• 정보처리학회논문지B
• /
• 제12B권2호
• /
• pp.215-222
• /
• 2005

본 연구에서는 웹에서 사용자가 방문하고자 하는 곳을 찾아가는 엔트리 페이지 검색을 위한 메타검색 방식을 제안한다. 기존의 연구에서 메타 검색이 여러 검색 엔진에서 많이 나타나는 중복된 문서를 강조하는 방식인 반면에 비해, 본 연구에서는 문서의 중복 개념을 확장하여 특정 도메인 및 디렉토리에서 나온 문서들도 중복되었다고 가정하여 메타검색에 이용하는 방식을 보인다. TREC에 제출된 시스템들의 결과물과 상용 검색 엔진의 결과물을 이용하여, 확장된 중복을 이용한 메타 검색의 유용성을 실험한다. 수행된 실험을 통해서 문서의 단순 중복을 이용하는 기존의 방식이 내용 기반 검색에 유용한 반면, 엔트리 페이지 검색에 있어서는 본 연구에서 제안하는 확장된 중복 방식이 기존 방식의 성능보다 $10\%$ 이상의 성능 향상을 얻을 수 있음을 알 수 있었다.

• 로봇학회논문지
• /
• 제14권1호
• /
• pp.40-49
• /
• 2019

Reinforcement learning has been applied to various problems in robotics. However, it was still hard to train complex robotic manipulation tasks since there is a few models which can be applicable to general tasks. Such general models require a lot of training episodes. In these reasons, deep neural networks which have shown to be good function approximators have not been actively used for robot manipulation task. Recently, some of these challenges are solved by a set of methods, such as Guided Policy Search, which guide or limit search directions while training of a deep neural network based policy model. These frameworks are already applied to a humanoid robot, PR2. However, in robotics, it is not trivial to adjust existing algorithms designed for one robot to another robot. In this paper, we present our implementation of Guided Policy Search to the robotic arms of the Baxter Research Robot. To meet the goals and needs of the project, we build on an existing implementation of Baxter Agent class for the Guided Policy Search algorithm code using the built-in Python interface. This work is expected to play an important role in popularizing robot manipulation reinforcement learning methods on cost-effective robot platforms.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 Proceedings of the Korea Automatic Control Conference, 11th (KACC); Pohang, Korea; 24-26 Oct. 1996
• /
• pp.243-246
• /
• 1996

Modular robot manipulator is a robotic system assembled from discrete joints and links into one of many possible manipulator configurations. This paper describes the design method of newly developed modular robot manipulator and the methodology of a task based reconfiguration of it. New locking mechanism is proposed and it provides quick coupling and decoupling. A parallel connection method is devised and it makes modular robot manipulator working well and the number of components on each module reduced. To automatically determine a sufficient or optimal arrangement of the modules for a given task, we also devise an algorithm that automatically generates forward and inverse manipulator kinematics, and we propose an algorithm which maps task specifications to the optimized manipulator configurations. Efficient genetic algorithms are generated and used to search for a optimal manipulator from task specifications. A few of design examples are shown.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 하계종합학술대회 논문집(3)
• /
• pp.27-30
• /
• 2000

Many human inventions were inspired by nature. The artificial neural network is one example. Another example is Genetic Algorithms(GA). GAs search by simulating evolution, starting from an initial set of solutions or hypotheses, and generating successive "generations" of solutions. This particular branch of AI was inspired by the way living things evolved into more successful organisms in nature. To simulate the process of GA in a computer, we must simulate many times according to varying many GA parameters. This paper describes the implementation of Java Package for efficient applications on Genetic Algorithms, called "JavaGA". The JavaGA used as a application program as well as applet provides graphical user interface of assigning major GA parameters.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2001년도 Proceedings ICPE 01 2001 International Conference on Power Electronics
• /
• pp.404-408
• /
• 2001

This paper deals with speed control of induction motors with a gain tuning based on simple Genetic Algorithms, which are search algorithms based on the mechanics of natual selection and genetics. Based on the designed control system structure, the indirect vector control system of induction motors is simulated. The simulation results show that the system has a strong robust to the parameter variation and is insensitive to the load disturbance. Thus, the proposed PI controller based on genetic algorithms is superior to manually tuned classical PI controller in improving the speed control performance of induction motors.

• 산업경영시스템학회지
• /
• 제20권41호
• /
• pp.15-23
• /
• 1997

Because of various request of consummer and rapidly chang, a product change and new-production come out variously. To satisfy the condition, companies must develop the product of rapidity and good quality. But, a product design difficults to consider many parameters and increase the level of each parameter. In order to solve this problem, this paper studies out algorithms taken into account more parameters and increased the level of parameters using the Genetic Algorithms. Because this algorithm can search detailed and wide for the level of parameter, in case of new-product development, we can use it for designing parameters of new-product.

• 산업경영시스템학회지
• /
• 제20권42호
• /
• pp.171-179
• /
• 1997

Genetic Algorithms face an inherent conflict between exploitation and exploration. Exploitation refers to taking advantage of information already obtained in the search. Exploration show that a pattern in bits coupled with another pattern elsewhere in the string is more effective. In this paper shows that the selection method has a major impact on the balance between exploitation and exploration. A more heavy-handed approach seeks to exploit the available information. If decisions must be made quickly, especially those in real-time trading environments, then quicker convergence through exploitation may be more desirable. Also this paper we present some theoretical and empirical the selection method in genetic algorithms for a GA-hard problem.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제14권1호
• /
• pp.43-56
• /
• 2010

The representative numerical algorithms to solve the time dependent Navier-Stokes equations are projection type methods. Lots of projection schemes have been developed to find more accurate solutions. But most of projection methods [4, 11] suffer from inconsistency and requesting unknown datum. E and Liu in [5] constructed the gauge method which splits the velocity $u=a+{\nabla}{\phi}$ to make consistent and to replace requesting of the unknown values to known datum of non-physical variables a and ${\phi}$. The errors are evaluated in [9]. But gauge method is not still obvious to find out suitable combination of discrete finite element spaces and to compute boundary derivative of the gauge variable ${\phi}$. In this paper, we define 4 gauge algorithms via combining both 2 decomposition operators and 2 boundary conditions. And we derive variational derivative on boundary and analyze numerical results of 4 gauge algorithms in various discrete spaces combinations to search right discrete space relation.

• 한국경영과학회지
• /
• 제32권2호
• /
• pp.109-121
• /
• 2007

This paper focuses on algorithms based on the evolution, which is applied to various optimization problems. Especially, among these algorithms based on the evolution, we investigate the simple genetic algorithm based on Darwin's evolution, the Lamarckian algorithm based on Lamark's evolution and the Baldwin algorithm based on the Baldwin effect and also Investigate the difference among them in the biological and engineering aspects. Finally, through this comparison, we suggest a new algorithm to find more various solutions changing the genotype or phenotype search space and show the performance of the proposed method. Conclusively, the proposed method showed superior performance to the previous method which was applied to the constrained minimum spanning tree problem and known as the best algorithm.

• 정보처리학회논문지B
• /
• 제9B권1호
• /
• pp.7-16
• /
• 2002

웹 검색 엔진의 검색 결과를 클러스터링하는 후처리 클러스터링 알고리즘은 그 특성상 일반적인 클러스터링 알고리즘과는 다른 요구조건을 갖는다. 본 논문에서는 이러한 후처리 클러스터링 알고리즘의 요구조건들을 최대한 만족하는 새로운 클러스터링 알고리즘을 제안하고자 한다. 제안된 Concept ART는 문서 클러스터링에 있어 여러 가지 장점을 갖는 개념 벡터와 실시간 클러스터링 알고리즘으로 알려진 Fuzzy ART를 결합한 형태로써, 후처리 클러스터링뿐 아니라 범용의 클러스터링 알고리즘으로도 응용이 가능하다.