• 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.

• Journal of the Korea Society for Simulation
• /
• v.33 no.2
• /
• pp.27-35
• /
• 2024

We aimed to shed light on the issue of perpetuating societal disparities by analyzing the impact of inherent biases present in datasets used for training artificial intelligence models on the predictions generated by Artificial Intelligence(AI). Therefore, to examine the influence of data bias on AI models, we constructed an original dataset containing biases related to gender wage gaps and subsequently created a de-identified dataset. Additionally, by utilizing the decision tree algorithm, we compared the outputs of AI models trained on both the original and de-identified datasets, aiming to analyze how data de-identification affects the biases in the results produced by artificial intelligence models. Through this, our goal was to highlight the significant role of data de-identification not only in safeguarding individual privacy but also in addressing biases within the data.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.22 no.1
• /
• pp.23-32
• /
• 1998

A shape estimation is needed to control actively a smart structure. A method is, hence, proposed to predict the deformed shape of the structure subjected to unknown external load using the signal from sensors attached to the structure. The shape estimation is based on the relationship between the deformation of the structure and the signal from the sensors. The matrix containing the relationship between the deformation and signal is obtained using fictitious force or eigenvector of global stiffness matrix. Then the deformed shape can be predicted using the linear matrix and signal from sensors attached to the structure. To verify this method, experiment and FEM were performed and it was shown that the shape estimation method based on the fictitious force predicts deflections well and more accurately than that based on eigenvector.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.1
• /
• pp.15-21
• /
• 2015

In support vector machine, the values of parameters included in kernels affect strongly generalization ability. It is often difficult to determine appropriate values of those parameters in advance. It has been observed through our studies that the burden for deciding the values of those parameters in support vector regression can be reduced by utilizing ensemble learning. However, the straightforward application of the method to large scale problems is too time consuming. In this paper, we propose a method in which the original data set is decomposed into a certain number of sub data set in order to reduce the burden for parameter tuning in support vector regression with large scale data sets and imbalanced data set, particularly.

• Journal of Intelligence and Information Systems
• /
• v.7 no.1
• /
• pp.63-79
• /
• 2001

Data Envelopment Analysis(DEA), a non-parametric productivity analysis tool, has become an accepted approach for assessing efficiency in a wide range of fields. Despite of its extensive applications and merits, some features of DEA remain bothersome. DEA offers no guideline about to which direction relatively inefficient DMUs improve since a reference set of an inefficient DMU, several efficient DMUs, hardly provides a stepwise path for improving the efficiency of the inefficient DMU. In this paper, we aim to show that DEA can be used to evaluate the efficiency of life insurance companies while overcoming its limitation with the aids of machine learning methods.

• Journal of Intelligence and Information Systems
• /
• v.10 no.2
• /
• pp.165-187
• /
• 2004

Constraint Programming in AS and Optimization in OR started and have grown in different backgrounds to solve common decision-making problems in real world. This paper tries to integrate results from those different fields by suggesting a hybrid solver as an integration framework. Starting with an integrating modeling language, a way to implement a hybrid solver will be discussed using Warren's abstract machine and an finite domain constraint programming solver structures. This paper will also propose some issues rising when implementing the hybrid solver and provide their solutions.

• Proceedings of the KSME Conference
• /
• 2003.04a
• /
• pp.734-739
• /
• 2003

Many types of smart materials and control laws are available to actively adjust the structure from various external disturbances. Usually, a certain type of control laws to activate a specific smart material is well established, but the effectiveness of the control scheme is limited by the choice of the smart materials and the responses of the structure. ER fluid is adequate to provide relatively large control force, on the other hand, the PZT patches are suitable to provide small but arbitrary control forces at any point along the structure. It was found that active vibration control mechanism using ER fluid failed to suppress the excitation off the resonant frequency with changed structural characteristics along the frequency response function of the closed loop of the control system. To compensate this additional peak of the closed loop system, PPF control using PZT as an actuator is added to construct a hybrid controller.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2002.11a
• /
• pp.321-327
• /
• 2002

본 논문은 기존의 귀납적 결정 트리 방식에서의 문제점 개선을 통한 사용자 관심 프로파일 구축을 목적으로 한다. 특히 사용자 관심 프로파일의 정확도 향상을 위한 속성 선택에 대한 연구에 초점을 맞추고 있다. 사용자의 관심, 비관심 문서를 대상으로 사용자 관심 키워드를 생성하고 이를 바탕으로 초기 문서들을 재표현한다. 재표현된 문서를 입력 집합으로 하여 기계학습을 진행한다. 본 논문의 의사 결정 트리 생성 알고리즘은 입력 집합을 클래스별로 가장 잘 나누는 속성을 선택하여 노드를 구성하는 면에서는 기존의 알고리즘과 같다. 그러나 기존의 의사 결정 트리 알고리즘에서는 hill-climbing.방식을 사용함으로써 사용자의 관심을 나타내는 중요한 단어가 사용자 관심 프로파일에서 숨겨질 경우가 발생한다. 이를 최소화하기 위해 특징 추출을 통해 선택된 속성을 그대로 학습의 입력 데이터로 사용하는 것이 아니라 입력데이터를 가장 잘 나누는 속성과 그 다음 속성을 대상으로 disjunctive 연산을 통해 새로운 속성을 생성하여 이것을 속성 집합에 포함시키고 이를 학습의 입력 데이터로 이용한다. 이와 같이 disjunctive operator를 이용하여 새로운 속성을 의사 결정 트리 형성 시 이용하면 사용자의 중요한 관심을 포함하는 의미 있는(semantic) 사용자 관심 프로파일 구축이 가능해지고, 사용자 관심 프로파일을 기반으로 사용자가 관심 있는 문서를 제공할 수 있는 개인화 서비스를 제공한다.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.17 no.7 s.94
• /
• pp.1649-1657
• /
• 1993

This paper reports on a proof-of-concept experimental investigation focused on evaluating the vibration characteristics and control of smart hollow cantilever beams filled with an electro-rheological(ER) fluid. The beams are considered to be of uniform viscoelastic materials and modelled as a viscously-damped harmonic oscillator. Electric field-dependent natural frequencies, loss factors and complex moduli are evaluated and compared among three different beams : two types of different volume fraction of ER fluid and one type of different particle concentration of ER fluid by weight. Modal characteristics of the beams are observed in both the absence and the presence of electric potentials. It is also shown that by constructing active control algorithm the removal of structural resonances and the suppression of tip deflection are obtained. This result provides the feasiblility of ER fluids as an active vibration control element.

• Proceedings of the KSME Conference
• /
• 2008.11a
• /
• pp.565-570
• /
• 2008

Recently, the intelligent service robot is applied for the purpose of guiding the building or providing information to the visitors of the public institution. The intelligent robot which is on development has a sensor to recognize its location at the bottom of it. Four wheels which are arranged in the form of a lozenge support the weight of the components and structures of the robot. The operating environment of this robot is restricted at the uneven place because the driving part and internal structure is designed in one united body. The impact from the ground is transferred to the internal equipments and structures of the robot. This continuous impact can cause the unusual state of the precise components and weaken the connection between each structural part. In this paper, a suspension system which can be applied to the intelligent robot is designed. The dynamic model of the robot is created, and the driving characteristics of the actual robot and the robot with suspension are compared. The road condition which the robot can operate is expanded by the application of the suspension system. Additionally, the suspension system is optimized to reduce the impact to the robot components.