• 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.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.11 no.5
• /
• pp.38-44
• /
• 2002

An experimental investigation was performed to observe the cooling performance of cutting oil and the effect of inclination angle on the transport behaviour of a inclined thermosyphons with low integral-fins. Relatively high rates of heat transfer have been achieved by operating this manner. Water has been used as the working fluid. The inclimation angle of thermoryphon and the ratio of Reynolds number of cooling to hot fluid have been used as the experimental parameters. Heat transfer coefficients at the condenser and the evaporator are estimated from experimental results. The experimental results have been assessed and compared with existing theory. Good agreement with the theory of Yiwei has been found. The maximum heat flow rate in the thermosyphon proved to depend upon the inclination angle.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.12 no.6
• /
• pp.14-21
• /
• 2003

With the increased use of lasers in industrial welding applications, techniques for monitoring and controlling these processes become increasingly important. It is very important that we understand the dynamic behaviors of the laser induced Plume in welding, because the laser induced plume has considerable effects on welding efficiency and the quality of materials. As the plume fluctuation was associated with keyhole instability, unstable vapor plume indicated the process was unstable and would result in poor welds. An Infrared Thermal-vision Camera can be utilized compensate for incurracies encountered in real-time monitoring during laser welding. We have results that instabilities of plume are closely related with hot cracking and defect of laser welding.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 2001.04a
• /
• pp.72-77
• /
• 2001

The use of hexagonal ceramic inserts for copper extrusion dies offers significant technical and economic advantages over other forms of manufacture. In this paper the data on the loading of the tools is determined from a commercial FEM package as the contact stress distribution on the die-workpiece interface and as temperature distributions in the die. This data can be processed as load input data for a finite element die-stress analysis. Process simulation and stress analysis are thus combined during the design, and a data exchange program has been developed that enables optimal design of the dies taking into account the elastic deflections generated in shrink fitting the die inserts and that caused by the stresses generated in the process. The stress analysis of the dies is used to determine the stress conditions on the ceramic insert by considering contact and interference effects under both mechanical and thermal loads.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.16 no.4
• /
• pp.29-35
• /
• 2007

The main objective of this paper is to propose the optimal design method of forging process using genetic algorithm. Design optimization of forging process was doing about one stage and multi stage. The objective function is considered the filling of die. The chosen design variables are die geometry in multi stage and initial billet shape in one stage. We performed FE analysis to simulated forging process. The optimized preform and initial billet shape was obtained by genetic algorithm and FE analysis. To show the efficiency of GA method in forging problem are solved and compared with published results.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.11 no.6
• /
• pp.105-110
• /
• 2002

ZnTe epilayers have been successfully grown on (100) CaAs substrate by hot wall epitaxy (HWE) with Zn reservoir. Optimum growth condition has been determined by a four-crystal rocking curve (FCRC). It was found that Zn partial pressure from h reservoir has a strong influence on the quality of grown films. Under the determined optimum growth condition, ZnTe epitaxial films with thickness of 0.72~24.8${\mu}m$ were grown for studying the effect of the thickness on crystalline quality. The FCRC results indicated that the quality of ZnTe films becomes higher rapidly with increase of thickness up to 6${\mu}{\textrm}{m}$. The best value of the FWHM of the few crystal rocking curve, 66 arcsec, was obtained on the film with $12{\mu}m$ in thickness. Until now, this result shows the best quality of ZnTe/GaAs films in reported.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.11 no.6
• /
• pp.111-117
• /
• 2002

The high quality and a nearly stoichometric growth of $Cd_{1-y} Zn_y$/Te(y＝0.04) epilayers have been successfully grown on GaAs substrate by hot wall epitaxy (HWE) by optimizing the growth condition including the preheating treatment and Cd reservoir temperature. The relationship between quality and thickness was examined and best value of FWHM from X-ray rocking curve of 121 arcsec are obtained. Also, emission peaks related to the recombination of free excitons such as the ground state and the first excited state were observed in the PL spectrum at 4.2K. The ($A^0$, X) emission related to Cd vacancy and deep level emission was not measured. These results indicated that the grown CZT/GaAs epilayer was high qualify and purity.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 1998.04a
• /
• pp.12-12
• /
• 1998

국제 공동 가스터빈 엔진 개발 프로그램인 ASE120 사업은 항공기용 터보팬 엔진을 Low Emission 산업용 엔진으로의 개조사업으로서 현대우주항공, AlliedSignal, AIDC(대만), SPHT(대만) 등이 참여하고 있다. 본 사업을 통한 주요 개조 부품인 연소기의 경우, 산업용 엔진의 특성상 장시간의 수명이 보장되어야 하고 Dry Low NOx 연소실 형태로 배기가스 규제의 요구조건을 만족하여야 하므로 일반 항공기용 연소기와 구조적으로 첨예한 변화가 발생하여 구조 및 수명해석을 통한 구조적 안전성과 요구 수명에 대한 평가가 절대적으로 요구되었다. 당 엔진의 요구수명은 30년으로 cold section part의 경우 180,000시간, hot section의 경우 90,000시간의 수명을 목표로 개발되었으며, 연소기 최적설계를 위해 구조/수명 해석결과가 개발과정 중 설계에 지속적으로 반영되어 추진되었다. 연소기의 구조해석은 해석용 tool로서 ANSYS53을 활용하였고 수명예측은 AlliedSignal사의 Material Database 및 Inconel사 재질 data를 근간으로 수행하였다.

• Journal of the Korean Institute of Rural Architecture
• /
• v.1 no.1
• /
• pp.23-40
• /
• 1999

In this paper, I made a comparative study of design process of Kuryong-Ri, Keumsong-Myun, Chechon City, Chungbuk Province which is one of rural communities in Korea and framing process of master plan for Hama street at Sirahama town on Wakayama Ken, famous hot-bath resort in Japan. Both processes have common points to adopt identical design philosophy and method called 'Pattern Language'. The aim of this study to provide valuable reference for applying at community design by the people in the future, though analyzing common and unlike points, laying stress on the process of user-participation rather than its final results. It is possible confirm Pattern Language is effective tool for carrying out community design by the people.

• Proceedings of the KAIS Fall Conference
• /
• 2004.06a
• /
• pp.89-91
• /
• 2004

This study examines a regeneration process using hot air heated by solar radiation to recover absorption potential by evaporating moisture in liquid desiccant. More specifically, this study is aimed at finding the optimum operating condition of the regenerator by utilizing a well-established statistical tool, so-called design of experiment, and optimization techniques. It is demonstrated that an optimization model to find the optimum operating condition can be obtained using the functional relationship between regeneration rate and affecting factors which is approximated on the basis experimental results.