A Real Time Strategy (RTS) Game is a game which simulates the real war and can be seen as a useful way to develop tactical, strategic and doctrinal solutions. To defeat enemies with equivalent competitive conditions is a novel and interesting issue in developing combating strategies in real-time strategy games.

Genetic algorithms (GA) are heuristic search methods which simulate the natural evolution according to Darwin's theory of evolution and appropriate for solving search and optimization problem. GA is effective in the problems which contain uncertainty and then suitable in the developing strategies in RTS game. Transfer Learning is a method to improve the learning in the new task via using the previous learning experience and result in the similar tasks. We can save or avoid the learning time in solving the new problem by using the transfer learning.

In this thesis, first we implement genetic algorithms by developing the military strategies and tactical formations in a simplified RTS game platform then build robust AI bots to defeat enemies with fixed formations and strategies. In the second part, we use transfer learning to improve and accelerate the evolving procedure of genetic algorithms by modifying the population via the evolved robust AI bots. The last part we combine the evolved result to construct a robust AI bot without evolution and gain the improvement. The winning percentage of the combination result is better than a robust bot which is evolved in another similar domain; it shows the success of applying transfer learning in this domain.

即時戰略遊戲為模擬戰爭之遊戲，並可視為一種用於發展戰術、策略之良好方法。於即時戰略遊戲中，如何於對等條件中，發展良好的策略及隊形以擊敗敵人為一值得探討的議題。

基因演算法(Genetic Algorithm)為以自然演化概念形成之演算法，適用於各式求解及最佳化問題，並於充斥不確定因素之問題環境有良好成效；而即時戰略遊戲之戰爭策略發展即為其適切之應用領域。轉移學習(Transfer Learning)為一利用先前機器學習所學習過的經驗及結果，改善或強化新問題之學習過程之方法，利用轉移學習，將可節省於新問題之機器學習時間，甚者更可直接利用經驗結果進行分析應用於新問題之解決，無需學習時間。

在此篇論文中，藉一簡化之即時戰略遊戲平台，首先以基因演算法進行戰術策略及戰鬥隊形之研發演化，用於建立強大人工智慧之機器人程序(AI bot)，其為一能擊敗特定敵人之隊形與戰略組合。次利用轉移學習進行基因演算法之改善，利用已演化之經驗結果之強大人工智慧機器人程序，改良原基因演算法之初始族群分布，以加速演化速度。第三部份以轉移學習之方式，將已演化之經驗結果進行組合，結合為無需演化過程之人工智慧機器人程序，並獲得初步成效，組合結果之勝率已優於直接利用演化結果進行對戰之勝率，說明轉移學習應用於此問題之成功。

S. Bakkes, P. Spronck, and J. van den Herik. Rapid and reliable adaptation of video game ai. Computational Intelligence and AI in Games, IEEE Transactions on, 1(2):93 {104, june 2009. ISSN 1943-068X. doi: 10.1109/TCIAIG.2009.2029084.

N. Beume, T. Hein, B. Naujoks, N. Piatkowski, M. Preuss, and S. Wessing. Intelligent anti-grouping in real-time strategy games. In IEEE Symposium on Computational Intelligence and Games (CIG 2008). Piscataway, NJ: IEEE Press,2008.

H. Danielsiek, R. Stuer, A. Thom, N. Beume, B. Naujoks, and M. Preuss. Intelligent moving of groups in real-time strategy games. pages 71 {78, 15-18 2008doi: 10.1109/CIG.2008.5035623.

S. Erdtman and J. Fylling. Pathnding with hard constraints. 2008.

Jonas Flensbak. Flock behavior based on inuence maps. 2007.

Ji-Lung Hsieh and Chuen-Tsai Sun. Building a player strategy model by analyzing replays of real-time strategy games. In Neural Networks, 2008. IJCNN2008. (IEEE World Congress on Computational Intelligence). IEEE International Joint Conference on, pages 3106 {3111, 1-8 2008a. doi: 10.1109/IJCNN.2008.4634237.

Ji-Lung Hsieh and Chuen-Tsai Sun. Building a player strategy model by analyzing replays of real-time strategy games. In Neural Networks, 2008. IJCNN2008. (IEEE World Congress on Computational Intelligence). IEEE International Joint Conference on, pages 3106 {3111, 1-8 2008b. doi: 10.1109/IJCNN.2008.4634237.

H. Jones and M. Snyder. Supervisory control of multiple robots based on a real-time strategy game interaction paradigm. In Systems, Man, and Cybernetics,2001 IEEE International Conference on, volume 1, pages 383 {388 vol.1, 2001.doi: 10.1109/ICSMC.2001.969842.

D. Keaveney and C. O'Riordan. Evolving robust strategies for an abstract real-time strategy game. In Computational Intelligence and Games, 2009. CIG2009. IEEE Symposium on, pages 371 {378, 7-10 2009. doi: 10.1109/CIG.2009.5286453.

S. Mulgund, K. Harper, K. Krishnakumar, and G. Zacharias. Air combat tacticsoptimization using stochastic genetic algorithms. In Systems, Man, and Cy-bernetics, 1998. 1998 IEEE International Conference on, volume 4, pages 3136 {3141 vol.4, 11-14 1998. doi: 10.1109/ICSMC.1998.726484.

S.J. Pan and Q. Yang. A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering, 2009.

T. Revello and R. McCartney. Generating war game strategies using a genetic algorithm. In Evolutionary Computation, 2002. CEC'02. Proceedings of the 2002 Congress on Evolutionary Computation, volume 2, pages 1086{1091, 2002.

R. R rmark. Thanatos-a learning rts game ai. Master's thesis, 2009.

M. Sharma, M. Holmes, J. Santamaria, A. Irani, C. Isbell, and A. Ram. Transfer learning in real-time strategy games using hybrid CBR/RL. In Proceedings of the Twentieth International Joint Conference on Articial Intelligence, 2007.

D. W. Aha, M. Molineaux, and M. Ponsen. Learning to win: Case-based plan selection in a real-time strategy game. Case-Based Reasoning Research and Development, pages 5{20, 2005.

Haibo Wang, P.H.F. Ng, Ben Niu, and S.C.K. Shiu. Case learning and indexing
in real time strategy games. In Natural Computation, 2009. ICNC '09. Fifth International Conference on, volume 1, pages 100 {104, 14-16 2009. doi: 10. 1109/ICNC.2009.729.

Fei Yang, Qing Wang, and Wu Zhendong. Cooperative combat system action planning method based on multi-agent system. In Education Technology and Computer Science (ETCS), 2010 Second International Workshop on, volume 1, pages 490 {493, 6-7 2010. doi: 10.1109/ETCS.2010.582.

Liu Yuefeng and Zhang An. Multi-agent system and its application in combat simulation. In Computational Intelligence and Design, 2008. ISCID '08 International Symposium on, volume 1, pages 448 {452, 17-18 2008. doi:10.1109/ISCID.2008.45.