• Proceedings of the Korean Statistical Society Conference
• /
• 2005.11a
• /
• pp.193-197
• /
• 2005

Classification tree is one of the most suitable base learners for ensemble. For past decade, it was found that bagging gives the most accurate prediction when used with unpruned tree and boosting with stump. Researchers have tried to understand the relationship between the size of trees and the accuracy of ensemble. With experiment, it is found that large trees make boosting overfit the dataset and stumps help avoid it. It means that the accuracy of each classifier needs to be sacrificed for better weighting at each iteration. Hence, split effect in boosting can be explained with the trade-off between the accuracy of each classifier and better weighting on the misclassified points. In bagging, combining larger trees give more accurate prediction because bagging does not have such trade-off, thus it is advisable to make each classifier as accurate as possible.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.9 no.6
• /
• pp.47-56
• /
• 2001

Nitrogen oxides(NOx) and smoke emissions of diesel engine are regarded as a source of air pollution, and there is a global trend to enforce more stringent regulations on these exhaust gas emissions. However, the trade-off relation of NOx and smoke is a main obstacle to reduce both of them simultaneously. In this paper, experiments were conducted with an oxygenated fuel(diethyl ether) as an effective way to improve the trade-off relation of NOx and smoke. Exhaust emissions of diesel fuels with DEE were influenced by the additive content of DEE and the injection timing. Especially, DEE effected more at the high engine speed and load than at the low engine speed and load. Diesel fuel blended with DEE 10% was a desirable blend for the simultaneous reduction of NOx and smoke.

• International Journal of Aeronautical and Space Sciences
• /
• v.2 no.2
• /
• pp.73-81
• /
• 2001

This paper describes the procedure to develop a robust estimator design method for a target tracker that accounts for both structured real parameter uncertainties and unknown inputs. Two robust design approaches are combined: the Mini-p-Norm. design method to consider real parameter uncertainties and the $H_{\infty}$ design technique for unknown disturbances and unknown inputs. Constant estimator gains are computed that guarantee the robust performance of the estimator in the presence of parameter variations in the target model and unknown inputs to the target. The new estimator has two design parameters. One design parameter allows the trade off between small estimator error variance and low sensitivity to unknown parameter variations. Another design parameter allows the trade off between the robustness to real parameter variations and the robustness to unknown inputs. This robust estimator design method was applied to the longitudinal motion tracking problem of a T-38 aircraft.

• ETRI Journal
• /
• v.37 no.5
• /
• pp.951-960
• /
• 2015

For a conventional power metal-oxide-semiconductor field-effect transistor (MOSFET), there is a trade-off between specific on-state resistance and breakdown voltage. To overcome this trade-off, a super-junction trench MOSFET (TMOSFET) structure is suggested; within this structure, the ability to sense the temperature distribution of the TMOSFET is very important since heat is generated in the junction area, thus affecting its reliability. Generally, there are two types of temperature-sensing structures-diode and resistive. In this paper, a diode-type temperature-sensing structure for a TMOSFET is designed for a brushless direct current motor with on-resistance of $96m{\Omega}{\cdot}mm^2$. The temperature distribution for an ultra-low on-resistance power MOSFET has been analyzed for various bonding schemes. The multi-bonding and stripe bonding cases show a maximum temperature that is lower than that for the single-bonding case. It is shown that the metal resistance at the source area is non-negligible and should therefore be considered depending on the application for current driving capability.

• Proceedings of the KSME Conference
• /
• 2004.04a
• /
• pp.525-530
• /
• 2004

In this paper, the optimum design of tire sidewall contour consisted of double plies for improving automobile maneuverability and tire durability is considered and a GUI program is developed for the purpose of the practical design. Each improvement of maneuverability and durability depends on the cord tension and strain energy distribution of tire sidewall. Satisfing trade-off method, which requires the judgment of aspiration levels, is used for the multi-objective optimization problem. Also, this paper presents the application to the practical sidewall contour design with the GUI program developed using visual Fortran.

• Transactions of the Korean Society of Machine Tool Engineers
• /
• v.14 no.6
• /
• pp.1-7
• /
• 2005

This paper deals with the position control for a feed drive system in CNC milling machine. The study shows that reduction in tacking error does not necessarily increase contouring accuracy, and proposes that a proper control scheme is able to exhibit the optimal tracking and contouring accuracy. The proposed scheme is to fix the parameter of the contouring control, and to find the optimal value of the parameter of the tracking control, which is automatically tuned. The effectiveness of the proposed scheme is confirmed through simulations.

• Aerospace Engineering and Technology
• /
• v.1 no.1
• /
• pp.55-64
• /
• 2002

A conceptual design of propulsion system for a geosynchronous communication satellite with 12 years design life is presented in this paper. Propellant mass budget for the design life is calculated using total velocity increment (ΔV) flowed-down from mission requirement analysis. Sizes of the fuel and oxidizer tank are derived based on the calculated propellant mass budget, and mass of the pressurant as well as the size and pressure of pressurant tank are calculated too. Thruster positioning, number of rocket engines, and position of tank are determined through Trade-Off Study with Structure & Mechanical Subsystem. Propulsion system configuration and its schematics are presented finally.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.11 no.4
• /
• pp.44-51
• /
• 2003

By blending of various low boiling point oxygenated agents to lower grade fuels, significant improvements were simultaneously obtained in smoke, CO, PM, SOF and BSEC. Especially, our trends were remarkably obtained by retarding injection timing, by decreasing boiling point and increasing blending contents of additives in case of oxygenated agents rather than non-oxygenated agents. Also, it was revealed that when 20vo1.% DMM added to high viscosity fuels and injection timing was retarded, NOx-smoke trade off relationship was much better than that of ordinary diesel fuel. Thus, lower grade fuels with high viscosity could be expected to be used efficiently and cleanly in diesel operation by blending low boiling point oxygenates.

• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.62 no.4
• /
• pp.216-222
• /
• 2013

Handover is a technical methodology to support mobility of wireless communication users, which mainly affects the capacity of wireless communication systems and the quality of service (QoS) of link level signal. However, in OFDMA-based cellular systems, there are few published technical reports investigating handovers with respect to diversity gains and resource consumption depending on what handover technique is adopted. In this paper, we propose handover schemes exploiting transmit diversity and macro-diversity in order to increase capacity of OFDMA-based wireless communication systems, and analyze their performance. In cellular environments, depending on what handover scheme is selected, average signal-to-interference-and-noise ratio is calculated first for a handover user in order to evaluate the link level QoS. Through this technical evaluation for handover schemes, we conclude what handover scheme is suitable for OFDMA-based cellular systems.

• Journal of Korea Multimedia Society
• /
• v.3 no.2
• /
• pp.184-191
• /
• 2000

In thid paper, we have analized the characteristics of $\Delta$-shaped filter based on preceeding study. The algorithm of $\Delta$-shaped filter for full screen display has been discussed including their edge-preserving characteristics and computaion complexity compared with the conventional algorithms. PSNR has been employed to compare the proposed method to the conventional algorithms. The proposed algorithm has been implemented using Synopsys VHDL tool. In addition, the features and trade off of the $\Delta$-shaped filter have been discussed.