本論文針對主成分分析(Principle Components Analysis, PCA)提出一個以Generalized Hebbian Algorithm (GHA)為基礎的高維度平行計算之硬體架構。
我們希望利用硬體的特性來達到平行計算能力，進而加速運算效能，同時希望透過擷取高維度的特徵向量來取得較好的分類成功率，在突觸權重向量更新單元，將原本m筆的資料切割成b等分，重複利用q份硬體電路來運算b次，即m=q×b，m指的是訓練資料的維度，b指的是我們將資料切割成幾等分，q指的是每一等分的資料量，如此一來就可達到硬體共享的機制，也將記憶單元共享給不同的計算元件使用，因此可以降低面積成本(Area Cost)，也能實現較高維度的硬體架構。
我們將硬體電路實作在可程式化系統晶片(System on a Programmable Chip，SOPC)的平台中，並且利用此平台來測試與驗證實驗數據，根據實驗結果來證明我們所提出的硬體架構，是具有較好的分類成功率及較低的硬體資源消耗，也與軟體做時間測量比較，來驗證硬體的加速效能。

[1] Yi-Tsan Hung, “ 以Generalized Hebbian Algorithm為基礎的主成分分析之硬體實現, ” M.S. thesis, Dept. Comput. Sci. and Inform. Eng., Nat. Taiwan Normal Univ., Taipei, Taiwan, 2010.
[2] Cyclone III Device Handbook, 2010, Altera Corporation. http://www.altera.com/literature/hb/cyc3/cyclone3_handbook.pdf
[3] Zeidenberg, Matthew. Neural Networks in Artificial Intelligence. 1990: Ellis Horwood Limited. 1990.
[4] Jolliffe, I. T., “ Principal Component Analysis., ” Springer-Verlag, pp. 487, 1986
[5] 133.Chin-Shu Chang, Teh-Lu Liao, Po-Yun Hsu, Kuo-Kuang Chen, “ Human face recognition system using modified PCA algorithm and ARM platform, ” Computer Communication Control and Automation (3CA), 2010 International Symposium on, On page(s): 294 - 297, Volume: 2 Issue: , 5-7 May 2010
[6] Sattler, M., Sarlette, R., and Klein, R., “ Simple and efficient compression of animation sequences, ” Eurographics/ACM SIGGRAPH Symposium on Computer Animation, 2005.
[7] Soderstrom, U., and Li, H., “ High Definition Wearable Video Communication, ” Lecture Notes in Computer Science, Vol. 5575, pp. 500-512, 2009.
[8] Kim, K., Franz, M.O., and Scholkopf, B., “ Iterative kernel principal component analysis for image modeling, ” IEEE Trans. Pattern Analysis and Machine Intelligence, pp.1351V1366, 2005.
[9] Gottumukkal, R., and Asari, V.K., “ An improved face recognition technique based on modular PCA approach, ” Pattern Recognit. Lett., vol. 25, no. 4, pp. 429–436, Mar. 2004.
[10] Navarrete, P., and Ruiz-del-Solar, J., “ Eigenspace-based recognition of faces: Comparisons and a new approach, ” in Proc. ICIAP, pp. 42–47, 2001.
[11] Perlibakas, V., “ Distance measures for PCA-based face recognition, ” Pattern Recognit. Lett., vol. 25, no. 6, pp. 711-724, Apr. 2004.
[12] Yambor, W.S., Draper, B.A., Beveridge, J.R., “ Analyzing PCA-based Face Recognition Algorithms: Eigenvector Selection and Distance Measures, ” in Christensen, H., Phillips, J. (eds.) Empirical Evaluation Methods in Computer Vision. World Scientific Press, 2002.
[13] Yang, J., Zhang, D., Frangi, A. F., and Yang, J.Y., “ Two-dimensional PCA: A new approach to appearance-based face representation and recognition, ” IEEE Trans. Pattern Anal. Mach. Intell., Vol. 26, no. 1, pp. 131–137, Jan. 2004
[14] Zuo, W., Zhang, D., and Wang, K., “ Bidirectional PCA with assembled matrix distance metric for image recognition, ” IEEE Trans. Systems, Man, and Cybernetics—PART B: Cybernetics, pp.863-872, Vol. 36, 2006.
[15] Partridge, M., and Calvo, R., “ Fast dimensionality reduction and Simple PCA. Intelligent Data Analysis, ” pp. 292–298, 1997.
[16] Sharma, A., and Paliwal, K.K., “ Fast principal component analysis using fixed-point algorithm, ” Pattern Recognition Letters, pp. 1151-1155, 2007.
[17] Storer, M., Roth P.M., Urschler, M., and Bischof, H., “ Fast-Robust PCA, ” Lecture Notes in Computer Science, Vol.5575, pp.430-439, 2009.
[18] Erkki Oja. “ Simplied neuron model as a principal component analyzer. ” Journal of Mathematical Biology, 15 (3): 267–273, November 1982.
[19] Haykin, S., Neural Networks and Learning Machines, 3rd Ed., Pearson, 2009.
[20] Sanger, T.D., “ Optimal unsupervised learning in a single-layer linear feedforward network. ” Neural Networks, 2:459–473, 1989.
[21] Hauck, S., and Dehon, A., Reconfigurable Computing, Morgan Kaufmann, 2008.
[22] Brown, S., FPGA architectural research: a survey, IEEE Design & Test of Computers, Vol. 13, pp.9-15, 1996.