臺灣博碩士論文加值系統

前饋式類神經網路(Feed-forward Neural Network, FNN)，為人工神經網路的一種，已大量運用在醫療診斷、資料探勘、證券市場分析上等領域，在FNN學習過程中，其主要目標是找到連接權重和偏差的最佳組合，以實現最小誤差。然而，大多數時候FNN會陷入局部(local)最佳解而非全域(global)最佳解。換句話說，學習過程中會將FNN引向局部最小值而不是域最小值。如何優化其連接權重和偏差，來達到誤差最小化，為本研究探討目的之一。
為了驗證本研究所提出的方法之有效性，利用公開的慢性腎臟疾病(chronic kidney disease, CKD)與間皮瘤疾病(mesothelioma diseases, MES)資料庫作為本研究的研究對象，該資料庫為醫院收集許多受試者的各項身體特徵與疾病狀況，彙整後提供給大眾使用。本研究方法首先將資料庫進行預處理，將資料進行標準化，使數據介於-1~1。利用FNN對每筆數據的特徵進行學習，並利用粒子群演算法(Particle Swarm Optimization, PSO)和引力搜尋演算法(Gravitational Search Algorithm, GSA)兩種基於觀察大自然現象所啟發的演算法，對FNN分類器的權重與偏差進行優化，來找到最佳組合。在演算法結合方面，參考González et al.所提出的FuzzyGSA，利用模糊規則對GSA演算法的參數進行優化，以提高演算法在分類器的性能，與Mirjalili et al.所提出的PSOGSA，將PSO中的社會思維能力(Gbest)與GSA的局部搜索能力相結合。本研究將模糊規則於優化PSOGSA演算法之參數，提出FuzzyPSOGSA，來加速演算法在後期收斂的速度。在結果的部分則探討各演算法於優化分類器所產生的學習誤差，並利用混淆矩陣來對所提出的方法進行評估。
本研究所提出的方法可以提供醫師在診斷慢性腎臟疾病與間皮瘤疾病病患時，能有一個更準確的支持決策系統，來幫助醫師更準確的判定病患是否患有疾病，以減少醫師於臨床診斷時，判別錯誤造成的延誤就醫和醫療資源之浪費。

The Feed-forward Neural Network (FNN) is a kind of artificial neural network and has been widely used in medical diagnosis, data mining, securities market analysis and other fields. In FNN, during the learning process, the goal is to find the best combination of connection weights and biases in order to achieve the minimum error. However, in many cases, FNNs converge to local optimum but not global optimum. How to optimize the connection weights and deviations to achieve the minimum error is one of the purposes of this study.
This study is to use the University of California Irvine (UCI) mechanical learning database of chronic kidney disease (CKD) and mesothelioma (MES) disease as the research object of this study. The research method firstly preprocesses the database and normalizes the data so that the data is between -1 and 1. Using FNN to learn the feature　of each data, and using particle swarm optimization (PSO) and gravitational search algorithm (GSA) to optimize the weights and biases of FNN classifiers based on the algorithms inspired by observation of natural phenomena. In addition, referring to the FuzzyGSA proposed by González et al., the fuzzy rules is used to optimize the parameters of the GSA algorithm to improve the performance of the algorithm in the classifier. PSOGSA, proposed by Mirjalili et al., combines the social thinking ability (Gbest) in PSO with the local search ability of GSA. In this study, fuzzy rules are used to optimize the parameters of the PSOGSA algorithm, and FuzzyPSOGSA is proposed to accelerate the convergence of the algorithm in the later stage. In the part of the results, the learning errors generated by each algorithm in optimizing the classifier are discussed, and the proposed method is evaluated by using the confusion matrix.
Our proposed method can provide doctors with a more accurate support decision-making system in the diagnosis of patients with CKD and MES cutaneous disease, to help doctors more accurately determine whether patients have diseases. In order to reduce the delay caused by mistakes in medical diagnosis and medical resources, doctors can reduce the waste of medical treatment and medical resources.

摘要 I
Abstract II
目 錄 IV
圖目錄 VI
表目錄 VII
第一章 緒論 1
1.1研究動機 1
1.2研究目的 3
第二章 文獻探討 5
2.1啟發式演算法 5
2.1.1粒子群演算法 5
2.1.2粒子群演算法之相關研究文獻 7
2.1.3引力搜尋演算法 8
2.1.4引力搜尋演算法之相關研究文獻 11
2.2模糊邏輯 13
2.3人工神經網路 14
第三章 研究方法 16
3.1資料收集 17
3.1.1慢性腎臟疾病資料庫 17
3.1.2間皮瘤疾病資料庫 18
3.2資料預處理 19
3.3模型建立 20
3.3.1適應函數 20
3.3.2權重、偏差矩陣 21
3.3.3前饋式類神經結合粒子群演算法 22
3.3.4前饋式類神經結合引力搜尋演算法 24
3.3.5前饋式類神經結合模糊引力搜尋演算法 26
3.3.6前饋式類神經結合粒子群與引力搜尋演算法 27
3.3.7前饋式類神經結合粒子群與模糊引力搜尋演算法 29
第四章 研究結果 31
4.1演算法誤差 31
4.2性能指標 31
4.3各演算法優化類神經結果 33
4.4討論相關工作 33
第五章 結論與建議 37
5.1結論 37
5.2建議 38
參考文獻 39

[1] W. Gulbinat, What is the role of WHO as an intergovernmental organisation, https://www.hon.ch/en/, 1997.
[2] G. A. Cuckler, A. M. Sisko, J. A. Poisal, S. P. Keehan, S. D. Smith, A. J. Madison, C. J. Wolfe, J. C. Hardesty, National Health Expenditure Projections, 2017-26: Despite Uncertainty, Fundamentals Primarily Drive Spending Growth, Health Aff., 37 (2018) 482-492.
[3] Irie, Miyake, Capabilities of three-layered perceptrons, IEEE 1988 International Conference on Neural Networks, 1988, pp. 641-648 vol.641.
[4] N. A. Mat Isa, W. M. F. W. Mamat, Clustered-Hybrid Multilayer Perceptron network for pattern recognition application, Applied Soft Computing, 11 (2011) 1457-1466.
[5] C. J. Lin, C. H. Chen, C. Y. Lee, A self-adaptive quantum radial basis function network for classification applications, 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541), 2004, pp. 3263-3268 vol.3264.
[6] B. Malakooti, Y. Zhou, Approximating polynomial functions by Feedforward Artificial Neural Networks: Capacity analysis and design, Appl. Math. Comput., 90 (1998) 27-51.
[7] K. Hornik, M. Stinchcombe, H. White, Multilayer feedforward networks are universal approximators, Neural Networks, 2 (1989) 359-366.
[8] H. Adeli, S. L. Hung, An adaptive conjugate gradient learning algorithm for efficient training of neural networks, Appl. Math. Comput., 62 (1994) 81-102.
[9] N. Zhang, An online gradient method with momentum for two-layer feedforward neural networks, Appl. Math. Comput., 212 (2009) 488-498.
[10] M. T. Hagan, M. B. Menhaj, Training feedforward networks with the Marquardt algorithm, IEEE Transactions on Neural Networks, 5 (1994) 989-993.
[11] J. R. Zhang, J. Zhang, T. M. Lok, M. R. Lyu, A hybrid particle swarm optimization–back-propagation algorithm for feedforward neural network training, Appl. Math. Comput., 185 (2007) 1026-1037.
[12] M. Gori, A. Tesi, On the problem of local minima in backpropagation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 14 (1992) 76-86.
[13] V. K. Bohat, K. V. Arya, An effective gbest-guided gravitational search algorithm for real-parameter optimization and its application in training of feedforward neural networks, Knowledge-Based Systems, 143 (2018) 192-207.
[14] S. Mirjalili, S. Z. Mohd Hashim, H. Moradian Sardroudi, Training feedforward neural networks using hybrid particle swarm optimization and gravitational search algorithm, Appl. Math. Comput., 218 (2012) 11125-11137.
[15] D. J. Armaghani, E. T. Mohamad, M. S. Narayanasamy, N. Narita, S. Yagiz, Development of hybrid intelligent models for predicting TBM penetration rate in hard rock condition, Tunnelling and Underground Space Technology, 63 (2017) 29-43.
[16] J. L. Salmeron, S. A. Rahimi, A. M. Navali, A. Sadeghpour, Medical diagnosis of Rheumatoid Arthritis using data driven PSO–FCM with scarce datasets, Neurocomputing, 232 (2017) 104-112.
[17] H. Melo, J. Watada, Gaussian-PSO with fuzzy reasoning based on structural learning for training a Neural Network, Neurocomputing, 172 (2016) 405-412.
[18] B. González, F. Valdez, P. Melin, G. Prado-Arechiga, Fuzzy logic in the gravitational search algorithm for the optimization of modular neural networks in pattern recognition, Expert Syst. Appl., 42 (2015) 5839-5847.
[19] F. Olivas, F. Valdez, P. Melin, A. Sombra, O. Castillo, Interval type-2 fuzzy logic for dynamic parameter adaptation in a modified gravitational search algorithm, Inf. Sci., 476 (2019) 159-175.
[20] F. Valdez, J. C. Vazquez, P. Melin, O. Castillo, Comparative study of the use of fuzzy logic in improving particle swarm optimization variants for mathematical functions using co-evolution, Applied Soft Computing, 52 (2017) 1070-1083.
[21] S. J. Russell, P. Norvig, Artificial intelligence: a modern approach, Malaysia; Pearson Education Limited2016.
[22] J. Kennedy, R. Eberhart, Particle swarm optimization, Proceedings of ICNN'95 - International Conference on Neural Networks, 1995, pp. 1942-1948 vol.1944.
[23] Z. Zhan, J. Zhang, Y. Li, H. S. Chung, Adaptive Particle Swarm Optimization, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 39 (2009) 1362-1381.
[24] S. Gambhir, S. K. Malik, Y. Kumar, PSO-ANN based diagnostic model for the early detection of dengue disease, New Horizons in Translational Medicine, 4 (2017) 1-8.
[25] J. Cervantes, F. Garcia-Lamont, L. Rodriguez, A. López, J. R. Castilla, A. Trueba, PSO-based method for SVM classification on skewed data sets, Neurocomputing, 228 (2017) 187-197.
[26] A. Subasi, Classification of EMG signals using PSO optimized SVM for diagnosis of neuromuscular disorders, Comput Biol Med, 43 (2013) 576-586.
[27] E. Rashedi, H. Nezamabadi-pour, S. Saryazdi, GSA: A Gravitational Search Algorithm, Inf. Sci., 179 (2009) 2232-2248.
[28] M. A. Behrang, E. Assareh, M. Ghalambaz, M. R. Assari, A. R. Noghrehabadi, Forecasting future oil demand in Iran using GSA (Gravitational Search Algorithm), Energy, 36 (2011) 5649-5654.
[29] Y. Jamshidi, V. G. Kaburlasos, gsaINknn: A GSA optimized, lattice computing knn classifier, Engineering Applications of Artificial Intelligence, 35 (2014) 277-285.
[30] L. A. Zadeh, Fuzzy sets, Information and Control, 8 (1965) 338-353.
[31] C. M. Bishop, Neural networks for pattern recognition, Oxford university press1995.
[32] S. Sinaie, Solving shortest path problem using gravitational search algorithm and neural networks, (2010).
[33] L. J. Rubini, P. Eswaran, Generating comparative analysis of early stage prediction of Chronic Kidney Disease, International Journal of Modern Engineering Research (IJMER), 5 (2015) 49-55.
[34] K. Murat, T. Kemal, Classification of chronic kidney disease with most known data mining methods, International Journal of Advances in Science Engineering and Technology, 5 (2017).
[35] S. B. Akben, Early Stage Chronic Kidney Disease Diagnosis by Applying Data Mining Methods to Urinalysis, Blood Analysis and Disease History, IRBM, 39 (2018) 353-358.
[36] O. Er, A. C. Tanrikulu, A. Abakay, F. Temurtas, An approach based on probabilistic neural network for diagnosis of Mesothelioma’s disease, Computers & Electrical Engineering, 38 (2012) 75-81.
[37] M. Nilashi, O. bin Ibrahim, H. Ahmadi, L. Shahmoradi, An analytical method for diseases prediction using machine learning techniques, Computers & Chemical Engineering, 106 (2017) 212-223.
[38] S. Mukherjee, Malignant Mesothelioma Disease Diagnosis using Data Mining Techniques, Applied Artificial Intelligence, 32 (2018) 293-308.