本研究之目的在嘗試將神經網路模型應用於會計盈餘預測。盈餘預測資訊可用來評估公司之獲利能力及真正價值，若能找出相當準確的盈餘預測模型，將有助於管理當局及投資大眾從事各種投資及理財決策。過去許多研究試圖以各種計量方法來找出較準確的盈餘預測模型，然因種種模型及統計上之限制，都未能獲致良好的結果。

神輕網路(Neural Networks) 在人工智慧領域中相當熱門，一般認為最適合用來從事預測及分類的工作。自1982年以來，國外有關神經網路的研究相當多，也已成功地將神經網路應用於許多不同的領域，但在財務上應用的研究則不多。既然大多數的研究均認為神經網路模式優於傳統方法，而且會計上傳統以統計模式進行盈餘預測的研究都未能獲致良好的結果，為什麼不試著將神經網路應用於盈餘預測的研究上呢？因此，本論文以國內上市公司中水泥業的臺灣水泥股份有限公司，自民國六十五年第三季至民國八十一年第四季的季盈餘及年盈餘為資料，嘗試用神經網路來進行盈餘預測之研究。

首先以預測民國七十九年度四季的季盈餘，及民國八十年度的年盈餘為對象，對神經網路的各項參數及架構加以操弄，探討不同的網路參數及架構對盈餘預測準確性的影響，並找出神經網路在盈餘預測上的最佳參數及架構之設定。

其次將最佳神經網路模式的預測結果，與國內外經實證支持的年盈餘及季盈餘預測模式的預測結果相比較，探討神經網路在盈餘預測上的準確性是否優於傳統模式。最後，以預測民國八十年第一季至八十一年第四季的季盈餘，及民國八十一年度的年盈餘為對象，將神經網路的預測結果，與國內外經實證支持的年盈餘及季盈餘預測模式的預測結果相比較，探討所找出的最佳參數及架構是否仍優於傳統模式，以判斷所找出的最佳設定是否為通解。

經實證後可得出下列的暫時性結論：

一、影響盈餘預測準確性的神經網路參數與架構

1 就臺灣水泥股份有限公司的盈餘預測而言，學習率的大小對準確性的影響並不明顯。

2 就臺灣水泥股份有限公司的盈餘預測而言，收斂條件的大小對準確性有影響，較大的收斂條件可能會產生較佳的預測值。

3.就臺灣水泥股份有限公司的盈餘預測而言，三層架構的網路較二層架構的網路可產生較佳的預測值。

4.隱藏層內處理單元的數目，只要有輸入及輸出層處理單元數二者之和的一半，就可產生不錯的預測值。

二、神經網路模式的盈餘預測準確性優於傳統模式

就臺灣水泥股份有限公司的盈餘預測而言，神經網路模式的盈餘預測準確性優於傳統的年及季盈餘預測方法。

三、最佳模式的設計並不是通解值

就臺灣水泥股份有限公司的盈餘預測而言，所找出的最佳神經網路參數及架構設定並非通解值，而是特定解。

一、中文部分
1 王鵲翔，以平行分散處理模式建立股市預測知識庫，國立臺灣大學商學研究所未出版碩士論文，民國七十九年
2 白晉榮，股價預測:專家系統給例學習法之研究，國立政治大學企業管理研究所未出版碩士論文，民國七十八年
3. 李順成譯，計量經濟學理論與應用(上，下冊) ，曉園出版社，民國七十八年
4. 林和譯，混沌:不測風雲的背後，天下文化出版，民國八十年
5. 林維婿，台灣上市公司盈餘預測:時間數列與公司預期之比較暨聯合效益分析，國立政治大學會計研究所未出版碩士論文，民國七十九年
6. 胡玉城，暢談類神經網路，倚天資訊公司，民國八十一年
7. 張維哲，人工神經網路，全欣資訊圖書公司，民國八十一年
8. 黃聰明，自動化知識擷取--神經網路於稅務查核之應用，國立臺灣大學商學研究所未出版碩士論文，民國八十一年。
9. 葉怡成，類神經網路模式應用與實作，儒林圖書公司，民國八十二年
10. 焦李成，神經網路系統理論，儒林圖書公司，民國八十年
11 陳燕慶及鹿浩，神經網路理論及其在控制工程中的應用，儒林圖書公司，民國八十一年
12 徐春美，期中報表預測能力之研究，國立政治大學會計研究所未出版碩士論文，民國六十七年
13.游萬淵，會計盈餘預測之準確性研究，國立政治大學會計研究所未出版碩士論文，民國七十八年
14.楊建民、張碧環、司怡平及蘇佩芬，在微平行電腦上發展類神經網路演算法以預測台灣股市行為，國科會研究計畫報告(NSC 81-0301 -H 000 4-18 ) 。
15.靳蕃、範俊波及譚永東，神經網路與神經計算機原理‧應用，民國八十一年。
16.鄭素鄉，我國上市公司季盈餘時間數列特性之研究，國立政治大學會計研究所未出版碩士論文，民國七十八年。
17.盧炳勳及曹登發合譯，類神經網理論與應用，全華科技圖書公司，民國八十一年。
二、英文部分
1 Albrecht, W.S., Lookabill, L.L. and McKeown, J.C., Time-series Properties of Annual Earnings, Journal of Accounting Research, 1977 , Vol. 15,pp. 226-244.
2 Anderson J.A., Pellionisz A. and Rosenfeld E. eds, Neurocomputing 2:Directions for Research, The MIT Press, 1990.
3. Bao, D.H., M.T. Lewis, W.T. Lin, J.G. Manegold, Applications of Time-Series Analysis in Accounting: A Review, Journal of Forecasting, 1983, Vol. 2, No.4, pp. 437-447 .
4. Blum Adam, Neural Networks in C++ - An Object-Orentied Framework for Building Connectionist Systems, John Wiley & Sons, 1992
5. Brocklebank, John C. and David A. Dickey, SAS System for Forecasting Time Series, SAS Institution Inc., 1986.
6. Brown, L.D. and Rozeff , M.S., Univariate Time-Series Models of Quarterly Accounting Earnings Per Share: A Proposed Model, Journal of Accounting Research, 1977, Vol. 15,179- 189.
7. Cadden, David T., Neural Networks and The Mathematics of Chaos - An Investigation of These Methodologies as Accurate Predictors of Corporate Bankruptcy, First International Confererce on Artifical Inteligence Applications on Wall Street, 1991, IEEE Computer Society Press.
8. California Scientific Software, BrainMaker User's Guide and Reference Manual, California Scientific Software, 1990.
9. Caudill Maureen, The View form Now, AI Expert, June 1992, p24-31
10. Caudill Maureen and Charles Butler, Naturally Intelligent Systems, The MIT Press, 1990.
11 Crooks Ted, Care and Feeding of Neural Networks, AI Expert, July 1992, pp. 36-41
12 Cybenko, G., Continuous Valued Neural Networks with Two Hidden Layers Are Sufficient, Technical Report, 1988, Department of Computer Science, Tufts University.
13. Dopuch, N. and Watts, R., "Using Time-Series Models to Assess the Significance of Accounting Changes, Journal of Accounting Research , 1972,Vol. 10, pp. 180-194.
14. Fant, L. Franklin and Pamela K. Coats, A Neural Network Approach to Forecasting Finacial Distress, The Journal of Business Forecasting ,Winter 1991 -92, pp. 9- 12
15. Foster, G., Quarterly Accounting Data; Times-Series Properties and Predictive-Ability results, The Accounting Review, 1977, Vol. 52, pp . 1-21
16. Gorman, R.P. and T.J. Sejnowski, Learned Classification of Sonar Targets Using a Massively-Parallel Network, IEEE Transactions on Acoustics, Speech, and Signal Processing 36, 1988.
17. Griffin, P .A., The Time-Series Behavior of Quarterly Earings: Preliminary Evidence, Journal of Accounting Research, 1977, Vol. 15 ,71-83 .
18. Harmon Paul, Neural Networks: Hot Air or Hot Technology? Part I, Intelligent Software Strategies, April 1992, Vol. 8, No.4, pp. 1-12
19.-----Neural Networks: Hot Air or Hot Technology? Part II, Intelligent Software Strategies, May 1992, Vol. 8, No.5, pp. 1-18.
20.----- Neural Networks: Hot Air or Hot Technology? Part III, Intelligent Software Strategies, July 1992, Vol. 8, No.7, pp. 1-15.
21 Hawley, Delvin D., John D. Johnson and Dijjitam Raina, Arficial Neural Systems: A New Tool for Financial Decision-MAking, Financial Analysis Journal, Nov-DecI990, pp. 63-72
22 Hecht-Nielsen, R., Theory of the Backpropagation Neural Network, Proceeding of International Joint Conference on Neural Networks 1, 1989, IEEE Computer Society Press, pp. 593-611
23. HertZ John, Anders Krogh and Richard G. Planner, Introduction to the Thoery of Neural Computation, Addison-Wesley, 1991
24. Khanna Tarun, Foundations of Neural Networks, Addison-Wesley, 1990.
25. Kolmogorov, A. N., On the Representation of Continuous Functions of Many Variables by Superposition of Continuous Functions of One Variable and Addition, Dokl. Akad. Nauk, USSR 114, 953 -956.
26. Lapeds A. and R. Farber, Nonlinear Singal Processing Using Neural Networks: Prediction and System Modelling, Technical Report LA-UR -87-2662, 1987,Los Alamos National Laboratory.
27.-----How Neural Nets Work, Neural Information Systems, American Institute of Physics, 1987, pp. 442-456.
28. Leftwich, R. W. and R. L. Watts, The Time Series of Annual Accounting Earnings, Journal of Accounting Research, 1977, Vol. 15, 253-271
29. Lin, Frank C. and Mei Lin, Analysis of Financial Data Using Neural Nets, AI Expert, Feb. 1993, pp. 37-41
30. Marose, Robert A., A Financial Neural-Network Application, AI Expert, May 1990,pp. 50-53 .
31 Nelson, Marilyn McCord & W. T. Illingworth, A Practical Guide to Neural Nets, Addison-Wesley, 1991
32 Neural Ware Inc., Neural Computing, Neural Ware Inc., 1991
33. Rauch-Hindin, Wendy B., A Guide to Commercial Artificial Intelligence - Fundamentals and Real World Applications, Prectice-Hall, 1988.
34. Refenes A. N., Azema-Barac and P. C. Treleaven, Financial Modelling Using Neural Networks, in Liddell H. (ed) "Commercial Parallel Processing", Unicorn, 1993.
35. Ripley, B.D., Statistical Aspects of Neural Networks, in Chaos and Networks - Statistical and Probabilistic Aspects (edit by O.E. Barndorff-Nielsen, D.R. Cox, J.L. Jensen & W.S. kendall), Chapman & Hall, 1993, pp. 1-70.
36. Simpson, Patrick K., Artificial Neural Systems - Foundations, Paragigms, Applications, and Implementations, Pregamon Press, 1990.
37. Stanley Jeannette, Introduction to Neural Networks, California Scientific Software, 1990.
38. Stein Roger, Selecting Data for Neural Networks, AI Expert, Feb. 1993, pp. 42-47.
39. Preprocessing Data for Neural Networks, AI Expert, March 1993,pp. 32-37
40. Surkan, Alvin J. and 1 Clay Singleton, Neural Network Performance in Emulations of Professional Bond Rating Judgements, Proceeding of International Neural Networks Conference, 1990, IEEE Computer Society Press, p394.
41 Tang Zaiyong, Chrys de Almeida, Paul A. Fishwick, Times Series Forecasting Using Neural Networks vs. Box-Jenkins Methodology, Simulation, Nov. 1991,pp. 303 -31 0.
42 Utans Joachim and John Moody, Selecting Neural Network Architectures via the Predictiion Risk: Application to Corporate Bond Rating Prediction, First International Confererce on Artifical Inteligence Applications on Wall Street, 1991, IEEE Computer
Society Press.
43. Varfis A. and C. Versino, Univariate Economic Time Series Forecasting by Connectionist Methods, Proceeding of International Neural Networks Conference, 1990, IEEE Computer Society Press, pp . 342-345.
44. Wasserman, Philip D., Neural Computing - Theory and Practice, Van Nostrand Reinhold, 1989.
45. Watts, R.L., The Times-Series Behavior of Quarterly Earnings, Research Paper, New South Wales: Department of Commerce, Univerity Of Newcastle,1975.
48. Weiss, Sholom M. and Casimir A. Kulikowski, Computer Systems That Learn - Classfication and Prediction Methods from Statistics, Neural Nets, Machine Learning, and Expert Systems, Morgan Kaufmann Publishers, 1991
47. White, Halbert, Neural-Network Learning and Stastics, AI Expert, December 1989, pp. 48-52
48. Wong, Francis and PanYong Tan, Neural Networks and Genetic Algorithm for Economic Forecasting, Technical Report, 1992, Institute of Systems Science,National University of Singapore.
49. Wu Berlin, How to Use Neural Networks in Nonlinear Time Series Forecasting, The National Chengchi University Journal, Vo1 66.
50. Wu, Fred. Y. and Kang K. Yen, Applications of Neural Network in Regression Analysis, Computer and Industrial Engineering, 1992, Vol. 23, pp. 93-95.