• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제9권2호
• /
• pp.83-89
• /
• 2005

In this paper a new algorithm of learning and evolving artificial neural networks using gene expression programming (GEP) is presented. Compared with other traditional algorithms, this new algorithm has more advantages in self-learning and self-organizing, and can find optimal solutions of artificial neural networks more efficiently and elegantly. Simulation experiments show that the algorithm of evolving weights or thresholds can easily find the perfect architecture of artificial neural networks, and obviously improves previous traditional evolving methods of artificial neural networks because the GEP algorithm imitates the evolution of the natural neural system of biology according to genotype schemes of biology to crossover and mutate the genes or chromosomes to generate the next generation, and the optimal architecture of artificial neural networks with evolved weights or thresholds is finally achieved.

• 산업경영시스템학회지
• /
• 제31권3호
• /
• pp.17-23
• /
• 2008

In this research article, scheduling a casting sequence in a job-shop type foundry involving a variety of casts made of an identical alloy but with different shapes and II weights, has been investigated. The objective is to produce the assigned mixed orders satisfying due dates and obtaining the highest ingot efficiency simultaneously. Implementing simple integer programming instead of complicated genetic algorithms accompanying rigorous calculations proves that it can provide a feasible solution with a high accuracy for a complex, multi-variable and multi-constraint optimization problem. Enhancing the ingot efficiency under the constraint of discrete ingot sizes is accomplished by using a simple and intelligible algorithm in a standard integer programming. Employing this simple methodology, a job-shop type foundry is able to maximize the furnace utilization and minimize ingot waste.

• 전기학회논문지
• /
• 제67권6호
• /
• pp.767-772
• /
• 2018

Recently, some meta-heuristic algorithms, such as GA(Genetic Algorithm) and GP(Genetic Programming), have been used to optimize CNN(Convolutional Neural Network). The CNN, which is one of the deep learning models, has seen much success in a variety of computer vision tasks. However, designing CNN architectures still requires expert knowledge and a lot of trial and error. In this paper, the recent attempts to automatically construct CNN architectures are investigated and analyzed. First, two GA based methods are summarized. One is the optimization of CNN structures with the number and size of filters, connection between consecutive layers, and activation functions of each layer. The other is an new encoding method to represent complex convolutional layers in a fixed-length binary string, Second, CGP(Cartesian Genetic Programming) based method is surveyed for CNN structure optimization with highly functional modules, such as convolutional blocks and tensor concatenation, as the node functions in CGP. The comparison for three approaches is analysed and the outlook for the potential next steps is suggested.

• 한국지능시스템학회논문지
• /
• 제15권5호
• /
• pp.626-631
• /
• 2005

본 논문에서는 Dynamic Programming(DP)을 적용한 트리구조 미로내의 목표물 탐색 알고리즘을 구현한다. DP는 큰 문제를 이루는 작은 문제들을 먼저 해결하고 작은 문제들의 최적해를 이용하여 순환적으로 큰 문제를 해결한다. 먼저 실험을 위해 적외선 센서론 부착한 소형 이동 로봇과, 'Y'형태로 갈라진 길을 연결한 트리 구조의 미로 환경을 구성한다. 실험에서는 두 개의 서로 다른 알고리즘 - 좌수법, DP - 을 사용하여 목표물 탐색을 시도한다. 마지막으로 위 실험을 통해 DP를 미로 탐색문제에 적용했을 때의 성능을 검증한다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제5권4호
• /
• pp.333-338
• /
• 2005

This paper presents the target object search algorithm under Dynamic Programming (DP) in the Tree-type maze. We organized an experimental environment with the concatenation of Y-shape diverged way, small mobile robot, and a target object. By the principle of optimality, the backbone of DP, an agent recognizes that a given whole problem can be solved whether the values of the best solution of certain ancillary problem can be determined according to the principle of optimality. In experiment, we used two different control algorithms: a left-handed method and DP. Finally we verified the efficiency of DP in the practical application using our real robot.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제5권4호
• /
• pp.626-648
• /
• 2011

Energy consumption of wireless access networks is in permanent increase, which necessitates development of more energy-efficient network management approaches. Such management schemes must result with adaptation of network energy consumption in accordance with daily variations in user activity. In this paper, we consider possible energy savings of wireless local area networks (WLANs) through development of a few integer linear programming (ILP) models. Effectiveness of ILP models providing energy-efficient management of network resources have been tested on several WLAN instances of different sizes. To cope with the problem of high computational time characteristic for some ILP models, we further develop several heuristic algorithms that are based on greedy methods and local search. Although heuristics obtains somewhat higher results of energy consumption in comparison with the ones of corresponding ILP models, heuristic algorithms ensures minimization of network energy consumption in an amount of time that is acceptable for practical implementations. This confirms that network management algorithms will play a significant role in practical realization of future energy-efficient network management systems.

• 한국경영과학회지
• /
• 제11권2호
• /
• pp.1-6
• /
• 1986

Capacitated facility location problems have received a great deal of attention in the past decade, resulting in a proliferation of algorithms for solving them. As is the case with mixed 1-1 integer programming problems, the computational success of such algorithms depends greatly on how to obtain lower bounds in good quality within a resonable time. The objective of this paper is to provide a comparative analysis of those algorithms in terms of lower bounds they produce. Analyses of the strategies for generating lower bounds as well as the quality of generated lower bounds are provided.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2006년도 학술발표회 논문집
• /
• pp.354-359
• /
• 2006

The analysis of large-scale water resources systems is often complicated by the presence of multiple reservoirs and diversions, the uncertainty of unregulated inflows and demands, and conflicting objectives. Reinforcement learning is presented herein as a new approach to solving the challenging problem of stochastic optimization of multi-reservoir systems. The Q-Learning method, one of the reinforcement learning algorithms, is used for generating integrated monthly operation rules for the Keum River basin in Korea. The Q-Learning model is evaluated by comparing with implicit stochastic dynamic programming and sampling stochastic dynamic programming approaches. Evaluation of the stochastic basin-wide operational models considered several options relating to the choice of hydrologic state and discount factors as well as various stochastic dynamic programming models. The performance of Q-Learning model outperforms the other models in handling of uncertainty of inflows.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.40-43
• /
• 1996

An off-line programming, OLP, system is widely used in automation fines. To help an on-line robot system to carry out desirable tasks planned by the off-line simulation, an approach to the real-time communication is presented. The OLP system developed consists of a software, a host computer(PC), a SCARA robot body, four servo drivers, and four independent joint controllers. This study focuses on the software where real-time communication is included. The software, can be used in teaching, trajectory planning, real-time running, and performance evaluation. The evaluation of different control algorithms is one of the merits of the software. The software can give servo commands for task running. A comparison of generated and corresponding actual trajectories provides the evaluation of task performance. The safety, of the OLP system is ensured by alarming malfuntions of the system. The OLP system developed can reduce the teaching time and increase the user's convenience.

• Structural Engineering and Mechanics
• /
• 제6권8호
• /
• pp.857-870
• /
• 1998

This paper deals with a special class of inverse problems in discrete structural plasticity involving the identification of elastic limits and hardening moduli on the basis of information on displacements. The governing equations lead naturally to a special and challenging optimization problem known as a Mathematical Program with Equilibrium Constraints (MPEC), a key feature of which is the orthogonality of two sign-constrained vectors or so-called "complementarity" condition. We investigate numerically the application of two simple algorithms, both based on the use of the general purpose nonlinear programming code CONOPT accessed via the GAMS modeling language, for solving the suitably reformulated problem. Application is illustrated by means of two numerical examples.