臺灣博碩士論文加值系統

因人工智慧領域的蓬勃發展，許多專家學者致力於人工智慧技術挑戰人腦的研究，尤其是在棋類遊戲中。棋士對弈表現有時會受到當時的身心理狀況及外在環境而有所影響，使得棋力表現較不穩定，增加棋力評估的困難度。因此，本論文結合知識本體、第一型模糊集合、第二型模糊集合及基因學習演算法，以上界信賴搜尋樹之模擬次數及棋局資訊做為輸入，進行模糊推論系統於圍棋棋力適性評估之應用。首先，我們邀請業餘段位棋士與台灣及法國研究團隊共同研發之適性MoGoTW電腦圍棋程式進行對弈，藉由動態調整模擬次數，以達到適性評估與之對弈棋士棋力之目的。接著，本論文執行第一型模糊推論系統、第二型模糊推論系統及分別結合基因學習演算法學習專家知識，以推論對弈棋士棋力。最後，本論文將學習前與學習後實驗結果以及Bradley Terry Model產生之實驗結果，分別與棋士真實段位進行比較分析，以找出最適合棋士棋力評估值。由實驗結果可知: (1)不論是第一型模糊推論系統及第二型模糊推論系統，學習後的表現皆優於學習前的表現; (2)學習前第二型模糊推論系統所得到的實驗結果優於學習前第一型模糊推論系統的結果及(3)本論文所提方法於推論業餘棋士可能段位是可行的。未來，希望能直接結合適性版MoGoTW，在棋士與電腦圍棋程式完成對弈後即可得知可能的棋力。

Owing to the thriving in artificial intelligence (AI) research, many researchers have devoted themselves to challenging humans based on the AI techniques, especially the applications to the board games. One Go player’s performance sometimes may be affected by his/her mental and physical situations and external environment, so this causes the difficulties in evaluating the rank of one Go player. Hence, according to simulation number of Upper Confidence Tree (UCT) and the basic information of the collected games, this thesis combines ontology with type-1 fuzzy sets (T1FSs), type-2 fuzzy sets (T2FSs), and genetic learning algorithm to apply these technologies to make an adaptive assessment on game of Go. We first invite amateur Go players to play with adaptive MoGoTW, developed by Taiwanese and French team members, to make an assessment on Go player’s rank by dynamically adjusting the simulation number of UCT. Next, based on the constructed ontology, we use T1FLS and T2FLS to infer the specific Go player’s rank before learning and after learning the knowledge of the domain experts, respectively. Additionally, this thesis also uses Bradley Terry Model to assess the rank of the Go player. Finally, we compare the experimental results with the actual rank of the Go player. The experimental results show that (1) the proposed approach performs better after learning regardless of T1FLS or T2FLS, (2) T2FLS can acquire much better results than T1FLS before learning, and (3) the proposed approach is feasible for the application to adaptive assessment on game of Go. In the future, it is hope to combine with the adaptive MoGoTW to allow the invite Go players to know their evaluated rank after playing some games with the computer Go program.

目錄 X
圖目錄 XII
表目錄 XIV
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機 1
1.3 章節簡介 3
第二章 相關研究與文獻探討 4
2.1 本體論 4
2.1.1 本體論架構 4
2.1.2 圍棋棋力適性評估本體論 5
2.2 適性評量 7
2.3 模糊理論 7
2.3.1 第一型模糊系統 8
2.3.2 第二型模糊邏輯系統 10
2.3.3 模糊標記語言 13
2.4 上界信賴搜尋樹 16
2.5 基因學習 17
第三章 圍棋適性評估 21
3.1 適性版台灣魔圍棋簡介 21
3.2 系統架構 21
3.3 基於UCT可調式評估圍棋段位機制 25
3.4 模糊集合建構機制 28
3.4.1 棋局模擬數 28
3.4.2 棋局權重及棋局勝率 31
3.5 Bradley-Terry Model 評估機制 32
第四章 模糊推論機制於棋力適性評估系統應用 34
4.1 第一型模糊推論機制 34
4.2 第二型模糊推論機制 38
第五章 基因學習於棋力適性評估系統之應用 43
5.1 基因學習機制 43
5.2 基因學習機制限制式 46
5.3 訓練資料及測試資料建置機制 47
第六章 實驗結果 50
6.1 實驗架構 50
6.2 系統介面 51
6.3 基因學習結果分析 56
6.3.1 應用基因學習於第一型模糊推論系統 59
6.3.2 應用基因學習於第二型模糊推論系統 62
6.4 基因學習前與學習後實驗比較 66
6.4.1 基因學習前後MSE比較 66
6.4.2 適應值變動分析 70
6.5 Bradley-Terry模型與模糊推論實驗分析 72
第七章 結論及未來研究方向 77
7.1 結論 77
7.2 未來研究方向 78
Reference 80
附錄A-1 第一型模糊推論系統知識庫語法 84
附錄A-2 第二型模糊推論系統知識庫語法 85
附錄B 模糊推論系統規則庫 87
附錄C-1 基因學習-Training Data 89
附錄C-2 基因學習-Testing Data 91
附錄D 適性圍棋比賽棋士與棋局資料 93
附錄E 基因學習結果G=2000、CR=0.9、MR=0.1 96
附錄F-1群組T1-A 101
附錄F-2群組T1-B 104
附錄F-3群組T1-C 107
附錄F-4群組T1-D 110
附錄G-1群組T2-A 113
附錄G-2群組T2-B 116
附錄G-3群組T2-C 119
附錄G-4群組T2-D 122
附錄H-1 (T1FIS/0.75/0.05) 125
附錄H-2 (T2FIS/0.65/0.1) 128
附錄I 語意及段位Mapping結果 131

[1]李健興, “第二型適性評量模糊本體論建構與應用(II)(III),” 102年度專題研究計畫申請書.
[2]李允中, 王小璠, 蘇木春, ”模糊理論及其應用”, 全華科技圖書股份有限公司, 2003.
[3]余民寧, “試題反應理論的介紹,” 2013 [Online]. Available: http://www.edutest.com.tw/e-irt/irt12.htm.
[4]M. Harmsen, B. Fischer, H. Schramm, T. Seidl, and T. M. Deserno, “Support vector machine classification based on correlation prototypes applied to bone age assessment,” IEEE Journal of Biomedical and Health Informatics, vol. 17, no. 1, pp. 190-197, Jan. 2013.
[5]S. Ji, W. Xu, M. Yang, and K. Yu, “3D convolutional neural networks for human action recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 1, pp. 221-231, Jan. 2013.
[6]A. Baylari and G. A. Montazer, “Design a personalized e-learning system based on item response theory and artificial neural network approach,” Expert Systems with Applications, vol. 36, no. 4, pp. 8013-8021, 2009.
[7]Y. C. Kiong, S. Palaniappan, and N. A. Yahaya, “Health ontology system,” in Proceeding of the 7th International Conference on Information Technology in Asia (CITA 11), Jul. 12-13, 2011, pp. 1-4.
[8]F. Bobillo, U. Straccia, “Fuzzy ontology representation using OWL2,” International Journal of Approximate Reasoning, vol. 52, no. 7, pp. 1073-1094, 2011.
[9]A. J. C. Trappey, C. V. Trappey, F. C. Hsu, and D. W. Hsiao, “A fuzzy ontological knowledge document clustering methodology,” IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, vol. 39, no. 3, pp. 806-814, 2009.
[10]C. S. Lee, Z. W. Jian, and L. K. Huang, “A fuzzy ontology and its application to news summarization,” IEEE Transactions on Systems, Man and Cybernetics Part B: Cybernetics, vol. 35, no. 5, pp. 859-880, 2005.
[11]C. S. Lee, M. H. Wang, and J. J. Chen, “Ontology-based intelligent decision support agent for CMMI project monitoring and control,” International Journal of Approximate Reasoning, vol. 48, vol. 1, pp. 62-76, 2008.
[12]Y. T. Yu and C. C. Hsu, “A structured ontology construction by using data clustering and pattern tree mining,” in Proceeding of 2011 International Conference on Machine Learning and Cybernetics (ICMLC), Guilin, China, Jul. 10-13, 2011, pp. 45-50.
[13]C. S. Lee, M. J. Wu, M. H. Wang, O. Teytaud, H. M. Wang and S. J. Yen, “T2FML-based adaptive assessment system for computer game of Go,” 2013 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE 2013), Hyderabad, India, Jul. 7-10, 2013.
[14]C. M. Chen and L. J. Duh, “Personalized web-based tutoring system based on fuzzy item response theory,” Expert Systems with Applications, vol. 34, no. 4, pp. 2298-2315, 2008.
[15]J. Yen and R. Langari, Fuzzy Logic-intelligence, control and information, New Jersey, Prentice Hall, 1999.
[16]L. A. Zadeh, “Fuzzy sets,” Information and Control, vol. 8, no. 3, pp. 338-353, 1965.
[17]J. M. Mendel, “Type-2 fuzzy sets and systems: an overview,” IEEE Computational Intelligence Magazine, vol.2, no.1, pp.20-29, Feb. 2007.
[18]Q. Liang, and J. M. Mendel, “Interval type-2 fuzzy logic systems: theory and design,” IEEE Transactions on Fuzzy Systems, vol. 8, pp. 535­550, 2000.
[19] J. M. Mendel, Uncertain Ruled-Based Fuzzy Logic Systems: Introduction and New Directions. Upper Saddle River, NJ: Prentice Hall, 2001.
[20]G. Acampora, V. Loia, C. S. Lee, and M. H. Wang, On the Power of Fuzzy Markup Language. Berlin, Germany: Springer-Verlag, Jan. 2013.
[21]C. B. Browne, E. Powley, D. Whitehouse, S. M. Lucas, P. I. Cowling, P. Rohlfshagen, S. Tavener, D. Perez, S. Samothrakis, and S. Colton, “A survey of Monte Carlo tree search methods,” IEEE Transactions on Computational Intelligence and AI in Games, vol. 4, no. 1, pp. 1-43, 2012.
[22]C. S. Lee, M. H. Wang, C Chaslot, J. B. Hoock, A. Rimmel, O. Teytaud, S. R. Tsai, S. C. Hsu, and T. P. Hong, “The computational intelligence of MoGo revealed in Taiwan''s computer Go tournaments,” IEEE Transactions on Computational Intelligence and AI in Games, vol. 1, no. 1, pp. 73-89, Mar. 2009.
[23]S. Samothrakis, D. Robles, S. Lucas, “Fast Approximate Max-n Monte Carlo Tree Search for Ms Pac-Man,” IEEE Transactions on Computational Intelligence and AI in Games, vol. 3, no. 2, pp. 142-154, Jun. 2011.
[24]S. Gelly and D. Silver, “Monte-Carlo tree search and rapid action value estimation in computer Go,” Artificial Intelligence, vol. 175, no.11, pp. 1856-1875, 2011.
[25]G. Chaslot, “Monte-Carlo tree search,” Maastricht University, The Netherlands, 2010.
[26]S. Sumino, A. Mutoh, and S. Kato, “Evolutionary approach of reward function for reinforcement learning using genetic programming,” in Proceeding of 2011 International Symposium on Micro-NanoMechatronics and Human Science (MHS), Nagoya, Japan, Nov. 6-9, 2011, pp. 385-390.
[27]Human vs. Computer Go Competition @ IEEE WCCI 2012, 2013 [Online]. Available: http://oase.nutn.edu.tw/wcci2012/.
[28]Invited Game fpr MoGoTW vs. Human Go Player, 2013 [Online]. Available: http://go.nutn.edu.tw/2012/.
[29]MoGo-Taiwan Research and Development Center, 2013 [Online]. Available http://mogotw.nutn.edu.tw/
[30]Use R for Statistics, 2013 [Online]. Available:
https://sites.google.com/site/rlearningsite/analysis/catagory/bradleyterry.
[31]A. Stein, J. Aryal, and G. Gort, “Use of the Bradley-Terry model to quantify association in remotely sensed images,” IEEE Transactions on Geoscience and Remote Sensing, vol. 43, no. 4, pp. 852-856, Apr. 2005.
[32]M. H. Wang, Y. T. Tsai, K. H. Lin, C. S. Lee, and C. H. Liu, “FML-based decision support system for solar energy supply and demand analysis,” 2013 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE 2013), Hyderabad, India, Jul. 7-10, 2013.
[33]C. H. Wang, T. P. Hong, and S. S. Tseng, “Integrating Fuzzy Knowledge by Genetic Algorithms,” IEEE Transactions on Evolutionay Computation, vol. 2, no. 4, pp. 138-149, Nov. 1998.
[34]C. H. Wang, M. B. Chang, T. P. Hong, and S. S. Tseng, "Integrating multiple rule sets by genetic algorithms," in Proceeding of IEEE International Conference on Systems, Man, and Cybernetics, Oct. 11-14,1998, vol. 2, pp.1524-1529
[35]C. S. Lee, S. M. Guo, and C. Y. Hsu, "A novel fuzzy filter for impulse noise removal," Lecture Notes in Computer Science, vol. 3174/2004, Springer-Verlag, 2004.
[36] C. S. Lee, M. H. Wang, H. Hagas, Z. W. Chen, S. T. Lan, S. E. Kuo, H. C. Kuo, and H. H. Cheng, "A novel genetic fuzzy markup language and its application to healthy diet assessment," (SCI) International Journal of Uncertainty, Fuzziness, and Knowledge-Based Systems, vol. 20, no. 2, pp. 247-278, Oct. 2012.