• 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.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.41 no.11
• /
• pp.775-784
• /
• 2017

The purpose of this study was to analyse the basic data of the engine tuning inspection by confirming the working possibility of effective engine tuning and identifying the characteristics of tuned engine that are no problem with the safety operation and environment in a driving gasoline car. The effects of tuned engine on the characteristics of air/fuel ratio and performance at a wide range of engine speeds were experimentally investigated by the actual driving car with a four-cycle, four-cylinder DOHC, turbo-intercooler, water-cooled gasoline engine operating under four types of non-tuning, and tuning 2-1, 2-2 and 2-3. The tuned parts of engine in a driving gasoline car include the intake manifold, intake pipe, air filter, exhaust manifold, exhaust pipe and silencer. In this experiment, the air-fuel ratio and torque of both non-tuned and tuned engines that one person took on board in the car with a five-speed automatic transmission were measured by the chassis dynamometer(Dynojet 224xLC). It was found that the maximum torque of tuned engine in a driving gasoline car was increased by 103.68% on average, while the maximum output was increased by 119.68% on average in comparison to the non-tuned engine.

• The KIPS Transactions:PartB
• /
• v.14B no.2
• /
• pp.127-134
• /
• 2007

There are many researches about the intelligent stock trading systems with the help of the advance of the artificial intelligence such as machine learning techniques, Though the establishment of the reasonable trading policy plays an important role in the performance of the trading systems most researches focused on the improvement of the predictability. Also some previous works, which treated the trading policy, treated the simplified versions dependent on the predictors in less systematic ways. In this paper, we propose the integrated multiple simulation' as a method of optimizing trading performance of stock trading systems. The propose method is adopted in the NXShell a development environment for neural network based stock trading systems. Under the proposed integrated multiple simulation', we simulate the multiple tradings for all combinations of the neural network's outputs and the trading policy parameters, evaluate the learning performance according to the various metrics and establish the optimal policy for a given prediction module based on the resulting performance. In the experiment, we present the trading policy comparison results using the stock value data from the KOSPI and KOSDAQ.

• The KIPS Transactions:PartB
• /
• v.11B no.3
• /
• pp.387-394
• /
• 2004

In this paper, we propose a modified unsupervised linear alignment algorithm for building an aligned corpus. The original algorithm inserts null characters into both of two aligned strings (source string and target string), because the two strings are different from each other in length. This can cause some difficulties like the search space explosion for applications using the aligned corpus with null characters and no possibility of applying to several machine learning algorithms. To alleviate these difficulties, we modify the algorithm not to contain null characters in the aligned source strings. We have shown the usability of our approach by applying it to different areas such as Korean-English back-trans literation, English grapheme-phoneme conversion, and Korean morphological analysis.

• Review of KIISC
• /
• v.33 no.4
• /
• pp.23-29
• /
• 2023

대부분의 임베디드 시스템은 기계장치와 전자기기 장치가 함께 작동되는 물리 장치로써, 이기종 네트워크, 복잡한 보안체계 등을 고려하여 가상화 기반 사이버훈련 환경이 구성되어야 한다. 또한, 차량을 대상으로 물리적인 실험환경에서 모의침투 등 사이버훈련을 수행한다는 것은 교통사고를 비롯한 안전사고 발생에 있어 위험이 존재한다. 본 논문에서는 가상 개발환경에서의 공격 기반 차량용 사이버훈련 프레임워크를 제안하고자 한다. 먼저, 공격 기반 차량용 사이버훈련 프레임워크의 작동은 자동 활성화되는 가상의 CAN 네트워크 인터페이스로 시작된다. 가상의 CAN 네트워크 인터페이스는 가상 머신에서 간단한 부트스트랩 명령어 실행을 통해 파이썬 패키지와 Ubuntu 서비스 목록 설치 명령이 자동으로 실행되면서 설치된다. 이후 내부 네트워크 시뮬레이터와 공격모듈과 관련된 UI가 자동으로 Ubuntu Systemd에 의해 백그라운드에서 실행되어 시작과 동시에 준비 상태를 유지하게 된다. 사이버훈련 UI 내 공격 모듈은 사용자에 의한 공격 선택 및 파라미터 셋팅 이후 차량의 이상 상태를 사이버훈련 UI에 다시 출력되게 된다. 본 논문에서 제안하는 가상 개발환경 기반의 차량용 사이버훈련 프레임워크는 자율주행 차량 사고의 위험이나 다른 특수한 제약 없이 사용자의 학습 경험을 확장시킬 수 있다. 또한, 기존의 가상화 기반 사이버훈련 교육 콘텐츠와는 달리 일반 사용자들이 접근하기 쉬운 형태로 확장 개발이 가능하다.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.34 no.5
• /
• pp.619-628
• /
• 2010

Shape Memory Alloy (SMA) has recently received attention in developing implantable surgical equipments and it is expected to lead the future medical device market by adequately imitating surgeons' flexible and delicate hand movement. However, SMA actuators have not been used widely because of their nonlinear behavior called hysteresis, which makes their control difficult. Hence, we propose a parameter, $t_1$, which is necessary for temperature control, by analyzing the open-loop step response between current and temperature and by comparing it with the values of linear differential equations. $t_1$ is a pole of the transfer function in the invariant linear model in which the input and output are current and temperature, respectively; hence, $t_1$ is found to be related to the state variable used for temperature control. When considering the parameter under heat treatment conditions, $T_{max}$ was found to assume the lowest value, and $t_1$ was irrelevant to the heat treatment.

• Korean Journal of Optics and Photonics
• /
• v.5 no.3
• /
• pp.423-430
• /
• 1994

We designed and fabricated the single mode fiber pigtailed DFB-LD module for 2.5 Gbps optical communication system. In the design of the DFB-LD module, we made the module divided into two parts of inner sub-module and outer 14-pin butterfly package and cylindrical shaped sub-module contain quasi confocal 2 lens system including optical isolator and electrical connection between these parts via hybrid substrate of bias T circuit. Laser welding was used to assemble the sub-module which requires accurate fixing between optical elements. The fabricated DFB-LD module showed optical coupling efficiency of 20% and - 3 dB small signal response of more than 2.6 GHz. We confirmed mechanical reliability of the module by temperature cycle test where the tested module exhibit optical power fluctuation of less than 10%. Finally we evaluated the performance of the fabricated DFB-LD module as light source of 2.5 Gbps optical communication system, sensitivity of - 30.2 dBm was obtained through 47 km optical fiber transmission under the criterion of $1\times10^{-10}$ BER and transmission penalties were 1.5 dB caused by extinction ratio and 1.0 dB caused by chromatic dispersion of normal single mode fiber. fiber.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.35 no.6
• /
• pp.577-583
• /
• 2011

To meet the requirements of the Tier 4 interim regulations for off-road vehicles, emissions of particulate matter (PM) and nitrogen oxides (NOx) must be reduced by 95% and 30%, respectively, compared to current regulations. In this research, both the DPF and HPL EGR systems were investigated, with the aim of decreasing the PM and NOx emissions of a 56-kW off-road vehicle. The results of the experiments show that the DOC-DPF system is very useful for reducing PM emissions. It is also found that the back pressure is acceptable, and the rate of power loss is less than 5%. By applying the HPL EGR system to the diesel engine, the NOx emissions under low- and middle-load conditions are reduced effectively because of the high differential pressure between the turbocharger inlet and the intake manifold. The NOx emissions can be decreased by increasing the EGR rate, but total hydrocarbon (THC) emission increases because of the increased fuel consumption needed to compensate for the power loss caused by EGR and DPF.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.41 no.10
• /
• pp.788-795
• /
• 2013

A magnesium bipolar plate whose surface was protected by thinly deposited silver layer was investigated as an alternative to existing graphite bipolar plate of PEM fuel cells. Thin silver layer of $3{\mu}m$ was deposited on a magnesium alloy substrate by physical vapor deposition (PVD) method in an environment of $180^{\circ}C$. A number of tests were conducted on the fabricated magnesium based bipolar plates to determine their suitability for use in PEM fuel cell stacks. The test on corrosion resistance in the same pH condition as in a PEM operation demonstrated the layer protected the magnesium alloy substrate, while unprotected substrate suffered from severe corrosion. The contact resistance of the fabricated bipolar plate was less than $20m{\Omega}-cm^2$ which was superior to the conventional bipolar plates. A single cell was constructed using the fabricated bipolar plates and power output was measured. Due to the enhanced conductivity caused by low contact resistance, slight increase was observed in current density and output voltage. With low density of the magnesium substrate and ease on machining, the weight reduction of the stack of 30~40 % is possible to produce the same power output.

• Journal of Broadcast Engineering
• /
• v.19 no.5
• /
• pp.578-589
• /
• 2014

Window image is displayed through a monitor screen when we execute the application programs on the computer. This includes webpage, video player and a number of applications. The webpage delivers a variety of information by various types in comparison with other application. Unlike a natural image captured from a camera, the window image like a webpage includes diverse components such as text, logo, icon, subimage and so on. Each component delivers various types of information to users. However, the components with different characteristic need to be divided locally, because text and image are served by various type. In this paper, we divide window images into many sub blocks, and classify each divided region into background, text and subimage. The detected subimages can be applied into 2D-to-3D conversion, image retrieval, image browsing and so forth. There are many subimage classification methods. In this paper, we utilize AdaBoost for verifying that the machine learning-based algorithm can be efficient for subimage detection. In the experiment, we showed that the subimage detection ratio is 93.4 % and false alarm is 13 %.