臺灣博碩士論文加值系統

近幾年來，人工智慧(Artificial Intelligence)的話題充斥著整個世界科學領域中，人工智慧中的機器學習(Machine Learning) 占最大一部份，藉由著軟體編程使電腦像人類依樣具有學習的能力，就如人類剛開始學習時需藉由著分類(Classification)，才能分析理解並且進行判斷，故我們希望藉由人工神經網路進行分類問題與系統辨識模擬反應出不同學習方法優劣性。在這篇研究中，為了增加神經網路對分類問題及非線性系統辨識之學習效果，利用一種為具有自我回饋因子的艾爾曼神經網路(Elman Neural Network)，經由資料的回饋過程增加其訊息聯想度及記憶性。本論文中為了針對分類問題及系統辨識問題進行資料學習，則利用以模仿物種在棲息地之間遷移的生物習性為基礎概念的「生物地理學演算法(Biogeography-Based Optimization)」，以雜草在土地空間中隨機分佈種子找尋空間最佳解為想法的「雜草入侵演算法(Invasive Weed Optimization」及為了增加植物之間分享特徵而加入生物遷徙機制的「混和型IWO/BBO演算法」，求解問題最佳化。為了探討啟發式演算法之間在訓練神經網路方面的效果，使用四種分類問題數據庫包含了鳶尾花數據庫、乳癌數據庫、心臟疾病數據庫和肺癌數據庫以及三種非線性系統辨識問題包含了Mackey-Glass、henon system和nonlinear plant，進行實驗模擬，最後實驗記錄中使用均方誤差(MSE)、標準差(STD)和正確率進行演算法之間效能比較。結果顯示，使用Hybrid IWO/BBO方法能得到比其他學習演算法更佳精確的正確率以及能夠收斂到較低的均方誤差。

The most popular topic of engineering science area is artificial intelligence recently. Therefore, as human beings begin to study through distinguish from category, need to analyze and understand by judgment and make a final decision. Moreover, it is hoped that the classification problem and system identification can reflect the advantages and disadvantages of different learning methods by artificial neural network. In order to improve the classification system capacity and accuracy of the solution, it has changed the new architecture what is named Elman neural network with self-feedback factor. It is used to increase the associative degree and memory through the data feedback process. In order to learn data for purpose of solving the classification problem and nonlinear identification problem, using the algorithm based on metaheuristic what named BBO, it is based on the biological habit that imitate species migration between habitats. Though randomly reproduce seeds to find the best solution in the search space which is called IWO. At last, Hybrid IWO/BBO algorithm increase the biological migration mechanism by sharing features among plants. With a view to exploring the effect of heuristic algorithms on training neural networks, using four kind of classification datasets (including Iris dataset, Breast cancer datasets, SPECT heart datasets and Lung datasets) and three plants of nonlinear system identification (including mackey-glass, Henon system and nonlinear plant). There are used to simulate and compare the result. The experimental results clearly indicate that Hybrid IWO/BBO method is better than other optimization algorithms. It means the Hybrid algorithm can present recognition rate and converge to a lower MSE.

目錄
摘要 i
Abstract ii
誌謝 iii
目錄 iv
表目錄 vii
圖目錄 viii
第一章、緒論 1
1.1 研究動機與目的 1
1.2 文獻回顧 2
1.3 論文架構 4
第二章、系統估測 5
2.1 分類問題 5
2.2 系統辨識 6
2.2.1 數學模型的種類 6
2.2.2 非線性系統辨識 7
2.2.3 混沌系統 8
2.2.4 混沌系統的特點 8
2.3 類神經網路 10
2.3.1 前饋式類神經網路: 11
2.3.2 回饋式類神經網路: 12
2.3.3 類神經網路學習方式 13
第三章、多層感知器網路及回饋式神經網路 15
3.1 多層感知器神經網路架構 15
3.1.1 梯度下降法(GD) 16
3.2 艾爾曼回饋式神經網路架構 20
3.2.1 動態Elman網路 21
3.2.2 即時回饋學習演算法 23
第四章、啟發式演算法 26
4.1 生物地理學以及生物地理學演算法 26
4.1.1 生物地理學介紹 27
4.1.2 生物地理學之數學模型 28
4.1.3 生物地理學優化演算法 32
4.2 雜草入侵演算法 37
4.2.1 雜草生態學 37
4.2.2 雜草入侵演算法 38
4.2.3 雜草入侵演算法流程圖 43
4.3 雜草入侵/生物地理學混和型演算法(Hybrid IWO/BBO 44
4.3.1 Hybrid IWO/BBO遷移機制 44
4.3.2 Hybrid IWO/BBO演算法流程圖 46
第五章、模擬步驟與結果 47
5.1 分類問題之實驗步驟與結果 49
5.1.1 鳶尾花數據庫 (Iris Dataset) 50
5.1.2 乳癌數據庫 (Breast Cancer Dataset) 52
5.1.3 心臟疾病數據庫 (SPECT Heart Dataset) 54
5.1.4 肺癌數據庫 (Lung Cancer Dataset) 55
5.2 非線性系統辨識之實驗步驟與結果 57
5.2.1 Mackey-Glass time series 58
5.2.2 Henon system 63
5.2.3 Nonlinear plant 66
第六章、結論與未來展望 71
6.1 結論 71
6.2 未來展望 72
參考文獻 73

參考文獻
[1]X. Wu, et al. “Top 10 Algorithms in Data Mining,” Know ledge and Information System, vol. 14(1), pp.1-37, 2008.
[2]Y.C. Chang, W.Y. Kao, C.P. Chua, and C.H. Chiu, “A learning style classification mechanism for e-learning,” Computers & Education, vol.53(2), pp. 273-285, 2009.
[3]S, Mirjalili, S. Mirjalili. A. Lewis, “Let a Biogeography-Based Optimizer Train Your Multi-Layer Perceptron,” Information Sciences 269, 188-209, 2004.
[4]O. Claveria, E. Monte, S. Torra, “Multiple-input Multiple-output vs. Single-input Single-output Neural Network Forecasting,”.
[5]J. Wang, J. Wang, M. Zeng, J. Wang, “Prediction of Internet Traffic Based on Elman Neural Network,” Chinese Control and Decision Conference, 2009.
[6]Elman JL. Finding structure in time. Cognitive Sci 1990;2(14):179–211. A. Kalinli, S. Sagiroglu, “Elman Network with Embedded Memory for System Identification,” Journal of Information Science and Engineering 22, 1555-1568, 2006.
[7]Koker R. , “Design and Performance of An Intelligent Predictive Controller for A Six Degree of Freedom Robot Using the Elman Network,” Information Sciences 2006;176(12):1781–99.
[8]R. Sarker, J. Kamruzzaman, and C. Newton, “Evolutionary optimization (evopt): A Brief Review and Analysis,” International Journal of Computational Intelligence and Applications, vol. 3, no. 4, pp. 311–330, 2003.
[9]D. Simon, “Biogeography-Based Optimization,” IEEE Trans. Evol. Computer., vol. 12, no. 6, pp. 702-713, Dec. 2009.
[10]G. Wang, L. Guo, H. Duan, H. Wang, L. Liu, and M. Shao, “Hybridizing harmony search with biogeography based optimization for global numerical optimization,” Journal of Computational and Theoretical Nanoscience, vol. 10, no. 10, pp. 2312–2322, 2013.
[11]W. Gong, Z. Cai, and C. X. Ling, “DE/BBO: A Hybrid Differential Evolution with Biogeography-Based Optimization for Global Numerical Optimization,” Soft Computing, vol. 15, no. 4, pp. 645–665, 2010.
[12]G. Khademi, H. Mohammadi, and D. Simon, “Hybrid Invasive Weed/Biogeography-Based Optimization,” Submitted for Publication, http://embeddedlab.csuohio.edu/BBO/IWO.html
[13]McCulloch, Warren S., Pitts, Walter. “A Logical Calculus of The Ideas Immanent in Nervous Activity,” The Bulletin of Mathematical Biophysics. 1943-12-01, 5 (4): 115–133. doi:10.1007/BF02478259. ISSN 0007-4985
[14]J .Han, M .Kamber , M. Pei , “Data Mining(Third Edition),” 2012.
[15]劉金琨，沈曉蓉，趙龍，“系統辨識理論及MATLAB仿真”，電子工業出版社，北京市，2013。
[16]Alan V. Oppenheim , Alan S. Willsky , S. Hamid Nawab , ”Signals and Systems,” 黃永達譯，東華書局，台北市，1999。
[17]劉金琨，“先進PID控制MATLAB仿真”，電子工業出版社，北京市，2004。
[18]張斐章，張麗秋， “類神經網路導論:原理與應用”， 滄海書局， 台中市，1999。
[19]Y.C Cheng, W.M Qi, W.Y Cai, “Dynamic Properties of Elman and Modified Elman Neural Network,” Proceedings of First International on Machine Learning and Cybernetics, Beijing, 2002.
[20]C. Darwin, The Origin of Species. New York: Gramercy, 1995.
[21]P. Judea, “Heuristics: Intelligent Search Strategies for Computer Problem Solving,” United States: Addison-Wesley Pub. Co., Inc, Reading, 1984.
[22]A. Wallace, “The Geographical Distribution of Animals (Two Volumes),” Boston, MA: Adamant Media Corporation, 2005.
[23]R. MacArthur and E, Wilson, “The Theory of Biogeography,” Princeton, NJ: Princeton Univ. Press, 1967.
[24]I. Hanski and M. Gilpin, “Metapopulation Biology,” New York: Academic, 1997.
[25]T. Wesche, G. Goertler, and W. Hubert, “Modified Habitat Suitability Index Model for Brown Trout in Southeastern Wyoming,” North Amer. J. Fisheries Manage., vol. 7, pp. 232-237, 1987.
[26]A.R. Mehrabian, C. Lucas, “A Novel Numerical Optimization Algorithm Inspired from Weed Colonization,” Ecological Informatics,355-366, 2006.
[27]A.R. Mehrabian, A. Yousefi-Koma, “A Novel Technique for Optimal Placement of Piezoelectric Actuators on Smart Structures.” Journal of the Franklin Institute, 12-23, 348(1), 2011.
[28]W. Zhang, Y. Wang, G. Cui, et al. “Application of a Novel IWO to the Design of Encoding Sequences for DNA Computing,” Computers and Mathematics with Applications, 57(11/12), 2009.
[29]UCI Repository of Machine Learning Databases.
http://www.ics.uci.edu/~mlearn/MLRepository.html (accessed 2016).
[30]C.J. Lin, C.H. Chen, “A Compensation-based Recurrent Fuzzy Neural Network for Dynamic System Identification,” European Journal of Operational Research, Vol.172, pp.696-715, 2006.