• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.535-538
• /
• 2003

We employ Genetic Programming (GP) which is optimized with Simulated Annealing (SA) to recognize characteristic of a plan. Its result is described in Laplace function. The algorithm proceeds with automatic PID designs for the plant.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2002.05a
• /
• pp.159-165
• /
• 2002

진화연산에는 교배, 돌연변이, 경쟁, 선택이 있다. 이러한 과정 중에서 선택은 새로운 개체를 생산하지는 않지만, 모든 해중에서 최적의 해가 될만한 해는 선택하고, 그러지 않은 해는 버리는 판단의 역할을 한다. 따라서 아무리 좋은 해를 만들었다고 해도, 취사 선택을 잘못하면, 최적의 해를 찾지 못하거나, 또 많은 시간이 소요되게 된다. 따라서 본 논문에서는 stochastic한 성질을 갖고 있는 Tournament selection에 Local selection개념을 도입하여, 지역 해에서 벗어나 전역 해를 찾는데, 개선이 될 수 있도록 하였고 Fast Evolutionary Programming의 mutation과정을 개선하고, Genetic Algorithm의 연산자인 crossover와 mutation을 도입하여 Parallel search로 지역 해에서 벗어나 전역 해를 찾는 하이브리드 알고리즘을 제안하고자 한다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04c
• /
• pp.425-427
• /
• 2003

RNA는 세포내에서 유전자 발현에 직, 간접적으로 중요한 역할을 하며, RNA 구조는 세포 내에서의 기능과 깊은 연관이 있기 때문에 RNA 구조를 예측하는 것은 중요한 의미를 갖는다, 본 논문에서는 진화연산의 한가지인 유전자 프로그래밍(genetic programming) 방법을 사용하여 염기서열 정보를 참고하는 RNA 구조 문법의 학습 방법을 보여 준다. 이 RNA 구조를 의미하는 문법을 트리(tree)형태의 함수로 코드화(encoding) 한 후 이것을 유전자 프로그래밍 방법으로 진화시킨다. 진화를 통해 최적의 적합도를 갖는 트리의 문법을 테스트 데이터를 통해 평가한 결과 0.893의 특이도(speicificity)와 0.752의 민감도(sensitivity)를 보였다.

• Journal of Ocean Engineering and Technology
• /
• v.17 no.6
• /
• pp.91-95
• /
• 2003

Yacht design is significantly affected by the hull geometrical characteristics. Therefore, it is necessary to closely examine the relation between hull and performance, before considering characteristics of sea condition. In this study, Genetic Programming is used to derive a formula the relationship between hull geometric characteristics and performance. Using the formula, a new guideline is proposed to determine performance of a yacht.

• The Transactions of the Korean Institute of Electrical Engineers A
• /
• v.48 no.9
• /
• pp.1154-1160
• /
• 1999

A rule based controller is constructed for multiple robots accomplishing a given task in unknown environments by using genetic programming. The example task is playing a simplified soccer game, and the controller for robots that governs emergent cooperative behavior is successfully found using the proposed procedure A neural network controller constructed using the rule based controller is shown to be applicable in a more complex environment.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.146.4-146
• /
• 2001

Various methods have been proposed for the time series prediction. Most of the conventional methods only optimize parameters of mathematical models, but to construct an appropriate functional form of the model is more difficult in the first place. We employ the Genetic Programming (GP) to construct the functional form of prediction models. Our method is distinguished because the model parameters are optimized by using Back-Propagation (BP)-like method and the prediction model includes discontinuous functions, such as if and max, as node functions for describing complicated phenomena. The above-mentioned functions are non-differentiable, but the BP method requires derivative. To solve this problem, we develop ...

• Proceedings of the KIEE Conference
• /
• 1998.07g
• /
• pp.2197-2199
• /
• 1998

Many multiagents cooperative problems, such as table transport problem, require several emergent behaviors and a proper coordination of these is essential for successful accompishment of the task. We study in this paper the genetic programming method, called fitness switching, to evolve cooperation strategies of robots in these kind of tasks and show simulation results to demonstrate its effectiveness.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.63 no.3
• /
• pp.389-394
• /
• 2014

Two evolutionary gait generation methods for Cartesian and Joint coordinates space are compared to develop a fast locomotion for quadruped robots. GA(Genetic Algorithm) based approaches seek to optimize a pre-selected set of parameters for the locus of paw and initial position in cartesian coordinates space. GP(Genetic Programming) based technique generate few joint trajectories using symbolic regression in joint coordinates space as a form of polynomials. Optimization for two proposed methods are executed using Webots simulation for the quadruped robot which is built by Bioloid. Furthermore, simulation results for two proposed methods are analysed in terms of different coordinate spaces.

• Computers and Concrete
• /
• v.7 no.3
• /
• pp.221-237
• /
• 2010

In this study 72 different high strength concrete (HSC) mixes were produced according to the Taguchi design of experiment method. The specimens were divided into four groups based on the range of their compressive strengths 40-60, 60-80, 80-100 and 100-125 MPa. Each group included 18 different concrete mixes. The slump and air-content values of each mix were measured at the production time. The compressive strength, splitting tensile strength and water absorption properties were obtained at 28 days. Using this data the Genetic Programming technique was used to construct models to predict mechanical properties of HSC based on its constituients. These models, together with the cost data, were then used with a Genetic Algorithm to obtain an HSC mix that has minimum cost and at the same time meets all the strength and workability requirements. The paper describes details of the experimental results, model development, and optimization results.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.10a
• /
• pp.420-426
• /
• 1998

Evolutionary algorithms are probabilistic optimization algorithms based on the model of natural evolution. Recently the efforts to improve the performance of evolutionary algorithms have been made extensively. In this paper, we introduce the research for improving the convergence rate and search faculty of evolution algorithms by using reinforcement learning. After providing an introduction to evolution algorithms and reinforcement learning, we present adaptive genetic algorithms, reinforcement genetic programming, and reinforcement evolution strategies which are combined with reinforcement learning. Adaptive genetic algorithms generate mutation probabilities of each locus by interacting with the environment according to reinforcement learning. Reinforcement genetic programming executes crossover and mutation operations based on reinforcement and inhibition mechanism of reinforcement learning. Reinforcement evolution strategies use the variances of fitness occurred by mutation to make the reinforcement signals which estimate and control the step length.