• Korean Journal of Mathematics
• /
• 제26권2호
• /
• pp.337-348
• /
• 2018

MNIST dataset is a database containing images of handwritten digits, with each image labeled by an integer from 0 to 9. It is used to benchmark the performance of machine learning algorithms. Neural networks for MNIST are regarded as the starting point of the studying machine learning algorithms. However it is not easy to start the actual programming. In this expository article, we will give a step-by-step instruction to build neural networks for MNIST dataset using MATLAB.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2007년도 춘계학술대회논문집
• /
• pp.609-612
• /
• 2007

The HDD (hard disk drive) using Load/Unload (L/UL) technology includes the benefits which are increased areal density, reduced power consumption and improved shock resistance than those of contact-start-stop (CSS). Dynamic L/UL has been widely used in portable hard disk drive and will become the key technology for developing the small form factor hard disk drive. The main design objectives of the L/UL mechanisms are no slider-disk contact or no media damage even with contact during L/UL, and a smooth and short unloading process. In this paper, we focus on lift-off force, pitch static attitude (PSA), roll static attitude (RSA) and dimple point. The "lift-off" force, defined as the minimum air bearing force, is another very important indicator of unloading performance. A large amplitude of lift-off force increases the ramp force, the unloading time, the slider oscillation and contact-possibility. PSA and RSA are also very important parameters in L/UL system and stability of slider is mainly determined by PSA and RSA. Dimple point by PSA and RSA is also important indicator. Therefore we find the optimal dimple point of SFF HDD suspension for improving the unloading performance.

• 전력전자학회논문지
• /
• 제17권3호
• /
• pp.246-251
• /
• 2012

This paper proposes a new position compensation method for misaligned Hall-effect sensors of BLDCM(Brushless DC Motor). If the Hall-effect sensors are installed at wrong position, the exact rotor position cannot be obtained. Therefore, when the BLDCM is controlled with this wrong position, the torque ripple can be increased and the average torque also decreases. The back-EMF of BLDCM can be obtained by using the voltage equation and by multiplying the back-EMF constant and rotor speed. At a constant speed, the estimated back-EMF by using the multiplication of the back-EMF constant and rotor speed is constant, but the estimated back-EMF from the voltage equation decreases at the commutation point because the line-to-line back-EMF of two conducting phases is start to decrease at this point. Therefore, by using the difference between these two estimated back-EMFs, the commutation point of the phase current can be determined and position compensation can be carried out. The proposed position correction method doesn't require additional hardware circuit and can be easily implemented. The validity of the proposed position compensation method is verified through several experiments.

• 한국해양공학회지
• /
• 제13권3B호
• /
• pp.38-46
• /
• 1999

Increasing stringent emissions legislation and requirement of more effective energy used for diesel engine demand the fine control of the fuel injection system. Recently, the electromagnetic fuel injection control system for diesel engine is tried to realize the optimum diesel combustion by the feel back sensing as optical signal of combustion flame. The photo detectors were made for the feed back signal of electromagnetic fuel injection control for small DI diesel engine. Their abilities to detect defining combustion events were examined. By evaluating test results, it was shown that the wider acceptable optical range design of optical probe window face, and selection of installation position and installation method of detector were important point for improving sensing ability. The detector was shown to detect start and end of diffused combustion and maximum point of flame intensity impossible for pressure sensor, and also shown that the maximum point of flame intensity was 75% of accumulated heat release point within the experimental conditions.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2001년도 춘계학술대회 논문집(한국공작기계학회)
• /
• pp.399-404
• /
• 2001

In CNC machining, a 3D(3-dimension) linear segment and a 2D(2-dimension) circular arc are general forms given by CAD/CAM system. Generally, the 2D circular arc machining is processed using dividing into some linear segments. A 3D circular arc also don't exist in the standard form of NC data. This paper present a algorithm and method for real-time machining of a circular arc(not only the 2D one, but also the 3D one). The 3D circular arc machining is based on the 2D circular arc machining. It only needs making a new coordinate system, converting given 3D points(a start point, a end point, and a center point of a 3D circular arc) into points of the new coordinate system, and processing a inverse transformation about a interpolated point. The proposed algorithm was implemented and simulated on PC system. It was confirmed to give a gcod result.

• 한국지반공학회논문집
• /
• 제36권3호
• /
• pp.25-36
• /
• 2020

동토의 세립분 함유량에 따른 파괴특성을 파악하기 위해 -10℃의 온도에서 다양한 세립분 함유량과 초기 노치(notch)의 위치를 조정한 직사각형 공시체를 제작하여 Three-point bending 시험을 수행하였다. 시험결과를 바탕으로 동토의 mode I 파괴인성(fracture toughness)을 산정하였으며, 하중-변형 곡선의 최대점까지의 fracture energy를 산정하여 동토의 mixed-mode(mode I + II) 파괴특성을 파악하였다. 시험결과, 최대하중 및 mode I 파괴인성은 세립분 함유량 10%까지 증가하다가 15%에서 다시 감소하는 경향을 나타내었다. 또한, 노치의 위치가 공시체 중심에서 멀어질수록 mode II 하중의 증가로 인해 균열이 진행하는데 필요한 fracture enenrgy가 증가하는 것으로 나타났으며, 세립분 함유량이 증가할수록 mode II 하중의 증가비율 또한 증가하는 것으로 나타났다.

• 아태비즈니스연구
• /
• 제12권1호
• /
• pp.151-163
• /
• 2021

Purpose - The break-even point refers to the point where total profit and total cost coincide, and from this point on, the entrepreneur's decision-making takes a different route. Strategic decisions can be made for more efficient operation and eventually for more likelihood for growth and sustainability if a startup figures out when it recoups the investment and switches to a net profit. Design/methodology/approach - 748 creativity and skill-based sole proprietors in manufacturing industry were examined to demonstrate the effect of the entrepreneur's entrepreneurial experience and education level, the business launch preparation time, or the self-financing on the achievement of break-even point. Findings - While the business launch preparation time lowered the likelihood of reaching a break-even point, self-financing increased the likelihood. As a result of further analysis by subdividing into subgroups according to skill level, only the business launch preparation time was statistically significant in the highly skilled industries. In the low skilled industries, in addition to the business launch preparation time, the CEO's education level and the self-financing were statistically significant. Research implications or Originality - The longer the business launch preparation time, the higher the start-up cost, which increases the burden of initial cost recovery, and the agile response to market changes is thereby delayed, resulting in the business idea losing its appeal. Self-financing not only provides stability and strong motivation for the business operation but also promotes careful spending which contributes to the achievement of break-even point. In particular, it is found that practical experience is more useful than theoretical knowledge in low skilled industries. Due to the limitation of secondary data based on the recollection, the time required to reach a break-even point, percentage of financing sources, etc. may include cognitive errors. In addition, variables are not included that explain the characteristics of creativity and skill-based sole proprietorship, so it is necessary to exercise caution with the actual application.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2007년도 춘계학술대회B
• /
• pp.2769-2774
• /
• 2007

In vacuum ejector-diffuser systems where a finite volume secondary chamber is used, the secondary jet exhibits transient characteristics during start-up. A steady state is achieved after some time in which mass entrainment prevails indefinitely inside the ejector, though there is no flow from the secondary chamber. An attempt is made in this work to study the infinite entrainment of secondary jet into the primary jet from a finite secondary chamber, with the help of a computational fluid dynamics method. The present study is also intended to identify the operating range of vacuum ejector-diffuser systems where the steady flow assumption can be applied without uncertainty. The results obtained show that the only condition in which an infinite mass entrainment is possible is the generation of a re-circulation zone near the primary nozzle exit. The flow in the secondary chamber attains a state of dynamic equilibrium at this point. Steady flow assumption is valid only after this point.

• 한국생물정보학회:학술대회논문집
• /
• 한국생물정보시스템생물학회 2005년도 BIOINFO 2005
• /
• pp.278-283
• /
• 2005

Cell cycle is regulated cooperatively by several genes. The dynamic regulatory mechanism of protein interaction network of cell cycle will be presented taking the budding yeast as a sample system. Based on the mathematical model developed by Chen et at. (MBC, 11,369), at first, the dynamic role of the feedback loops is investigated. Secondly, using a bifurcation diagram, dynamic analysis of the cell cycle regulation is illustrated. The bifurcation diagram is a kind of ‘dynamic road map’ with stable and unstable solutions. On the map, a stable solution denotes a ‘road’ attracting the state and an unstable solution ‘a repelling road’ The ‘START’ transition, the initiation of the cell cycle, occurs at the point where the dynamic road changes from a fixed point to an oscillatory solution. The 'FINISH' transition, the completion of a cell cycle, is returning back to the initial state. The bifurcation analysis for the mutants could be used uncovering the role of proteins in the cell cycle regulation network.

• 한국정밀공학회지
• /
• 제16권12호
• /
• pp.182-189
• /
• 1999

In this paper, we present an ultrasonic sensor based path planning method using fuzzy logic for obstacle avoidance of an intelligent vehicle in unknown environments. Generally, Robot navigation in unknown terrains is a very complex task difficult to control because of the great amount of imprecise and ambiguous sensor information that has to be considered. In this case, fuzzy logic can satisfactorily deal with such information in quite efficient manner. In this study, we propose two fuzzy logic controller which is composed of steering controller and velocity controller respectively. Our object is to develop a fuzzy controller that can enable a mobile robot to navigate from a start point to a goal point without collisions, in the least possible travel time. The ability and effectiveness for the proposed algorithm will be demonstrated by simulation and expeiment.