This paper provides theoretical basis for the weighting factors, which depend on image noise and contrast between an object and background.

• Journal of the Korean Society of Costume
• /
• v.57 no.3 s.112
• /
• pp.23-36
• /
• 2007

It is important that ubiquitous technology changes paradigm of thought, not simple definition in the 21st digital era. Characteristics of ubiquitous computing are pervasive, disappearing, invisible, calm through environment. As IT Technology develops, designers, computer scientists, chemists, performance artists cooperate in order to find out the best way to make desirable digital clothing in the future, with the merit of each part. Digital clothing defines clothes of new generation equipped computer, digital installations. Digital clothing design demands intercept of electromagnetic waves, light-weight and esthetic appearance, for it is attached high-technology equipment near body. The purpose of this study is to analyze design features of digital clothing according to ubiquitous characteristics. The methods of this study are documentary research of previous study and case study. In the theoretical study, ubiquitous characteristics are function-intensive by convergence, interactivity, embedded mobility and human & emotion-oriented attributes. Based on ubiquitous characteristics, digital clothing design classified function-intensive design by convergence, design for Interactivity and multi-sensible & emotion-oriented design, because embedded mobility is a basic element of ubiquitous environment. The early days digital clothing design is function-intensive design, and have esthetic appearances and design for interactivity increasingly. Recently digital clothing design is expressed multi-sensible and emotion-oriented design.

• Journal of the Korean Magnetics Society
• /
• v.10 no.3
• /
• pp.99-105
• /
• 2000

One dimensional magneto-photonic crystal having structure of (A/B)$^{k/W}$(B/A)$^{k}$ , where M is a magnetic layer of highly Bi-substituted iron garnet, A and B are dielectric layers of $SiO_2$ and T $a_2$O$_{5/}$, and k is the stacking number of the dielectric layers, has been numerically analyzed as a function of the thickness (d$_{M}$) of M (1∼535 nm) and the stacking numer of k (5∼15). The transmittance, Faraday rotation, and figure of merit of the magneto-photonic crystal have been investigated both in the visible and infrared wavelengths. A factor of several and several tens greater Faraday rotation and figure of merit have been obtained compared to the single layer of M, at many localized modes. In the visible the maximum figure of merit of 0.15 was obtained ( = 720 nm) when k = 11 and d$_{M}$ = 375 nm with T : 0.54, $\theta$$_{F}$ = 8.13$^{\circ}$, which was a factor of 30 greater than that of single garnet layer. Much greater maximum figure of merit, 0.285, was obtained in the infrared ( = 1114 nm) when k = 11 and d$_{M}$ = 800 nm with T = 0.66, $\theta$$_{F}$ = 18$^{\circ}$, which was a factor of 100 greater than that of single garnet layer.

• Journal of the Korean Society for Precision Engineering
• /
• v.19 no.6
• /
• pp.77-85
• /
• 2002

An analysis method which calculates corrective machining information for improving the motion accuracy of linear motion guide Is proposed in this paper. The method is composed of two algorithms. One is the algorithm fur prediction of the motion errors from rail form error. The other is the algorithm for prediction of rail form error from the motion errors of table. Transfer function is utilized in each algorithm, which represents the ratio of bearing reaction force variation to unit magnitude of spatial frequencies of raid from error. As the corrective machining information is acquired from the measured motion errors of table, the method has a merit not to measure rail form error directly. Validity of the method is verified both theoretically and experimentally. By applying the method, linear motion error of test equipment is reduced from 5.97$\mu$m to 0.58$\mu$m, and reduced from 32.78arcsec to 6.21 arcsec in case of angular motion error. From the results, it is confirmed that the method is very effective to improve the motion accuracy of linear motion guide.

• Transactions on Electrical and Electronic Materials
• /
• v.14 no.5
• /
• pp.242-245
• /
• 2013

Sn-doped $In_2O_3$ (ITO) thin films were prepared by radio frequency magnetron sputtering without intentional substrate heating on bare glass and $TiO_2$-deposited glass substrates to investigate the effect of a $TiO_2$ buffer layer on the electrical and optical properties of ITO films. The thicknesses of $TiO_2$ and ITO films were kept constant at 5 and 100 nm, respectively. As-deposited ITO single layer films show an optical transmittance of 75.9%, while $ITO/TiO_2$ bi-layered films show a lower transmittance of 76.1%. However, as-deposited $ITO/TiO_2$ films show a lower resistivity ($9.87{\times}10^{-4}{\Omega}cm$) than that of ITO single layer films. In addition, the work function of the ITO film is affected by the $TiO_2$ buffer layer, with the $ITO/TiO_2$ films having a higher work-function (5.0 eV) than that of the ITO single layer films. The experimental results indicate that a 5-nm-thick $TiO_2$ buffer layer on the $ITO/TiO_2$ films results in better performance than conventional ITO single layer films.

• Journal of the Korean Vacuum Society
• /
• v.6 no.1
• /
• pp.69-76
• /
• 1997

P-type ($Bi_{0.5}Sb_{1.5}Te_3$) and n-type ($Bi_2Te_{2.4} Se_{0.6}$) thermoelectric thin film were deposited on glass and Teflon substrates by the flash evaporation technique. The changes in thermoelectric properties, such as Seebeck coefficient, electrical conductivity, carrier concentration, carrier mobility, thermal conductivity, and figure of merit, were investigated as a function of film thickness and annealing condition. Figures of merit of the thin films annealed at 473 K for 1 hour were improved to be $1.3{\times}10^{-3}K^{-1}$ for p-type and $0.3{\times}10^{-3}K^{-1}$ for n-type, and they were almost independent of film thickness. Temperature sensors were fabricated from the thin films having the above mentioned properties. And thermo-emf, sensitivity, and time constant of the sensors were measured to evaluate their characteristics for temperature sensors. Thin film sensors deposited on Teflon substrates showed better performance than those on glass substrates, and their sensitivity and time constant were 2.91 V/W and 28.2 sec respectively for the sensor of leg width 1 mm$\times$length 16 mm.