• 한국컴퓨터정보학회논문지
• /
• 제7권3호
• /
• pp.42-48
• /
• 2002

배치(Placement)는 VLSI 회로의 physical design에서 중요한 단계로서 회로의 성능을 최대로 하기 위하여 회로 모듈의 집합을 배치시키는 문제이며, 배치 문제에서 최적의 해를 얻기 위해 클러스터 성장(cluster growth), 시뮬레이티드 어닐링(simulated annealing; SA), ILP(integer linear programming)등의 방식이 이용된다. 본 논문에서는 배치 문제에 대하여 유전자 알고리즘(genetic algorithm; GA)을 이용한 해 공간 탐색(solution space search) 방식을 제안하였으며, 제안한 방식을 시뮬레이티드 어닐링 방식과 비교, 분석하였다.

• Journal of the Korean Statistical Society
• /
• 제1권1호
• /
• pp.18-24
• /
• 1973

Study on convexity has been improved in many statistical fields, such as linear programming, stochastic inverntory problems and decision theory. In proof of main theorem in Section 3, M. Loeve already proved this theorem with the $r$-th absolute moments on page 160 in [1]. Main consideration is given to prove this theorem using convex theorems with the generalized $t$-th mean when some convex properties hold on a real linear vector space $R_N$, which satisfies all properties of finite dimensional Hilbert space. Throughout this paper $\b{x}_j, \b{y}_j$ where $j = 1,2,......,k,.....,N$, denotes the vectors on $R_N$, and $C_N$ also denotes a subspace of $R_N$.

• 한국항공우주학회지
• /
• 제47권11호
• /
• pp.787-794
• /
• 2019

본 논문에서는 다수 이종 근접 방어 시스템(Closed-In Weapon System, CIWS)의 최적 무장 할당 문제를 제시하고, 이를 혼합정수선형계획법(Mixed Integer Linear Programming, MILP)으로 변형해 해결하는 기법을 제안한다. 일반적인 무장 할당 문제의 경우 다양한 경우의 수를 고려해야하기 때문에 계산 시간이 기하급수적으로 증가하는 경우가 잦다. 하지만 주어진 문제를 MILP와 같은 혼합정수 최적화 문제로 변형하면 준실시간 내에 전역 최적해를 찾을 수 있다. 본 논문에서는 다수 위협이 각각 다른 시점에 다른 방향에서 방어 자산을 공격하는 상황을 고려한다. 또한, 제원이 다른 다수 CIWS를 동시 운용하는 경우를 추가로 고려한다. 본 논문에서는 이와 같은 문제 상황을 비선형 혼합정수계획 문제로 정식화하고, 이를 MILP로 변형하는 기법을 제시하였다. 또한, 이를 상용 최적화 프로그램으로 구현해 최적화 성능을 검증하였다.

• Journal of Information Processing Systems
• /
• 제8권3호
• /
• pp.409-420
• /
• 2012

The decision-making by agents in games is commonly based on reinforcement learning. To improve the quality of agents, it is necessary to solve the problems of the time and state space that are required for learning. Such problems can be solved by Macro-Actions, which are defined and executed by a sequence of primitive actions. In this line of research, the learning time is reduced by cutting down the number of policy decisions by agents. Macro-Actions were originally defined as combinations of the same primitive actions. Based on studies that showed the generation of Macro-Actions by learning, Macro-Actions are now thought to consist of diverse kinds of primitive actions. However an enormous amount of learning time and state space are required to generate Macro-Actions. To resolve these issues, we can apply insights from studies on the learning of tasks through Programming by Demonstration (PbD) to generate Macro-Actions that reduce the learning time and state space. In this paper, we propose a method to define and execute Macro-Actions. Macro-Actions are learned from a human subject via PbD and a policy is learned by reinforcement learning. In an experiment, the proposed method was applied to a car simulation to verify the scalability of the proposed method. Data was collected from the driving control of a human subject, and then the Macro-Actions that are required for running a car were generated. Furthermore, the policy that is necessary for driving on a track was learned. The acquisition of Macro-Actions by PbD reduced the driving time by about 16% compared to the case in which Macro-Actions were directly defined by a human subject. In addition, the learning time was also reduced by a faster convergence of the optimum policies.

• 한국정밀공학회지
• /
• 제15권1호
• /
• pp.78-86
• /
• 1998

This paper presents a numerical method of the minimum-time trajectory planning for a robot manipulator amid obstacles. Each joint displacement is represented by the linear combination of the finite-term quintic B-splines which are the known functions of the path parameter. The time is represented by the linear function of the same path parameter. Since the geometric path is not fixed and the time is linear to the path parameter, the coefficients of the splines and the time-scale factor span a finite-dimensional vector space, a point in which uniquely represents the manipulator motion. The displacement, the velocity and the acceleration conditions at the starting and the goal positions are transformed into the linear equality constraints on the coefficients of the splines, which reduce the dimension of the vector space. The optimization is performed in the reduced vector space using nonlinear programming. The total moving time is the main performance index which should be minimized. The constraints on the actuator forces and that of the obstacle-avoidance, together with sufficiently large weighting coefficients, are included in the augmented performance index. In the numerical implementation, the minimum-time motion is obtained for a planar 3-1ink manipulator amid several rectangular obstacles without simplifying any dynamic or geometric models.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2012년도 추계학술대회
• /
• pp.481-484
• /
• 2012

개정교육과정의 정보교과에서는 프로그래밍 학습을 통하여 문제해결능력을 신장할 수 있도록 하였으며, 실제로 많은 부분이 할애되었다. 그러나 컴퓨터 만을 활용한 프로그래밍 교육은 학습자의 몰입을 가져오게 하기 쉽지 않고, 학습 시 많은 논리적 부담을 일으키게 한다. 따라서 로봇을 활용한 프로그래밍 교육에 대한 연구가 많이 진행되고 있다. 또한 최근 몇 년 동안 스마트폰은 급속도로 보급되고 있는데, 스마트폰에 따른 몰입 현상과 부작용의 문제가 대두되고 있다. 본 연구에서는 로봇을 활용한 프로그래밍 교육에 스마트폰의 몰입효과를 이용하여 상승효과를 일으킬 수 있는 프로그래밍 교육 모형을 개발하고자 한다. 프로그래밍 학습 분야에서 절실히 요구되는 몰입을 스마트폰의 특징을 도입시켜 효과적인 프로그래밍 학습에 도움이 되도록 하고자 한다.

• 한국경영과학회지
• /
• 제25권4호
• /
• pp.55-65
• /
• 2000

This paper presents a linear programming based algorithm for pattern classification. Pattern classification is being considered to be critical in the area of artificial intelligence and business applications. Previous methods employing linear programming have been aimed at two-group discrimination with one or more linear discriminant functions. Therefore, there are some limitations in applying available linear programming formulations directly to general multi-class classification problems. The algorithm proposed in this manuscript is based on quadratic or polynomial discriminant functions, which allow more flexibility in covering the class regions in the N-dimensional space. The proposed algorithm is compared with other competitive methods of pattern classification in experimental results and is shown to be competitive enough for a general purpose classifier.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2003년도 춘계학술대회 논문집
• /
• pp.1731-1734
• /
• 2003

We propose a unmaned integrating control system based-on Windows XP version Off-Line Programming System which can simulate a Robot model in 3D Graphics space in this paper. The robot with 4 and 6 axes modeled SM5 and AM1 respectively were adopted as an objective model. Forward kinematics, inverse kinematics and robot dynamics modeling were included in the developed off-line program. The interface between users and the off-line programming system in the Windows XP's graphic user interface environment was also studied. The developing language is Microsoft Visual C++. Graphic libraries, OpenGL, by silicon Graphics, Inc. were utilized for 3D Graphics.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2003년도 춘계학술대회 논문집
• /
• pp.319-326
• /
• 2003

We propose a unmaned integrating control system based-on Windows XP version Off-Line Programming System which can simulate a Robot model in 3D Graphics space in this paper. The industrial robot with 4 and 6 axes modeled SM5 and AMI respectively were adopted as an objective model. Forward kinematics, inverse kinematics and robot dynamics modeling were included in the developed off-line program. The interface between users and the off-line programming system in the Windows XP's graphic user interface environment was also studied. The developing language is Microsoft Visual C++. Graphic libraries, OpenGL, by silicon Graphics, Inc. were utilized for 3D Graphics.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1999년도 제14차 학술회의논문집
• /
• pp.292-297
• /
• 1999

Evolving in artificial agent is an extremely difficult problem, but on the other hand, a challenging task. At present the studies mainly centered on single agent learning problem. In our case, we use simulated soccer to investigate multi-agent cooperative learning. Consider the fundamental differences in learning mechanism, existing reinforcement learning algorithms can be roughly classified into two types-that based on evaluation functions and that of searching policy space directly. Genetic Programming developed from Genetic Algorithms is one of the most well known approaches belonging to the latter. In this paper, we give detailed algorithm description as well as data construction that are necessary for learning single agent strategies at first. In following step moreover, we will extend developed methods into multiple robot domains. game. We investigate and contrast two different methods-simple team learning and sub-group loaming and conclude the paper with some experimental results.