臺灣博碩士論文加值系統

本論文主要研究一具有模糊特性之量子類神經網路於非凸集系統辨證上，其利用量子類神經網路中隱藏單元的模糊特性，將輸入訊號予於量化至若干不同程度以解決未知的、不明確的、混亂的、複雜的訊號分類問題。
因傳統類神經網路是以一非線性辨別曲線去明確分割樣本，這樣的分割特性對於樣本資料中含有不明確的訊息時，往往無法有效分割。再者現實世界中，信息訊號存在著複雜且不完美的問題，傳統類神經網路受本身的限制以不膚使用，故吾人以具有模糊分割特性之量子類神經網路為研究方向，內容包括量子神經元(quantum neuron)特性、量子區間(quantum interval)、量子演算法(quantum algorithm)。模擬實驗於兩組非凸集樣本之辨識能力且與傳統類神經網路作比較，並驗證量子類神經網路優於傳統類神經網路。

In this thesis, the non-convex system identification by using quantum neural network (QNN) is studied and simulated. The signals with fuzziness and uncertainties are expected can be effectively identified since the structure of hidden units with various graded levels.
As we know, NN has been popularly applied in areas such as pattern recognition, signal processing, system identification and so on due to its powerful nonlinear mapping property and its ability to learn from training examples. Usually, the NN classifier can sort a set of feature patterns into clear and appropriate classes. However, in real world, many signals do exist the problem that the environment of signal information is complex and ill defined. Many unknown or uncertain factors or disturbances are encountered. Such a phenomenon make NN have an ill learning and then its performance is not so satisfactory. Therefore, in our studies, QNN is studied and expected to deal with the signals with fuzziness since its hidden units having various graded levels that are capable of classifying the features of signals. In this thesis, two sets of non-convex systems are studied and simulated.
For demonstrating the QNN do have the capability to detect the presence of fuzziness in the signals and can effectively deal with the signals whose features are overlapped, same simulations are performed as the comparisons by using traditional NN techniques.

誌謝……………………………………………………………………………i
中文摘要……………………………………………………………………ii
英文摘要……………………………………………………………………iii
目錄…………………………………………………………………………v
圖目錄………………………………………………………………………vii
表目錄………………………………………………………………………ix
第一章 前言…………………………………………………………………1
1.1 背景簡介…………………………………………………………1
1.2 研究動機…………………………………………………………1
1.3 論文架構…………………………………………………………2
第二章 類神經網路介紹……………………………………………………3
2.1 類神經元模型……………………………………………………3
2.2 網路架構 ………………………………………………………6
2.3 倒傳遞演算法……………………………………………………7
第三章 量子類神經網路…………………………………………………9
3.1 傳統類神經網路的限制…………………………………………9
3.2 量子類神經網路之架構…………………………………………11
3.2.1 量子神經元…………………………………………12
3.2.2 量子區間……………………………………………13
3.3 量子類神經網路之學習法則……………………………………15
3.3.1 權值更新……………………………………………15
3.3.2 量子區間更新………………………………………18
3.3.3 QNN之演算法…………………………………………20
第四章 模擬結果…………………………………………………………23
4.1 實驗模擬一：Book……………………………………………23
4.1.1 Theta的學習率之選擇……………………………24
4.1.2 初始Theta之範圍的選擇…………………………24
4.1.3 QNN v.s CNN………………………………………25
4.2 實驗模擬二：Taiji……………………………………………36
4.2.1 Theta的學習率之選擇……………………………36
4.2.2 初始Theta之範圍的選擇…………………………37
4.2.3 QNN v.s CNN………………………………………37
第五章 結論……………………………………………………………59
參考文獻……………………………………………………………………61

[1] R. O. Duda and P. E. Hart, Pattern Classification and Scene Analysis, New York:Wiley, 1973.
[2] T. Yoshida, and S. Omatu, Pattern recognition with neural networks, IEEE International Conference on Geoscience and Remote Sensing Symposiums , Vol. 2,pp. 699-701, 2000.
[3] Y. L. Li, K. Q. Wang, D. Zhang, Region-mapping neural network model for pattern recognition, IEEE International Conference on Machine Learning and Cybernetics, Vol. 3, pp. 1541-1545, 2002.
[4] J. Zhao, J. W. Wen, and S. W. Luo, A multi-moment pattern recognition method based on neural networks, IEEE International Conference on Signal Processing,Vol. 3, pp. 1675-1678, 2000.
[5] S. W. Moon and S. G. Kong, Pattern recognition with block-based neural networks,IEEE International Conference on Neural Networks , Vol. 1, pp. 992-996, 2002.
[6] J. L. Mitrpanont, and A. Srisuphab, The realization of quantum complex-valued back-propagation neural network in pattern recognition problem, IEEE International Conference on Neural Information Processing, Vol. 1, pp. 462-466, 2002.
[7] R. G. Hutchins, Neural network topologies and training algorithms in nonlinear system identification, IEEE International Conference on Systems, Vol. 3, pp.2512-2515, 1995.
[8] J. Baruch, F. Thomas; R. Garrido, and E. Gortcheva, A hybrid multimodel neural network for nonlinear systems identification, IEEE International Conference on Neural Networks, Vol. 6, pp. 4278-4283, 1999.
[9] I. S. Baruch, J. M. Flores, F. Thomas, and E.Gortcheva, A multimodel recurrent neural network for systems identification and control, IEEE International Conference on Neural Networks, Vol. 2, pp. 1291-1296, 2001.
[10] W. Yu, and X. Li, Some new results on system identification with dynamic neural networks, IEEE Transactions on Neural Networks, Vol. 12, pp. 412-417, 2001.
[11] S. K. Pal, S. Mitra, Multilayer perceptron, fuzzy sets, and classification, IEEE Transactions on Neural Networks, Vol. 3, pp. 683-697, 1992.
[12] J-S. R. Jang, ANFIS: adaptive-network-based fuzzy inference system, IEEE Transactions on Systems, Vol. 23 , pp. 665-685, 1993.
[13] S. Kak, On quantum neural computing, Information Sciences
, Vol. 83, pp. 143-160, 1995.
[14] I. E. Lagaris, A. Likas, D. I. Fotiadis, Artificial neural network methods in quantum mechanics, Computer Physics Communication, Vol. 104, pp. 1-14, 1997.
[15] G. Purushothaman and N. B. Karayiannis, Quantum neural networks (QNN’s): Inherently fuzzy feed-forward neural networks, IEEE Transactions on Neural Network, Vol. 8, pp. 679-693, May 1997.
[16] E. C. Behrman, L. R. Nash, J. E. Steck, V. G. Chandrashekar, and S. R. Skinner, Simulations of quantum neural networks, Information Sciences, Vol. 128, pp. 257
-269, 2000.
[17] Jie Zhou, Automatic detection of premature ventricular contraction using quantum neural networks, IEEE Symposium on Bioinformatics and Bioengineering, pp.169-173, 2003.
[18] Dokur, Z., Kurnaz, M.N., Olmez, T., Segmentation of ultrasound images by using quantizer neural network, IEEE Symposium on Computer-Based Medical Systems, pp. 257-261, June 2002.
[19] Fei Li, Shengmei Zhao, Baoyu Zheng, Quantum neural network in speech recognition, IEEE International Conference on Signal Processing, Vol. 2, pp. 1267-1270, 2002.
[20] J. Zhou, Q. Gan, A. Krzyzak, and C. Y. Suen, Recognition of handwritten numerals by quantum neural network with fuzzy features, International Journal on Document Analysis and Recognition(IJDAR), vol. 2, no. 1,pp. 30-36, 1999.
[21] H. Ishibuchi, M. Nii, Possibilistic fuzzy classification using neural networks, IEEE International Conference on Neural Networks, Vol. 3 , pp. 1433-1438, 1997.
[22] I. I. Sakhnin, M. T. Manry and H. Chandrasekaran, Iterative Improvement of trigonometric networks, IEEE International Conference on Neural Networks, Vol. 2, pp. 13-18, 1999.
[23] 蘇木春, 張孝德, 機器學習:類神經網路、模糊系統以及基因演算法, 全華科技圖書有限公司, 民國91年。
[24] 林昇甫, 洪成安, 神經網路入門與圖樣辨識, 全華科技圖書有限公司, 民國85年。
[25] 葉怡成, 類神經網路模式應用與實作, 儒林圖書有限公司, 民國87年。