本研究使用支援向量迴歸(Support Vector Regression, SVR)來建立產品外型的感性預測系統。研究以須快速設計的手機產品為例，以A形態-物理量值、既定規格參數資訊，B形態-細部設計、特徵區塊參數資訊，C形態-外觀輪廓比例參數資訊。此三種造形參數化形態來拆解產品外型；再者，以五組感性語彙對來表示消費者對產品的偏好程度；最後，用支援向量迴歸來建立產品外型參數與感性語彙之間的非線性預測模式。此研究同步以倒傳遞類神經網路法進行，作為比對與驗証，支撐向量迴歸預測的結果都將與過去常用的倒傳遞類神經網路進行比較，檢視支撐向量迴歸對於工業設計領域之成效。
由預測模式的訓練時間比較分析、預測時間比較分析、預測結果錯誤率比較分析、訓練運算時間之t-test檢定與預測錯誤率之t-test檢定等五個方向分析結果顯示：1. 在手機造形設計的A、B、C各種參數化形態資訊組合的預測表現上，SVR與BPN預測模式皆以A+B+C全部組合一起，最多輸入欄位的ABC組合之平均錯誤率最低。2. 在兩種預測模式的預測效果表現上，BPN模式平均預測錯誤率與SVR模式之比約為1.26:1，但在挑出SVR與BPN分別的最佳預測模式時，預測錯誤率之比約為1.1:1，十分相近。3. 在兩種預測模式的時間效率表現上，SVR模式的訓練運算速度總平均與BPN模式之比約19.4:1。將兩種模式平均運算速度(總花費時間/欄位數)列表分析結果中發現，依進行運算的欄數持續增多，SVR預測模式和BPN預測模式的訓練速度比成正成長，顯示差距將更為加大。
綜合以上SVR、BPN兩種預測模式，其資訊形態、效果、效率三方面的結論，本研究推得「時間效率」之項目為SVR導入設計預測模式所得到的最大優勢。

In this paper, a state-of-the-art machine learning approach known as “Support Vector Regression (SVR)”, has been introduced to develop a cell phone product image prediction model. The form features of product samples are first examined in three styles: A. quantification and specification data, B. detail feature label data, C. contour and scale data. Adjectives are used to describe the product images of product samples and five linguistic labels are used to linguistically evaluate the ratings toward product samples. Finally, the prediction model is constructed using support vector regression by training a series of product samples consisting of product form features along with the product linguistic label. A common used method known as “Back-Popagation Network (BPN)”,is also introduced to develop the same prediction model simultaneously. Two prediction models have compared in effectiveness and efficiency to verify the advantage of using the SVR method.
Five prediction model statistics has been analyzed: training time, prediction time, prediction root mean square error (RMSE), test time t-test and prediction accuracy t-test. The results show that: 1. both SVR and BPN prediction models present the best results using the ABC combine style. 2. in effectiveness, the average RMSE between BPN to SVR is 1.26: 1, but for the best model the average RMSE between BPN to SVR is 1.1: 1, the difference is very small. 3. in efficiency, the average training time between SVR to BPN is 19.4: 1, and the difference in training computing speed will be enlarged when more product form feature inputs from SVR and BPN prediction models.
Conclude the SVR and BPN models’ performance in three above-mentioned parts: product form styles, effectiveness and efficiency, authors find out the “efficiency in computing time” is the principal advantage using support vector regression method in cell phone product image prediction model.

英文部分
[1] Aizerman, M. A., Braverman, E. M., and Rozonoer, L. I., “Theoretical Foundations of the Potential Function Method in Pattern Recognition Learning,” Autom. Remote Control, vol. 25, pp.821–837, 1964.
[2] Brown, M., Grundy, W., Lin, D., Cristianini, N., Sugnet, C., Furey, T., Ares, M., and Haussler, D., “Knowledge-base analysis of microarray gene expression data by using support vector machines,” Proceedings of the National Academy of Sciences, Vol.97, pp.262–267, 1999.
[3] Burr, D. J., “Experiments on neural net recognition of spoken and written text,” IEEE Trans. On ASSP, Vol.36, No.7, pp.1162–1168, 1988.
[4] Carpenter, G. A., “Neural network models for pattern recognition and associative memory,” Neural Network, Vol.2, pp. 243–257, 1989.
[5] Cao, L., ”Support vector machines experts for time series forecasting,” Neurocomputing, Vol. 51, pp. 321-339, 2003.
[6] Chang, Guanghsu A., “A Neural Network Model for The Handling Time of Design for Assembly,” Journal of the Chinese Institute of Industrial Engineers, Vol. 9, No. 1, pp. 35-48, 2001.
[7] Chen, Kuan-Yu, Wang, Cheng-Hua, “Support vector regression with genetic algorithms in forecasting tourism demand,” Tourism Management, Vol.28, pp.215–226, 2005.
[8] Desai, Kiran, Badhe, Yogesh, Tambe, Sanjeev S., Kulkarni, Bhaskar D., “Soft-sensor development for fed-batch bioreactors using support vector regression,” Biochemical Engineering Journal, Vol.27, pp.225–239, 2005.
[9] Gustavo, Camps-Valls, Luis, Gómez-Chova, Jordi, Muñoz-Marí, Joan, Vila-Francés, Julia, Amorós-López, Javier, Calpe-Maravilla, “Styling and design: intuition and analysis in industrial design,” Remote Sensing of Environment, Vol.36, pp.23–33, 2006.
[10] Hsiao, Shih-Wen, and Huang, H. C., “A Neural Network Based Approach for Product Form Design,” Design Studies, Vol. 23, pp. 67-84, 2002.
[11] Ince, Huseyin, Trafalis, Theodore B., “A hybrid model for exchange rate prediction,” Decision Support Systems, Vol.42, pp.1054–1062, 2005.
[12] Joachims, T., “Text categorization with support vector machines : Learning with Many Relevant Features,” Proceedings of European Conference on Machine Learning (ECML), pp.137–142, 1998.
[13] Karras, D. A. and Mertzios, B. G., “Time series modeling of endocardial border motion in ultrasonic images comparing support vector machimes multilayer perceptrons and linear estimation technique, ” Measurement, Vol. 36, pp.331-345, 2004.
[14] Kim, K. J., “Financial time series forecasting using support vector machines, ” Neurocomputing, Vol. 55, pp. 307-319, 2003.
[15] Liu, Xu, Lu, Wencong, Jin, Shengli, Li, Yawei, Chen, Nianyi, “Support vector regression applied to materials optimization of sialon ceramics,” Chemometrics and Intelligent Laboratory Systems, Vol.82, pp.8–14, 2006.
[16] Mizuno, H, “Application of neural network to technical analysis of stock market prediction,” Studies in Informatic and Control, Vol.7, pp.111-120, 1998.
[17] Nagamachi, Mituo., “Special Issue-Kansei Engineering and Comfort,” Industrial Ergonomics, Vol. 48, No.2, pp.426–444, 1997.
[18] Oliveira, Adriano L.I., “Estimation of software project effort with support vector regression,” Neurocomputing, Vol.69, pp.1749–1753, 2006.
[19] Paul, K., Phua, H., “Neural network with genetically evolved algorithms for stocks prediction,” Asia-pacific Journal of Operational Research, Vol. 18, pp. 103-107, 2001.
[20] Pontil, M. and Verri, A., “Object recognition with support vector machines,” IEEE Trans. On PAMI, Vol.20, pp.637-646, 1998.
[21] Tay, F. E. H., and Cao, L., “Application of support vector machines in financial time series forecasting,” Omega, Vol. 29, pp. 309-317, 2001.
[22] Tay, F. E. H., and Cao, L. J., “Modified support vector machines in financial time series forecasting,” Neurocomputing, Vol. 48, pp. 847-861, 2002.
[23] Thissen, U., Brakel, R., Weijer, A. P., Melssen, W. J., and Buydens, L. M. C., “Using support vector machine for time series prediction,” Chemometrics amd Intelligent Laboratory System, Vol. 69, pp. 35-49, 2003.
[24] Tovey, M., “Styling and design: intuition and analysis in industrial design,” Design Studio, Vol. 18, No. 1, pp. 5-32, January, 1997.
[25] Vapnik, V. N., Lisboa, P. J. G. and Vaughan, J., “Neural Networks in Business: A Survey of Applications(1992-1998),” Exper Systems with Applications, Vol. 17, pp.51-70, 1999.
[26] Whiteson, D.O. and Naumann, N.A., “Support vector regression as a signal discriminator in high energy physics,” Neurocomputing, Vol.55, pp.251–264, 2003.
[27] Yang, Shansheng, Lu, Wencong, Chen, Nianyi, Hu, Qiannan, “Support vector regression based QSPR for the prediction of some physicochemical properties of alkyl benzenes,” Journal of Molecular Structure, Vol.719, pp.119–127, 2004.
[28] Zhang, G., Patuwo, B. E., and Hu, M.Y., “Forecasting with Artificial Neural Networks：The State of the Art,” International Journal of Forecasting, Vol. 14, pp.35–62, 1998.
[29] Zimmermann, J., “Statistical learning methods: Basics, control and performance,” Nuclear Instruments and Methods in Physics Research, Vol.559A, pp.106–114, 2005.
中文部分
[30] 林忠志，“應用特徵導向與類神經網路於產品造形衍生之研究”，國立成功大學工業設計研究所碩士論文，2003。
[31] 林彥呈，“應用非線性推論模式於產品色彩、造形與意象關係之研究”，國立成功大學工業設計研究所碩士論文，2001。
[32] 施韋名，“以類神經網路探討眼鏡造形與感覺意象之對應關係” 工業設計, 第26卷, 第3期, pp.160-164，1997。
[33] 馬居正，“應用類神經網路於汽車造形特徵輔助設計之研究”，國立成功大學工業設計研究所碩士論文，2006。
[34] 莊明振、馬永川，“產品意象語彙與造形呈現對應關係之研究”，第三屆設計學會學術研究成果研討會論文集，pp.113-118，1998。
[35] 莊盈祺，“複合式感性意象下產品造形的建構”，國立成功大學工業設計研究所碩士論文，2002。
[36] 陳青海，“功能導向的產品造形衍生與整合模式研究”，國立成功大學工業設計研究所碩士論文，1996。
[37] 管倖生、林彥呈，“應用類神經網路於手機色彩與造形搭配之研究”，工業工程學刊，第18卷，第6期，pp.84-94，2001。
[38] 黃敏菁，“支撐向量機在財務時間序列預測之應用”，輔仁大學金融研究所碩士論文，2005。
[39] 葉怡成，“應用類神經網路”二版，儒林圖書，台北，1999。
[40] 張建成譯，John Chris Jones 著，”設計方法”，六和出版，1992。
[41] 張華城，“應用類神經網路模式於產品造形特徵辨識之研究”，國立成功大學工業設計研究所碩士論文，2000。
[42] 蔡明錡，“電腦輔助產品造形設計模式於網際網路上之應用研究”，國立成功大學工業設計研究所碩士論文，2000。
[43] 蔡宏政，“電腦輔助產品造形、色彩設計與客製化商務系統建構之研究”，國立成功大學工業設計研究所碩士論文，2004。
[44] 鄭宇杰，“電腦輔助設計於產品操作介面開發階段之研究”，國立成功大學工業設計研究所碩士論文，2002。
[45] 劉孟昌，“形態漸變之產品造形衍生與整合模式研究”，國立成功大學工業設計研究所碩士論文，1999。
[46] 薛承甫，“消費性產品涉入程度與造形選擇關係之研究－以行動電話為例”，國立成功大學工業設計研究所碩士論文，2000。
[47] 魏士超，“應用網際網路建立產品造形意象設計系統之研究”，國立成功大學工業設計研究所碩士論文，2001。
[48] 魏雅萍，“設計師與一般消費者對造形認知差異研究”，國立成功大學工業設計研究所碩士論文，2000。