바둑은 적어도 2,500년 이상의 역사를 지녔고 그동안 인간 고유의 게임 영역으로 여겨왔으나, 2016년 컴퓨터 바둑인 알파고에 의해 제압된 마지막 보드게임이 되었다. 바둑에서의 사활문제는 컴퓨터 바둑을 구축 시 반 드시 해결해야 되는 기본 문제 영역이 된다. 연역적 추론은 이미 알고 있는 판단을 근거로 새로운 사실을 추 론하는 논리학의 용어이다. 본 논문에서는 연역적 추론을 위해 제약 충족 방법을 활용하여 사활문제와 직결 되는 3궁, 4궁, 5궁, 6궁의 원형 안형을 표현하는 4-튜플의 형식을 찾고자 했다. 이후 생성된 4-튜플의 형식 을 갖고 점 패턴 매칭을 활용하여 각 궁도의 원형 안형의 갯수를 파악하고자 했다. 실험 결과에 따른 4-튜플 형식의 갯수는 3궁 1개, 4궁 3개, 5궁 4개, 6궁 8개가 있음을 알 수 있었다. 또한 각 궁도의 원형 안형의 갯 수는 3궁 2개, 4궁 5개, 5궁 12개, 6궁 35개가 존재함을 찾아냈다. 마지막으로 컴퓨터 바둑에서의 사활문제 해결을 위해 원형 안형들을 4-튜플 형식의 변형인 5-튜플 형식으로도 제시하였다.

바둑게임은 중국에서 적어도 2,500년 전에 창안되었으며, 게임을 이기기 위해 상대방보다 더 많은 영역을 둘러싸는 것을 목표로 한다. 간단한 규칙에도 불구하고 바둑은 인공지능(AI) 분야에서 아주 복잡한 전략적인 보드게임이다. 몬테카를로 트리탐색(MCTS)은 국면 평가의 어려움을 극복하고 게임트리 내의 엄청난 분기수를 줄여주는 최상우선탐색 알고리즘이다. 컴퓨터바둑 게임에서 MCTS는 수많은 플레이아웃을 수행한 후, 게임트리 내에 있는 현재노드로부터 단말노드까지의 임의의 수순을 표본 추출하여 다음 후보 착수에 대한 승률을 근사적으로 계산해낸다. AlphaGo를 포함한 모든 강력한 컴퓨터바둑 프로그램은 컴퓨터바둑 게임 진행에 있어 MCTS를 이용하여 가장 유망한 착수를 구해왔다. 본 연구는 소형바둑 컴퓨터게임에 있어 순수 MCTS를 이용하여 가장 유망한 일련의 수순을 찾고자 했다. 실험결과에 의하면 순수 MCTS는 소형바둑에서 19줄바둑게임에서 벌어지는 포석, 중반전, 끝내기라는 3단계가 아닌 중반전과 끝내기라는 단지 2단계를 진행하였으며, 아울러 순수 MCTS는 맥과 연단수와 같은 하위개념의 지식을 이해 못하는 것을 보였다.

Monte-Carlo Tree Search (MCTS) is a best-first search algorithm to evaluate states of the game tree in game playing, and has been successfully applied to various games, especially to the game of Go. Upper Confidence Bounds for Trees (UCT), which is a variant of MCTS, uses the UCB1 formula as selection policy, and balances exploitation and exploration of the states. Rapid Action-Value Estimation (RAVE), which is a All-Moves-As-First (AMAF) heuristic, treats all moves in a simulation as the first move, and therefore updates the statistics of all children of the root node. In this paper, we evaluate the performance of RAVE and UCT playing against each other in the game of Tic-Tac-Toe. The experimental results show that the first player RAVE is much inferior to the second player UCT (13.0±0.7%); on the other hand, the first player UCT is far superior to RAVE (99.9±0.1%).

최근 기후변화에 따라 전 세계적으로 큰 규모의 산불 발생이 증가하고 있으며, 우리나라도 산불발생이 전반적으로 증가하는 경향을 보인다. 최근 우리나라와 인접한 북한에서 발생한 산불이 비무장 지대 등의 국내 영토로 번지는 사건이 많이 발생하고 있다. 우리나라에서는 오픈 API(application programming interface)를 통하여 과거 산불발생 기록을 검색할 수 있으나 행정기관에서 수집한 자료여서 국내 지역으로 정보가 국한되어 있다. 이에 본 연구에서는 접근불능지역인 북한을 포함한 한반도의 산불발생에 대한 장기 시계열 정보를 생산하기 위하여 2000년부터 2015년까지의 MODIS(Moderate Resolution Imaging Spectroradiometer) 위성자료를 수집 및 재가공하여 일자별 산불발생 현황의 GIS 데이터베이스를 구축하고자 한다. 데이터베이스 구축을 위한 입력자료로서 Terra 위성의 MOD14A1 산출물과 Aqua 위성의 MYD14A1 산출물을 사용하였으며, 16년간의 일자별 영상을 일괄작업으로 재처리하여 산불의 발생일자, 발생 위치, 탐지 신뢰도, 방사열 에너지(fire radiative power: FRP) 등의 정보를 추출하고 이들을 결합하여 Shapefile 형태로 생성하였다. 본 연구에서 구축한 장기시계열 GIS 데이터베이스는 다른 연구자들에 의해 한반도 산불정보의 시공간 특성 분석에 활용될 수 있으며, 이를 위하여 인터넷에 결과 파일을 탑재하여 자유롭게 다운로드 가능하도록 하였다.

Go is an extremely complex strategic board game despite its simple rules and is the great challenging classic game for AI due to its enormous search space. The computer program AlphaGo finally defeated Fan Hui, the European Go champion, without handicaps on a full-sized 19 ×19 board in October 2015. Monte-Carlo Tree Search (MCTS) is a widely-used algorithm for game-tree search in game playing. MCTS based on statistical sampling is a best-first tree search technique to evaluate states; UCT which is a variant of MCTS uses the UCB1 formula as selection policy. In this paper, we evaluate the performance of MCTS and UCT playing against each other in the game of Tic-Tac-Toe. The experimental results show that the first player UCT is slightly superior to the second player MCTS (54.3±1.0%), the first player is always advantageous to the second player regardless of the MCTS and UCT players, and the result of each game should be a tie if both players do their best in Tic-Tac-Toe.

Go is an extremely complex strategic board game despite its simple rules. Recently computer Go based on MCTS plays at human-master level and also has defeated top professional players with handicap games in 19×19 Go. Before implementing computer Go, in this paper we show weakness of pure MC algorithm for playing robust Tic-Tac-Toe game and present alternative method to make up the weakness. Furthermore we show how UCB algorithm works for balancing exploration and exploitation in game tree and discuss the need of a hybrid algorithm combined with UCB and strategy based MCTS, for implementing an enhanced computer Go.

Evolutionary computation is a powerful tool for developing computer games. Back-propagation neural network(BPNN) was proved to be a universal approximator and genetic algorithm(GA) a global searcher. The game of Tic-Tac-Toe, also known as Naughts and Crosses, is often used as a test bed for testing new AI algorithms. We tried to recognize the strategic fitness of a finished Tic-Tac-Toe game when the parameters, such as a sequence of moves, its game depth and result, are provided. To implement this, we've constructed an evolutionary model using GA with back-propagation NNs(GANN). The experimental results revealed that GANN, in the very long training time, converges very slowly; however, performance of recognizing the strategic fitness does not meet we expected and, further, increase of the population size does not significantly contribute to the performance of GANN.

The game of Go is an oriental strategic board game originated from China at least more than 2,500 years ago. The Monte-Carlo Tree Search (MCTS) algorithm in Go is a method that uses a large number of simulations to approximately estimate the winning rate of candidate moves by sampling the game. The two computer Go programs called Crazy Stone and Mogo defeated human Go professionals on the 9⨯9 board in 2006. Prior to our implementing MCTS into computer Go, we tried to find out the best move sequence in playing Tic-Tac-Toe game as a test bed. The experimental results revealed that the first player should play the center to ensure the highest winning rate, and the game result becomes a draw if two players do their best.

One of the oldest board games in the world is the game of Go which is originated at least more than 2,500 years ago. In spite of its long history, the theoretical studies concerning to Go openings are still insufficient. We firstly used related-samples t-test using SPSS to the three countries' Go mean openings to find out their similarities and differences. Experimental result shows that there are no significant differences (p=.959) between Korean and Chinese mean openings, but meanwhile, there are just some slight similarities (p=.061) between Korean and Japanese mean openings. We secondly applied PCA and LDA classifiers with Euclidean distance to correctly classify a pro player's opening into his/her class obtained from the training openings. Result shows that the recognition rate (42.1%) of dependent LDA classifier is much better than that (27.9%) of PCA classifier; and also that dependent LDA classifier outperforms independent LDA classifier with the recognition rate of 35.0% .

Although the history of the game of Go is more than 2,500 years, the theoretical studies of Go are still insufficient. In recent years a lot of studies using Artificial Intelligent (AI) have been conducted, but they do not provide the prominent theoretical reality. We applied traditional Principal Component Analysis (PCA) algorithm to the Go openings, which are the early stage in Go, to analyze them especially focused on the Go game records of the Korean top 10 professional Go players. We firstly analyzed the number of most significant eigenvectors capturing most of variance. Experimental result shows that among the 361 eigenvectors the eight most significant eigenvectors capture most of the variance (96.2%). We secondly used PCA classifier with Euclidean distance to recognize a pro player's opening to a class obtained from the training openings. Result shows that the best average recognition rate of 22% is so much lower than the recognition rates reported in face recognition research.