• Journal of KIISE
• /
• v.45 no.1
• /
• pp.22-29
• /
• 2018

Over the decades, neural networks have been successfully used in numerous applications from speech recognition to image classification. However, these neural networks cannot explain their results and one needs to know how and why a specific conclusion was drawn. Most studies focus on extracting binary rules from neural networks, which is often impractical to do, since data sets used for machine learning applications contain continuous values. To fill the gap, this paper presents an algorithm to extract logic rules from a trained neural network for data with continuous attributes. It uses hyperplane-based linear classifiers to extract rules with numeric values from trained weights between input and hidden layers and then combines these classifiers with binary rules learned from hidden and output layers to form non-linear classification rules. Experiments with different datasets show that the proposed approach can accurately extract logical rules for data with nonlinear continuous attributes.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.16 no.2
• /
• pp.149-158
• /
• 1998

The aim of this paper is to propose the technique of the two dimensional geodetic coordinates conversion between WGS84 and Bessel spheroids by the two-dimensional affine transformation modeling based on the UTM plane coordinates without the local geoid model which is essential to three dimensional coordinates conversion. Area of approximately $25\times{11}$ square km in the city of Inchon was selected as the test area. The transformation parameters were determined using the eleven triangulation control points in test area. And then, those paraneters were applied to the fifteen cadastral control points which selected as checking points for precision checking of transformation parameters. The average and standard deviations of the absolute values of the conversion residuals of checking points in latitude/longitude and N/E(UTM) and/or x/y(TM) are $\pm0.006"$$\pm0.013"$ and $\pm{17cm/}\pm{30cm}$ respectively. Also, coefficients for 7-parameters, 3-parameters and UTM model transformation computed according as sizes of transformed area, and then the transformed characteristics of checking points according to transformation methods analyzed synthetically.hetically.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.16 no.7
• /
• pp.4405-4410
• /
• 2015

Two-dimensional displacements are obtained from the sandwiched patterns, which superpose two linearly-periodic patterns orthogonally, respectively. The transparent top pattern is identified by deflection of the laser beam due to a difference of refractivity and the opaque bottom pattern is identified by deviation of the beam intensity due to a difference of reflectance. In the sample setup, the top pattern made up of build-up film is manufactured by UV laser machining and the bottom pattern is manufactured by ultra-precision trench machining and deposition for aluminum plate. The proposed decoding method is verified experimentally using the $10{\mu}m$ equally spaced sample patterns and the devised optical system. The Korea Academia-Industrial cooperation Society.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.33 no.10
• /
• pp.98-103
• /
• 2005

The Flight Model of MAC (Medium-sized Aperture Camera), a 2.5m GSD class earth observation camera has been aligned and assembled. Topics discussed in this paper include the ground MTF performance of the MAC system, and the alignment of the focal plane assembly. MTF was measured by a knife-edge scanning technique, and a 450 mm diameter Cassegrain collimator with diffraction-limited performance was made and used for the MTF measurements. System MTF was used as the figure-of-merit to find the best focus of the focal plane assembly.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.26 no.7
• /
• pp.106-111
• /
• 1989

A new EMG signal processing algorithm for SMUAP pattern classification is proposed. It checks the combination and regularity of ISI using a spike counter as a decision making routine, and performs SMUAP waveform alignment in frequency domain and selects spikes through FIR filtering. As a result, with the EMG signals recorded during 5 seconds at 10-50% MVC force level, the SMUAP ranged from five to nine units were classified and identification rate is greater than 55 percent using a concentric needle electrode. In the IBM PC/AT the processing time typically required 2 minutes.

• Journal of the Korean Society for Precision Engineering
• /
• v.10 no.1
• /
• pp.42-51
• /
• 1993

Recently, WC-Co have some excellent properities as the material for the mechanical component such as metallic moulding parts, ball dies parts, and punch parts. This paper describes the surface roughness and grinding force caused by experimental study on the surface grinding of WC-Co with ultra-precision like a mirror shape using diamond wheel. Also, some investigations are carried out using WA grinding wheel to increase improved ground surface roughness such as polishing, lapping effect. Some important results obtained here are summarized as follow. 1) Within this experimental grinding condition, we can be obtained $R_{max}.\;2\mu\textrm{m}\;R_a\;0.3\mu\textrm{m}$ whichare the most favourable ground surface roughness using #140 diamond wheel, and improved surface roughness values about 20 .approx. 25% throughout 5 times sparkout grinding 2) The value of surface roughness is Rmax. $0.49\mu\textrm{m},\;R_a\;0.06\mu\textrm{m}$ using #600 diamond wheel. 3) The area of no rack zone is less than $F_{n}$ 0.27N/mm, Ft 0.009N/mm

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2003.05b
• /
• pp.136-139
• /
• 2003

Thermal imaging system is electro-optical imaging device which can make visible the difference of infrared energy naturally emitted by objects. It is acquire the same images at any time of the day or night. There it has been readily available to the night observation such as fire control systems. In this study, we are manufacturing thermal image Al flat mirror. The surface roughness 3.539nm Ra and power 0.382 fringe(at 632.8nm), irregularity 0.835 fringe(at 632.8nm) for form waviness of thermal image Al flat mirror are very satisfied. The results will be reflected for development of the ultra precision application. And a brief review of Ultra-precision system in the field of Ultra-precision at Korea photonics technology institute (KOPTI) is present in this paper.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.11 no.9
• /
• pp.817-821
• /
• 2001

In this paper, a new learning methodology for kernel methods that results in a sparse representation of kernel space from the training patterns for classification problems is suggested. Among the traditional algorithms of linear discriminant function, this paper shows that the relaxation procedure can obtain the maximum margin separating hyperplane of linearly separable pattern classification problem as SVM(Support Vector Machine) classifier does. The original relaxation method gives only the necessary condition of SV patterns. We suggest the sufficient condition to identify the SV patterns in the learning epoches. For sequential learning of kernel methods, extended SVM and kernel discriminant function are defined. Systematic derivation of learning algorithm is introduced. Experiment results show the new methods have the higher or equivalent performance compared to the conventional approach.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2016.05a
• /
• pp.57-61
• /
• 2016

교각 주위에서의 국부세굴은 교각을 지나는 유체의 복잡한 흐름에 의해 발생한다. 이를 해석하기 위하여 많은 난류모형을 이용한 실내실험 및 수치실험을 수행하였으나 발생하는 와류를 하천 규모에서 전부 계산하기는 매우 어려운 문제다. 따라서 국부세굴 관련으로 최대 관심사인 최대 세굴심은 인공지능 기술에 근거한 다양한 기법을 적용해 계산하여 예측하기도 한다. 본 연구에서는 기계학습 분야 중 하나인 서포트 벡터 머신 (Support Vector Machines)을 이용하여 교각주위 국부세굴을 예측하였다. SVM은 본래 초평면을 이용하여 데이터를 분류시키는 기법이나 Vapnik(1995)이 제안한 ${\varepsilon}$ 서포트 벡터 회귀 (${\varepsilon}$-support vector regression)방법을 통해 회귀분석에도 활용할 수 있게 되었다. 학습을 위해 Charbert and Engeldinger (1956), Shen et al. (1969), Jain and Fischer (1979), 그리고 Dey et al. (1995)의 실험 자료를 이용하였고 검증을 위해 Yanmaz and Altinbilek (1991)의 실험 자료를 이용하였다. 커널함수로는 다항식 함수와 방사 기저 함수를 이용하였고 각 계수는 적합한 값을 찾기 위해 시행착오법을 사용하였다. 민감도 분석을 통해 각 계수들 중 ${\varepsilon}$의 변화가 결과에 가장 민감하게 변화를 일으키는 것을 확인하였고 검증 결과 SVM가 충분히 국부세굴을 잘 예측하는 것을 확인하였다.

• Investigative Magnetic Resonance Imaging
• /
• v.7 no.1
• /
• pp.22-30
• /
• 2003

Purpose : The efficacy of magnetic resonance imaging for evaluating coronary artery disease has been reported. In this study, we evaluated the usefulness of breath-hold segmented K-space cine MR imaging for evaluating the patency of coronary artery bypass grafts (CABG). Materials and Method s : Thirty eight patients with a total of 92 CABGs (36 internal thoracic arteries and 56 saphenous vein grafts) were evaluated using segmented K-space cardiac-gated fast gradient echo sequence (2D-FASTCARD) MR imaging. MR magnitude images were evaluated from the hard copies by two independent observers. A graft was defined as patent if it was seen as a bright small round area on at least two consecutive images throughout the cardiac cycle at a position consistent with the expected location for that graft. Results : MR images were obtained successfully for 23 patients (61%). The sagittal planes were most helpful in visualizing the cross-section of sapheneous vein bypass graft to left circumflex artery branch, whereas the transverse planes were used for identification of internal mammary artery grafts to left anterior descending coronary artery or its branch and identification of saphenous vein grafts to right coronary artery. Forty five grafts were visible using this MR technique, while the grafts were not visible on seven saphenous vein grafts and two internal mammary artery grafts. In two patients showing symptoms of myocardial ischemia, one or two bypass grafts were not visible. Imaging, perpendicular plane to a CABG was important to visualize the flow inside the CABG with maximum sensitivity. Conclusion : Evaluation of patency of the bypass graft was clinically feasible by 2D-FASTCARD MR imaging, whereas any invisible bypass grafts should be further studied by contrast-enhanced MR angiography or by conventional angiography for confirmation of abnormalities.