營建工程管理決策分析過程中，常使用資料探勘(Data mining)技術建立模擬模式，以提供決策者評估之參考資訊。但以傳統方法論建立之模擬模式輸出結果正確率不佳，本文採用類神經網路等作為建立模式工具，以解決建立模擬模式資料之品質不佳或問題複雜關係導致分析結果不精確之問題。
本文研究標的為建構營建管理績效評估模式。該模式為解決缺乏監督式學習網路所需案例輸出值問題，以聚類分析進行資料預處理，然後建構分類分析模式而發展之多階段類神經網路模式，該模式使用類神經網路為組合元素。因其原始網路(自組織映射網路Self-Organizing Feature Mapping, SOM,倒傳遞網路Back Propagation, BP)並非針對此類問題發展；故本文對於建構分析模式採用類神經網路有提昇輸出結果品質需求。為提昇多階段類神經網路模式輸出資訊之參考性，本文進行相關研究，以期能提供營建管理有效決策資訊。
因SOM網路對於聚類圖形分辨能力有待提昇，故本文發展多聚類中心競爭理論之改良權重調整機制，形成多聚類中心SOM網路(Multi-stage Clustering centers SOM, MCC-SOM)，可加強聚類分析能力。且藉由AEI值與Davies-Bouldin index數值評估MCC-SOM網路對傳統SOM網路之聚類資訊保存率、聚類歸屬明確化改善程度。
BP網路其輸出推論值正確率有待提升，本文結合互動式調整學習係數機制，與依互動式調整學習係數需求發展之專屬更新頻率，提出互動式學習BP網路(Individual Inference Adjusting Learning Rate BP Network, IIALR)，以有效提昇BP網路之分類推論值正確率。
本文為印證MCC-SOM與IIALR網路之改善幅度，發展其模擬程式，並以臺灣地區甲級營造廠之財務比率為案例進行分析。本文經此案例印證結果可驗證改良式多階段類神經網路模式應用於營建管理績效評估模式分析能力之優越性。

The data mining has been used to construct the simulating model in the decision making process of the construction management. The result of the simulating model can offer as the referring information for the decision making. The training data of model constructing is hard to obtain by experiment or observation. Moreover, the case background of the construction engineering is also hard to keep consisting. Thus, this paper adopts the neural network as the constructing tool of the simulating model. It solves the problems of the error result from the evaluating model by the poor quality of training data.
This paper proposes an evaluating model for the decision making construction management. The model overcomes the problem of supervised learning analysis model that is short of the desired output of training data. The desired output is undefined in the practice. Therefore, the data preprocess uses the cluster analysis. This paper adopts the Self Organizing Feature Mapping network (SOM) as the original tool for the clustering analysis. Moreover, the SOM is modified by the clustering center electing to promote the topology preserving and the cluster diagram clarifying of the SOM. Those results help to make the clustering analysis more precise.
Besides, this paper proposes the Back Propagation network (BP) as the original tool of the classified analysis. The BP uses the result of data preprocess as the desire output of training data. Due to the structure of the BP is flexible, the fixed learning rate makes the correctness percent of estimated output of the BP poor. Thus, this paper develops the Individual Inference Adjusting Learning Rate BP network (IIALR) to search the best weight combination in the error space. The best weight combination enhances the correctness percent of estimated output of the BP.
The above two mentioned modified neural networks are combined as the multi-stage neural network model (MSOM-IIALR). Through the case study of financial rating of the large-scale contractor in Taiwan, this paper verifies the model to provide more effective referring information for the decision making of the construction management.

參考文獻
[1]. Wang, Ching-Hwang , Song, Yi-Zhe and Kao, Chih-Han,” A model of financial credit rating for construction contractor”, Journal of the Chinese Institute of Civil and Hydraulic Engineering, vol.15, No.1 , pp.25-35, （2003）.
[2]. Cheng, Min-Yuan and Ko Chien-Ho, "Object-Oriented Evolutionary Fuzzy Neural Inference System for Construction Management", Journal of Construction Engineering and Management, ASCE, Vol. 129, No. 4, pp.461-469, (2003).
[3]. Yu, W. D., and Skibniewski, M. J., “A neuro-fuzzy computational approach to constructability knowledge acquisition for construction technology evaluation”, Automation in Construction, Vol. 8, No. 5, pp. 539-552, (1999).
[4]. Leu, Sou-Sen and Cheng, Fei-Wen, "Performance evaluation of parallel genetic algorithm on resource-constrained construction scheduling," submitted to Journal of Construction Management and Economics, (2002).
[5]. Ritter, H., Schulten K., On the stationary state of Kohonen’s self-organizing sensory mapping, Biol. Cybern., Vol.54, pp.99-106, (1986).
[6]. Vesanto, Juha and Alhoniemi, Esa, “Cluster of the Self-Organizing Map”, IEEE Transactions on Neural Networks, Vol.11, No.3, pp.586-600, (2000).
[7]. Kohonen, T., “Self-Organized Formation and Topologically Correct Feature Maps”, Biological Cybernetics, Vol.43, pp.59-69, (1982).
[8]. Baraldi, Andrea and Blonda, Palma, ”A Survey of Fuzzy Cluster Algorithms for Pattern Recognition-Part ІⅡ”, IEEE Transactions on Systems, Man, and Cybernetics, Vol.29, No6, pp.778-801, (1999).
[9]. Villmann, Thomas, Der Ralf, Michael Herrmann and Martinetz, M. Thomas, ”Topology Preservation in Self-Organizing Feature Maps: Exact Definition and Mearsurement”, IEEE Transaction on Neural Networks, Vol.8, No.2, pp.256-266, (1997).
[10]. Lo, Zhen-Ping and Bavarian, B., “Improved Rate of Convergence in Kohonen Neural Networks”, IEEE, ICANN, Vol.Ⅱ, pp.201-206, (1991).
[11]. Lo, Zhen-Ping and Bavarian, “Analysis of Neighborhood Interaction in Kohonen Neural Networks”, IEEE, ICANN, Vol.Ⅱ, pp.246-251, (1991).
[12]. Lo, Zhen-Ping and Bavarian, “On the rate of convergence in topology prevserving neural networks”, Biol. Cybernet, Vol.65, pp.55-63, (1991).
[13]. Erdogan, S., Ng, G. S. and K.H.C. Patrick, “Measurement criteria for neural network pruning, Digital Signal Processing Applications”, Proceedings of the IEEE TENCON, Vol.1, pp.83-89, (1996).
[14]. Giudici, Paolo, “Applied data mining：statistical methods for business and industry”, Wiley, Chichester, (2002).
[15]. Berkhin, P., “Survey of Clustering Data Mining Techniques”, Accrue Software Inc., (2002).
[16]. Haykin, Simon, “Neural Networks : A Comprehensive Foundation, Prentice Hall”, Upper saddle river, (1999).
[17]. Barnard, E., “A comparative study of optimization techniques for backpropagation”, Neurocomputing, Vol.6, pp.19-30,（1994）.
[18]. Barnard, E., “Optimization for Training Neural Nets”, Transactions on Neural Networks, IEEE, Vol.3, No.2, pp.232-240,（1992）.
[19]. Carpenter, G., Grossberg, S. and Rosen, D. B., “Fuzzy ART: Fast stable learning and categorization of analog patterns by an adaptive resonance system”, Neural Networks, Vol.4, pp.759–771, (1991).
[20]. Carpenter, G., Grossberg, S., Maukuzon, N., Reynolds, J. and Rosen, D. B., “Fuzzy ARTMAP: A neural network architecture for incremental supervised learning of analog multidimensional maps,” IEEE Trans. Neural Networks, Vol. 3, No. 5, pp. 698–713, (1992).
[21]. Jeong, Jeong-Won, Shin, C. D., S. and Marmarelis, V. Z. Do, “Segmentation methodology for automated classification and differentiation of soft tissues in multiband images of high-resolution ultrasonic transmission tomography”, Medical Imaging, IEEE Transactions, Vol.25, No.8, pp.1068 - 1078, (2006)
[22]. FRITZKE, Bernd, “Let it grow- Self-organizing feature maps with problem dependent cell structure”, Artificial Neural Networks, pp.403-408, (1991).
[23]. FRITZKE, Bernd, ”Unsupervised Clustering with Growing Cell Structures, IEEE, ICANN, Vol.Ⅱ, pp.531-536, (1991).
[24]. FRITZKE, Bernd, “Growing grid- A self-organizing network with constant neighborhood range and adaptation strength”, Neural Processing Lett., (1995).
[25]. Voegtlin, Thomas, “Recursive self-organizing maps”, Neural Networks, Vol.15, pp.979–991, (2002).
[26]. Sun, Yi, “On quantization error of self-organizing map network”, Neurocomputing, Vol.34, pp.189-193, (2000).
[27]. Liebert, W., Pawelzik, K. and Schuster, H.G.., “Optimal embeddings of chaotic attractors from topological considerations”, Europhys. Lett., Vol.14, pp521-526, (1991).
[28]. Bauer, Hans-Ulrich and Villmann, Thomas, “Growing a Hypercubical Output Space in a Self-Organizing Fearture Map”, IEEE Transaction on Neural Networks, Vol.8, No.2, (1997).
[29]. Bauer, Hans-Ulrich and Pawelzik, Klaus R., “Quantifying the Negihborhood Preservation of Self-Organizing Feature Maps”, IEEE Transaction on Neural Networks, Vol.3, No.4, pp.570-579, (1992).
[30]. Dittenbach, Michael, Merkl, Dieter and Rauber, Andreas, “The Growing Hierarchical Self-Organizing Map”, IEEE, ICANN, pp.15-19, (2000).
[31]. Si, J., Lin, S. and Vuong, M.-A., “Dynamic topology representing networks”, Neural Networks, Vol.13, pp. 617-627, (2000).
[32]. Cottrell, M. and Fort. J.C., “A stochastic model of retinotopy: a self-organizing process”, Biol. Cybern., Vol.53, pp.405-411, (1986).
[33]. Mitra S. and Pal S. K., “Self-organizing neural network as a fuzzy classifier", IEEE Trans., Systems Man Cybernetics, 24, 3, pp.385-399, (1994).
[34]. Graepel, Thore, Obermayer, Klaus and Burger, Matthias, “Self Organizing Maps:Generalizations and New Techniques”, Neurocomputing, pp.12-44, (1998).
[35]. Cervera, E. and Del Pobil, A.P., “Multiple self-organizing maps: A hybrid learning scheme”, Neurocomputing, Vol.16, pp.309-318, (1997).
[36]. Merkl D. and Rauber A., “Uncovering the hierarchical structure of text archives by using an unsupervised neural network with adaptive architecture”, PADKK, LNAI1805, pp.384-395, (2000).
[37]. Krishnapuram, R. and Keller, J. M., “A possibilistic approach to clustering”, IEEE Tran. Fuzzy System, Vol.1, No.2, pp.98-110, (1993).
[38]. Dave, R. N. and Krishnapuram, R., “Robust clustering method: unified view”, IEEE Tran. Fuzzy Syst., Vol.5, No.2, pp.270-293, (1997).
[39]. Alahakoon, D., Halagmuge, S. K. and Srinivasan, B., “Data Mining with Self Generating Neuro-Fuzzy Classifiers”, IEEE, IFSCP, Vol.Ⅱ, pp.1096-1101, (1999).
[40]. Sohn, Sunghwan and Dagli, Cihan H., “Advantages of Using Fuzzy Class Memberships in Self-Organizing Map and Support Vector Machines”, IEEE, ICANN, pp.1886-1890, (2001).
[41]. Martinetz, T.M., Berkovich, S.G. and Schulten, K., ”Neural-gas network for vector quantization and its application to time-series prediction”, IEEE Trans. Neural Networks, Vol.4, No.4, pp.558-569, (1993).
[42]. Martinetz, T.M.. and Schulten, K., “Topology representing networks”, Neural Networks, 7, pp.507-522, (1994).
[43]. Liu, Yong, Zhao, Bin, Xia, Shaowei and Zhao, Ming-sheng, “A Self-Organizing Network with Fuzzy Hyperellipsoidal Classifying and It Application in Handwritten Numeral Recognition”, IEEE, pp.2859-2862, (1999).
[44]. Kim, Kyung-Joong and Cho, Sung-Bae, “Fusion of Structure Adaptive Self-Organizing Maps Using Fuzzy Integral”, IEEE, ICANN, pp.28-33, (2003).
[45]. Kuo, R. J., Chi, S. C. and Den, B.W., “A Fuzzy Kohonen’s Feature Map Neural Network with Application to Group Technology”, IEEE, pp.3098-3101, (1999).
[46]. DeSieno, D., “Adding a conscience to competitive learning”, IEEE Proceedings of the International Conference on Neural Networks, Vol.I, pp.117-124, (1988).
[47]. Bauer, H.-U., Herrmann, M. and Villmann, T., “Neural maps and topographic vector quantization”, Neural Networks, Vol.12, pp.659-676, (1999).
[48]. Dimitriadou, Evgenia, Dolnicar, Sara and Weingessel, Andreas, “An examination of indexes for determining the number of clusters in binary data sets”, Psychometrika, Vol.67, No1, pp.137-160, (2002).
[49]. Davies, D. L. and Bouldin, D. W., “A cluster separation measure”, IEEE Trans. Patt. Anal. Machine Intell., PAMI-1, pp.224-227, (1979).
[50]. Rumelhart, D., “Parallel Distributed Processing”, Cambridge, MA,(1974).
[51]. Maa, Chia-Yiu, “A Two-Phase Optimization Neural Network”, IEEE Transactions on Neural Networks, Vol.3, No.6, pp.1003-1009, (1992).
[52]. Drucker, Harris, “Improving Generalization Performance Using Double Backpropagation”, IEEE Transactions on Neural Networks, Vol.3, No.6, pp.991-997, (1992).
[53]. Jacobs, Robert A., “Increased Rates of Convergence Through Learning Rate Adaptation”, Neural Networks, Vol.1, pp.295-307, (1998).
[54]. Yamada, K., Kami, H. and Tsukumo, J., “Handwritten numeral recongnition by multi-layered neural network with improved learning algorithm”, IJCNN-89, pp.259-266,（1989）.
[55]. Gian, Paolo Drago and Sandoro, Ridella, “Statistically Controlled Activation Weight Initialization”, Transactions on Neural Networks IEEE, Vol.3, No.4, pp.627-631, (1992).
[56]. Qin, Si-Zhao, “Comparison of Four Neural Net Learning Methods for Dynamic System Identification”, IEEE Transactions on Neural Networks, Vol. 3, No. 1, pp.122-130, (1992).
[57]. Wilson, D. Randall and Martinez, Tony R., ”The general inefficiency of batch training for gradient descent learning”, Neural Networks, Vol.16, pp.1429-1451, (2003).
[58]. Kumar, Ashwani, Agrawal, D. P. and Joshi, S. D., “Multiscale rough set data analysis with application to stock performance modeling”, Intelligent Data Analysis, Vol.8, pp.197-209, (2004).
[59]. W. K. Harold (editor), Classics in Game theory, Princeton, NJ: Princeton University Press, (1997), ISBN 0691011931.
[60]. Wang, Ching- Hwang, Kao, Chih-Han and Lee, Wei-Hsien, „A New Interactive Model for Improving the Learning Performance of BPNN, Automation in Construction”, Vol.16 , pp.745-758, (2007).
[61]. De Angulo, Vicente Ruiz and Torras, Carme, “A deterministic algorithm that emulates learning with random weights”, Neurocomputing, Vol.48, pp.975-1002, (2002).
[62]. Wang, Ching- Hwang, Kao, Chih-Han and Tsai, Chia-Chang, „Improving Network Learning Performance using an Interactive Adjusting Algorithm“, WSEAS Transactions of Computers, Vol.4, No.6, pp.715, (2007) .
[63]. Wu, T. K., “Performance evaluation and prediction model for construction subcontractor”, MS these of National Taiwan University of Science and Technology, Taipei, Taiwan, (2001).