• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2016년도 추계학술대회
• /
• pp.405-408
• /
• 2016

바둑, 체스, 장기와 같은 게임은 사람들의 두뇌발달에 도움을 주어왔다. 이 게임들은 컴퓨터 프로그램으로도 개발되었으며, 혼자서도 게임을 즐길 수 있도록 많은 알고리즘들이 개발되었다. 사람을 이기는 체스 프로그램은 1990년대에 개발된 것에 비해 바둑은 경우의 수가 너무 많아서 프로 바둑기사를 이기기는 불가능한 것으로 여겨졌다. 하지만 MCTS(Monte Carlo Tree Search)와 CNN(Convolutional Neural Network)의 이용으로 바둑 알고리즘의 성능은 큰 향상을 이루었다. 본 논문에서는 CNN과 MCTS를 사용하여 바둑 알고리즘의 개발을 진행하였다. 바둑의 기보가 학습된 CNN을 이용하여 최적의 수를 찾고, MCTS를 이용하여 게임의 시뮬레이션을 진행하여 이길 확률을 계산한다. 또한 기존 기보를 이용하여 바둑의 패턴 정보를 추출하고, 이를 이용하여 속도와 성능 향상을 도모하였다. 이 방법은 일반적으로 사용되는 바둑 알고리즘들에 비해 성능 향상이 있었다. 또한 충분한 Computing Power가 제공되면 더욱 성능이 향상될 것으로 보인다.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2011년도 정기총회 및 추계학술대회 논문집
• /
• pp.607-613
• /
• 2011

This paper reviews search methods of multiple reasonable paths to implement multimodal public transportation network of Seoul. Such a large scale multimodal public transportation network as Seoul, the computation time of path finding algorithm is a key and the result of path should reflect route choice behavior of public transportation passengers. Search method of alternative path is divided by removing path method and deviation path method. It analyzes previous researches based on the complexity of algorithm for large-scale network. Applying path finding algorithm in public transportation network, transfer and loop constraints must be included to be able to reflect real behavior. It constructs the generalized cost function based on the smart card data to reflect travel behavior of public transportation. To validate the availability of algorithm, experiments conducted with Seoul metropolitan public multimodal transportation network consisted with 22,109 nodes and 215,859 links by using the deviation path method, suitable for large-scale network.

• ETRI Journal
• /
• 제42권3호
• /
• pp.366-375
• /
• 2020

The network traffic prediction of a smart substation is key in strengthening its system security protection. To improve the performance of its traffic prediction, in this paper, we propose an improved gravitational search algorithm (IGSA), then introduce the IGSA into a wavelet neural network (WNN), iteratively optimize the initial connection weighting, scalability factor, and shift factor, and establish a smart substation network traffic prediction model based on the IGSA-WNN. A comparative analysis of the experimental results shows that the performance of the IGSA-WNN-based prediction model further improves the convergence velocity and prediction accuracy, and that the proposed model solves the deficiency issues of the original WNN, such as slow convergence velocity and ease of falling into a locally optimal solution; thus, it is a better smart substation network traffic prediction model.

• 한국시뮬레이션학회:학술대회논문집
• /
• 한국시뮬레이션학회 2002년도 추계학술대회 논문집
• /
• pp.163-170
• /
• 2002

Terrain surfaces have to be modeled in very detail and wheel-surface contacting geometry must be well defined in order to obtain proper ground-reaction and friction forces for realistic simulation of off-road vehicles. Delaunay triangulation is one of the most widely used methods in modeling 3-dimensional terrain surfaces, and T-search is a relevant algorithm for searching resulting triangular polygons. The T-search method searches polygons in successive order and may not allow real-time computation of off-road vehicle dynamics if the terrain is modeled with many polygons, depending on the computer performance used in the simulation. In order to accelerate the searching speed of T-search, a terrain database of triangular polygons is modeled in multi-levels by adopting the LOD (Level of Detail) method used in realtime computer graphics. Simulation results show that the new LOD search is effective in shortening the required computing time. The LOD search can be even further accelerated by introducing an NN (Neural Network) algorithm, in the cases where a appropriate range of moving paths can be predicted by cultual information of the simulated terrain, such as lakes, houses, etc.. Numerical tests show that LOD-NN search almost double the speed of the original T-search.

• 한국국방경영분석학회지
• /
• 제30권1호
• /
• pp.30-47
• /
• 2004

This research considers a Network Diversion Problem (NDP) in the directed graph, which is to identify a minimum cost set of links to cut so that any communication paths from a designated source node to a destination node must include at least one link from a specified set of arcs which is called the diversion arcs. We identify a redundant constraint from an earlier formulation. The problem is known to be NP-hard, however a detailed proof has not been given. We provide the proof of the NP-hardness of this problem. We develop a tabu search algorithm that includes a preprocessing procedure with two steps for removing diversion arcs as well as reducing the problem size. Computational results of the algorithm on instances of general graphs and grid graphs are reported.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.1034-1038
• /
• 2005

This paper presents a novel method to speed up neural network (NN) based face detection systems. NN-based face detection can be viewed as a classification and search problem. The proposed method formulates the search problem as an integer nonlinear optimization problem (INLP) and develops an improved genetic algorithm (IGA) to solve it. Each individual in the IGA represents a subwindow in an input image. The subwindows are evaluated by how well they match a NN-based face filter. A face is indicated when the filter response of the best particle is above a given threshold. Experimental results show that the proposed method leads to a speedup of 83 on $320{\times}240$ images compared to the traditional exhaustive search method.

• 한국통신학회논문지
• /
• 제18권3호
• /
• pp.360-371
• /
• 1993

Flooding 탐색 알고리즘은 고도의 생존성과 신뢰성이 있으므로 전술응용을 위한 효과적 라우팅 메커니즘이라고 알려져 있다. 그러나 망 효율성 측면에서는 큰 결점을 갖고 있다. 본 논문에서는 음성 트래픽에 대하여 최대 4개의 링크와 2개의 우선순위클래스를 갖는 전술 회선교환 격자망을 MFDL 경로기법을 사용하여 호설정시간 증가 및 알고리즘의 프로세서로딩없이 회선교환망의 블로킹 확률에 대한 성능개선 방법을 제안하였다.

• 전자공학회논문지B
• /
• 제33B권12호
• /
• pp.95-105
• /
• 1996

This paper presents structure optimization of a feedforward neural netowrk controller using the genetic algorithm. It is important to design the neural network with minimum structure for fast response and learning. To minimize the structure of the feedforward neural network, a genralization of multilayer neural netowrks, the genetic algorithm uses binary coding for the structure and floating-point coding for weights. Local search with an on-line learnign algorithm enhances the search performance and reduce the time for global search of the genetic algorithm. The relative fitness defined as the multiplication of the error and node functions prevents from premature convergence. The feedforward neural controller of smaller size outperformed conventional multilayer perceptron network controller.

• 유통과학연구
• /
• 제14권2호
• /
• pp.41-45
• /
• 2016

Purpose - Finding an optimal path is an essential component for the design and operation of smart transportation or logistics network. Many applications in navigation system assume that travel time of each link is fixed and same. However, in practice, the travel time of each link changes over time. In this paper, we introduce a new transportation problem to find a latest departing time and delivery path between the two nodes, while not violating the appointed time at the destination node. Research design, data, and methodology - To solve the problem, we suggest a mathematical model based on network optimization theory and a backward search method to find an optimal solution. Results - First, we introduce a dynamic transportation problem which is different with traditional shortest path or minimum cost path. Second, we propose an algorithm solution based on backward search to solve the problem in a large-sized network. Conclusions - We proposed a new transportation problem which is different with traditional shortest path or minimum cost path. We analyzed the problem under the conditions that travel time is changing, and proposed an algorithm to solve them. Extending our models for visiting two or more destinations is one of the further research topics.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제6권12호
• /
• pp.565-572
• /
• 2017

본 연구는 기존의 수요 예측 등의 시계열 연구에서 주로 사용되는 ARIMA 모형의 어려움을 극복하고자 인공신경망(Artificial neural network) 모형을 이용하여 한국 프로 야구 관중 수를 예측하였다. 훈련 자료로는 2015년 3월부터 9월까지의 일별 KBO 관중 수 자료를 대상으로 하였다. 전방향 신경망(Feedforward neural network)의 모형 훈련 과정에서, 그리드 탐색(Grid search)을 적용하여 최적의 초모수(Hyperparameter)를 찾고자 하였다. 그 결과, 그리드 탐색법의 최적 모형을 이용한 평균 절대 백분율 오차(MAPE)는 평균 20.9% 였다. 앙상블 기법을 이용한 모형의 MAPE는 평균 20.0%였다. 이는 다중회귀와 비교해보았을 때, 평균적으로 각각 26.3%, 30.3% 높은 예측력을 보인다.