臺灣博碩士論文加值系統

本論文主要研究的方向有兩個，第一個研究方向為運用圖形識別中的各種非監督式分類法來分析2006年世界盃足球賽的球隊實力分為幾類，分類方法有K-Means分類演算法、Fuzzy C-Means分類演算法、Hierarchical分類演算法、及Self-Organizing Feature Map分類演算法，經由2006年世界盃足球賽官方網站所得到的統計數據，除分析各種非監督式分類法的分類正確率及分類結果之外，同時也針對數據資料作分類個數的有效性驗證。第二個研究方向為運用監督式類神經網路中的多層感知器及倒傳遞學習演算法，並根據2006年世界盃足球賽前一階段的比賽數據來預測下一階段比賽兩隊的勝率，經由實驗結果，本論文建構的勝率預測模型準確率達62.5%，若不含比賽結果平手的場次，則準確率達76.9%。

There are two main research aspects in this thesis. The first research aspect is that we use 4 unsupervised clustering methods to analyze the team level of 2006 FIFA World Cup Football Games. These clustering methods include K-means clustering algorithm, Fuzzy C-means clustering algorithm, hierarchical clustering algorithm, and self-organizing feature map algorithm. Furthermore, we use 3 clustering validity methods to verify whether these football statistical data possesses a clustering property. The second research aspect is that we use the supervised multi-layer perceptron with back propagation learning rule to predict the winning rate based on two team’s previous football game’s records. According to the experimental results, it shows that the correct rate achieves 62.5% by using the adopted prediction model. If the draw games are excluded, the correct rate can achieve 76.9%.

中文摘要............................................. i
英文摘要............................................. ii
誌謝................................................. iii
目錄................................................. iv
表目錄............................................... vi
圖目錄 ........................................... vii


第一章 圖形識別與類神經網路於2006年世界盃足球賽球隊實
力之分類 ..................................... 1

1 介紹 ........................................... 1
2 採用的非監督式分類法簡介 ....................... 4
2.1 K-Means分類法 ................................ 4
2.2 Fuzzy C-Means分類法 .......................... 5
2.3 凝聚階層式分類法 ............................. 6
2.4 Self-Organizing Feature Map Neural Networks
SOFM）分類法 ................................. 7
2.5 順序性群聚分類演算法 ......................... 8
3 採用的分類有效性驗證法 ......................... 12
3.1 以BSC及MBSC為基礎的有效性驗證法 .............. 13
3.2 以K-Means為基礎的有效性驗證法 ................ 14
4 非監督式分類法用於2006年世界盃足球賽球隊實力的分
類 ............................................. 16
4.1 原始資料的擷取及轉換 ......................... 16
4.2 比賽結果的分類原則 ........................... 21
4.3 分類正確率計算 ............................... 22
5 實驗結果 ....................................... 24
5.1 非監督式分類法的實驗結果 ..................... 24
5.2 2006年世界盃足球賽統計數據分類個數有有效性的驗
證結果 ....................................... 34
5.3 主分量分析法用於2006年世界盃足球賽統計數據之分
類實驗結果 ................................... 36
6 結論 ........................................... 44
7 參考文獻 ....................................... 45


第二章 多層感知器類神經網路於2006年世界盃足球賽的勝率
預測 ......................................... 46

1 介紹 ........................................... 46
2 多層感知器(MLP)及倒傳遞(BP)學習演算法 .......... 50
2.1 多層感知器(MLP)網路架構 ...................... 50
2.2 倒傳遞(BP)學習規則 ........................... 50
3 2006年世界盃足球賽的勝率預測 ................... 53
3.1 原始資料的擷取與轉換 ......................... 53
3.2 網路訓練的目標值 ............................. 54
3.3 訓練資料的選取及預測隊伍之輸入資料的轉換 ..... 55
3.4 MLP的BP學習法則參數的決定 .................... 57
4 2006年世界盃足球賽勝率預測的結果與討論 ......... 61
4.1 勝率預測結果 ................................. 61
4.2 預測結果的討論 ............................... 64
5 結論 ........................................... 67
6 參考文獻 ....................................... 68

附錄一 2006世界盃足球賽賽程表 ...................... 69
附錄二 2006世界盃足球賽單場比賽報告 ................ 71
附錄三 2006世界盃足球賽比賽結果 .................... 72

第一章 圖形識別與類神經網路於2006年世界盃足球賽球隊實
力之分類

[1] 黃國源、張生平、張文龍、董安晉、及陳楷儒,“以類神經
網路分析2006世界盃足球賽球隊實力”, 研究報告, 1 - 4
頁, 六月八日, 2006.
[2] The match schedule, matches, results, and statistics
reports of 2006 FIFA world cup Germany,
http://fifaworldcup.yahoo.com/06/en/.
[3] Official website of FIFA,
http://www.fifa.com/en/index.html.
[4] Sergios Theodoridis and Konstantinos Koutroumbas,
Pattern Recognition, 3rd Edition, Academic Press,
Inc., New York, 2006.
[5] C. T. Lin and C. S. G. Lee, Neural Fuzzy Systems: A
Neuro-Fuzzy Synergism to Intelligent Systems,
Prentice Hall, 1996, 797 pages.
[6] Kou-Yuan Huang, Neural Networks and Pattern
Recognition, Weikeg Publishing Co., Taipei, Taiwan,
March 2003, 406 pages.
[7] Rui Xu and Donald Wunsch II, “Survey of clustering
algorithm,” IEEE Transactions on Neural Networks,
Vol.16, No.3, 2005, pp. 645-678.
[8] E. W. Forgy, “Cluster analysis of multivariate data
efficiency vs interpretability of classifications,”
Biometrics 21, 1965, pp. 768-769.
[9] J. C. Bezdek, Fuzzy Mathematics in Pattern
Classification, PhD Thesis, Cornell University,
Ithaca, NY. 1973.
[10] T. Kohonen, H. Riittinen, E. Reuhkala, and S.
Haltsonen, “On-line recognition of spoken words from
a large vocabulary,” Information Sciences Journal,
Vol. 33, Issue 1-2, 1984, pp. 3-30.
[11] T. Kohonen, “Self-organized formation of
topologically correct feature maps,”
Neurocomputting, 1988, pp. 509-521.
[12] T. Kohonen, “The elf-organizing map,” Proceedings of
the IEEE, Vol.78, N0.9, 1990, pp. 1464-1480.
[13] S. Kaski and T. Kohonen, “Winner-take-all networks
for physiological models of comp- petitive learning,”
Neural Networks, Vol. 7, Issue 6-7, 1994, pp. 973-984.
[14] T. Kohonen, “Physiological interpretation of the self-
organizing map algorithm,” Neural Networks, Vol. 6,
Issue 7, 1993, pp. 895-905.
[15] A. V. Hall, "Methods for demonstrating resemblance in
taxonomy and ecology," Nature, Vol. 214, 1967, pp. 830-
831.
[16] M. A. Vogel and A. K. C. Wong, “PFS clustering
method,” IEEE Transactions on Pattern Analysis and
Machine Intelligence, Vol, PAMI-1, No. 3, 1979, pp.
237-245.


第二章 多層感知器類神經網路於2006年世界盃足球賽的勝率預測

[1] 2006 FIFA world cup Germany – match schedule, matches
and results, and statistics reports,
http://fifaworldcup.yahoo.com/06/en/.
[2] Official website of FIFA,
http://www.fifa.com/en/index.html.
[3] C. P. Michael, “Neural networks quarterbacking-how
different training methods perform in calling the
games,” IEEE Potentials, August/September 1996, pp.
9-15.
[4] J. Park and M. E. J. Newman, “A network-based ranking
system for US college football,” Journal of
Statistical Mechanics: Theory and Experiment, IOP
publishing, Issue 10, 2005, pp.1-14.
[5] A. P. Rotshtein, M. Posner, and A. B. Rakityyanskaya,
“Football predictions based on a fuzzy model with
genetic and neural tuning,” Cybernetics and Systems
Analysis, Vol.41, No.4, 2005, pp.619-630.
[6] D. E. Rumelhart, G. E. Hinton, and R. J. Williams,
“Learning representations by backpropagating
errors,” Nature, Vol.233, 1986, pp.533-536.
[7] David E. Rumelhart and James L. McClelland, Parallel
Distributed Processing: Explorations in the
Microstructure of Cognition, Vol.1, MIT Press:
Cambridge, MA, 1986.
[8] Kou-Yuan Huang, Neural Networks and Pattern
Recognition, Weikeg Publishing Co., Taipei, Taiwan,
March 2003, 406 pages.
[9] R. A. Jacobs, “Increase rates of convergence through
learning rate adaptation,” Neural Networks, Vol.1,
1988, pp. 295-307.
[10] J. M. Zurada, “Lambda learning rule for feedforward
neural networks,” Proceedings of the IEEE
International Conference in Neural Networks, Vol.3,
1993, pp.1808-1811.
[11] V. Kecman, Learning and Soft Computing: Support Vector
Machines, Neural Networks, and Fuzzy Logic Models, MIT
Press: Cambridge, MA, 2001.