資料探勘的技術，能讓我們從大量資料中，找出隱含或未被發現的模式，進而萃取出有用的知識。分類問題是目前資料探勘常被討論的主題之一，根據已知分類結果與其特徵屬性值之訓練點，利用探勘技術建立資料分類的模型，依此模型判定未知類別之新進資料的所屬分類。
　　支援向量機 ( support vector machine；SVM ) 是一種較新的分類方法，以統計學習理論為基礎，並充分實現風險最小化原則，近年來在處理分類問題上亦被廣泛應用。由於 SVM 是根據訓練實例來建構分類模型，大量歷史數據中，若存在相互矛盾衝突的訓練資料，就會使得 SVM 無法學習到正確的模式，導致分類正確率下降，進而造成分類錯誤成本的增加。因此，如何在資料量與資料複雜度增加的狀況下，避免訓練集衝突狀況的情形發生，使得分類器顯著地提升正確率及效能，是本研究的方向。
　　本研究將單調性的概念應用在改善分類器績效的目標上，對於可能存在衝突資料的訓練集，提出一種基於「避免訓練資料違反單調性」的方法。借助專家智慧經驗，擷取資料集中輸入變數與輸出結果的單調性規則，接著進行資料的前置處理，針對資料集中違反單調性的資料，利用標籤修正 ( relabeling ) 處理程序進行修正，最後 SVM 採用修正後的數據來完成學習任務。經實驗證實，以具有單調性限制的改良資料集進行分類器模型訓練時，確實能有效地增加分類器的預測成效。

　　Data mining techniques enable us to discover hidden patterns and extract valuable knowledge from huge databases. A central problem in data mining is how to assign data examples to one of several predefined classes and infer some key patterns. This kind of problem in data mining is referred to as the classification problem. An algorithm designed to solve this type of problem is consequently called classification algorithm, or classifier. With the help of a classifier, analysts can predict the class of a new instance on the basis of some of its attributes.
　　SVM is the state-of-the-art neural network technology based on statistical learning. By exploiting the mechanics of structural risk minimization, it can handle the overfitting problem quite well. Therefore, SVM have been widely applied to many kinds of fields in the past few years. Since SVM classifier is established by training examples, so if contradictory exists in volumes of historical data then we cannot learn the correct pattern from them. As a result, the accuracy of the classifier will decrease and the costs of misclassification increases. This research is focused on finding the way to enhance the predictive performance and avoid the conflict of training dataset while data was enlarged with volume and complexity.
　　In this study, we offer a new approach which deals with resolving the possible conflict in the training dataset by using the monotonic constraints. Monotonicity of relations between a response variable and predictor variable is a form of prior knowledge that can often be elicited reliably from domain experts. To be aimed with the training instance which violates monotonic constraints, we proceed a relabeling of the data point in order to make the training dataset monotone. We have shown that the use of the new relabeled dataset within monotonic constraints can give substantial improvements in predictive performance over the original dataset via SVM classifier.

[1] 石文俐。中文語音合成之韻律產生器的改良與研究。國立清華大學資訊工程學系碩士論文，2005年。
　
[2] 林昕潔。以SVM 與詮釋資料設計書籍分類系統。國立交通大學資訊科學與工程研究所碩士論文，2005年。
　
[3] Archer, N. P., & Wang, S. Application of the back propagation neural network algorithm with monotonicity constraints for two-group classification problems. Decision Sciences, 24(1), 60-75, 1993.
　
[4] Burbidge, R., Trotter, M., Buxton, B., & Holden, S. Drug design by machine learning: support vector machines for pharmaceutical data analysis. Computers and Chemistry, 26(1), 5-14, 2001.
　
[5] Chandrasekaran, R., Ryu, Y. U., Jacob, V. S., & Hong, S. Isotonic separation. INFORMS Journal on Computing, 17(4), 462-474, 2005.
　
[6] Chang, C. C., & Lin, C. J. LIBSVM: A library for support vector machines, 2001.
Software Available At
http://www.csie.ntu.edu.tw/~cjlin/libsvm
　
[7] Cho, S. B., & Won, H. H. Machine learning in DNA microarray analysis for cancer classification. Proceedings of the 2003 First Asia-Pacific bioinformatics conference on Bioinformatics, 19, 189-198, 2003.
　
[8] Cortes, C., & Vapnik, V. Support-vector networks. Machine Learning, 20, 273-297, 1995.
　
[9] Dembczynski, K., Kotlowski, W., & Slowinski, R. Ensemble of decision rules for ordinal classification with monotonicity constraints. Rough Sets and Knowledge Technology, 5009, 260-267, 2008.
　
[10] Duivesteijn, W., & Feelders, A. Nearest neighbour classification with monotonicity constraints. Proceedings of the 2008 European Conference on Machine Learning and Knowledge Discovery in Databases, 301-316, 2008.
　
[11] Furey, T. S., Cristianini, N., Duffy, N., Bednarski, D. W., Schummer, M., & Haussler, D. Support vector machine classification and validation of cancer tissue samples using microarray expression data. Bioinformatics, 16(10), 906-914, 2000.
　
[12] Hsu, C. W., Chang, C. C., & Lin, C. J. A practical guide to support vector classification, 2003. Available at
http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf
　
[13] Hsu, C. W., & Lin, C. J. A comparison of methods for multiclass support vector machines. IEEE Transactions on Neural Network, 13(2), 415-425, 2002.
　
[14] Lee, Y., & Lee, C. K. Classification of multiple cancer types by multicategory support vector machines using gene expression data. Bioinformatic, 19(9), 1132-1139, 2003.
　
[15] Li, S. T., Shiue, W., & Huang, M. H. The evaluation of consumer loans using support vector machines. Expert Systems with Application, 30(4), 772-782, 2006.
　
[16] Naisbitt, J. Megatrends: Ten new directions transforming our lives. Warner Books, 1988.
　
[17] Pazzani, M. J., Mani, S., & Shankle, W. R. Acceptance of rules generated by machine learning among medical experts. Methods of Information in Medicine, 40, 380-385, 2001.
　
[18] Pendharkar, P. C. A data envelopment analysis-based approach for data preprocessing. IEEE Transactions on Knowledge and Data Engineering, 17(10), 1379-1388, 2005.
　
[19] Pendharkar, P. C., & Rodger, J. A. Technical efficiency-based selection of learning cases to improve forecasting accuracy of neural networks under monotonicity assumption. Decision Support Systems, 36(1), 117-136, 2003.
　
[20] Platt, J. C., Cristianini, N., & Taylor, J. S. Large margin dags for multiclass classification. Advances in Neural Information Processing Systems, 547–553, 2000.
　
[21] Potharst, R., & Feelders, A. J. Classification trees for problems with monotonicity constraints. ACM SIGKDD Explorations Newsletter, 4(1), 1-10, 2002.
　
[22] Quah, J. K. H. The monotonicity of individual and market demand. Econometrica, 68(4), 911-930, 2000.
　
[23] Rademaker, M., Baets, D. B., & Meyer, D. H. On the role of maximal independent sets in cleaning data for supervised ranking. Proceedings of the 2006 IEEE International Conference on Fuzzy Systems, 1619–1624, 2006.
　
[24] Ryu, Y. U., Chandrasekaran, R., & Jacob, V. S. Breast cancer prediction using the isotonic separation technique. European Journal of Operational, 181(2), 842–854, 2007.
　
[25] Tay, F. E. H., & Cao, L. Application of support vector machines in financial time series forecasting. Omega, 29, 309–317, 2001.
　
[26] Vapnik, V. Statistical learning theory. Wiley, 1998.
　
[27] Vapnik, V. The nature of statistical learning theory. Springer, 1995.
　
[28] Wang, S. Adaptive non-parametric efficiency frontier analysis: A neural-network -based model. Computers and Operations Research, 30(2), 279-295, 2003.
　
[29] Wang, S. The unpredictability of standard back propagation neural networks in classification applications. Management Science, 41(3), 555-559, 1995.
　
[30] Witten, I. H., & Frank, E. Data mining: Practical machine learning tools and techniques, second edition. Morgan Kaufmann, 2005.
　
[31] Wu, D., Yang, Z., & Liang, L. Using DEA-neural network approach to evaluate branch efficiency of a large canadian bank. Expert Systems with Applications, 31(1), 108-115, 2006.