國立東華大學應用數學系
學 生 演 講

一、 主講人：林家良
講　題：Use “Multiple Discriminant Analysis” for Prediction of Corporate Bankruptcy
時　間：99年6月11日(星期五) 15:10-15:30
摘　要：
在這個報告中，要利用 Multiple Discriminant Analysis (MDA)來作公司破產的預測，內容來自Financial Ratios, Discriminante Analysis and The Prediction of Corporate Bankruptcy，作者為Altman於1968年發表，當中包含如何替 Discriminant Model 挑選適合的變數以便利用這個模型來預測破產，並且去確認這個模型的可行性。Comment: A brief description of MDA is not provided. Is it a statistical method? What kinds of questions can MDA address?”

二、 主講人：張貴雄
講 題：Introduction of (to) The Skew-normal Distribution
時 間：99年6月11日(星期五) 15:30-15:50
摘　要：
The skew normal (SN) distribution is a superset (?) of normal family that … 其特點為何?. In the paper “The Skew-normal Distribution and Related Multivariate Families” by Azzalini, A. (2005) (缺主詞) provide … the univariate SN distribution and some properties. In this presentation, we will understand the definition (?) of SN distribution by using a useful lemma (出自何處? 還是自已提出的? 或是Azzalini 有未盡事宜, 所以…What’s the connection of the lemma with Azzalini), and know what is (change to “are” and move to the end of the sentence) the moment generating function, expectation, variance of SN distribution.

三、 主講人：林紳宏
講　題：Line and Hyper-Plane Fitting ( not coherent!)
Gradient method and Newton method
時　間：99年6月11日(星期五) 15:50-16:05
摘　要：
In statistics, like Regression analysis, to know (how about explore?) the relationship between variables, (缺主詞) use the mean square error to do (Wrong statement! Why not using the lmethod minimizing sum of squared errors) line or hyper-plane fitting (connection with regression analysis is not clear?) and use the concept on iterative approach for modeling the function of data distribution (Why? 前面在說 the relationship between variables, not data distribution). (應寫摘要而非大綱)

References: program of Numerical Analysis (太空泛! What is it? A book or a program (in Matlab or something else?)

四、 主講人：陳思融
講　題：Function Approximation Use Levenberg-Marquardt Learning Radial Basis Function
時　間：99年6月11日(星期五) 16:05-16:25
摘　要：
在做函數近似有許多方法可以應用，而Levenberg-Marquardt Method 和其他方法相較之下為很有效的方法，Radial Basis Function是其中之一(?)。(What do you want to present?)

References:
吳建銘教授教學教材
Soft margins for AdaBoost by G. Ratsch, T. Onoda and K.-R. Muller
(Year? Publisher? Italic font for title. )
上列演講皆於理學院A324會議室舉行
※※※ 歡 迎 參 加 ※※※se990611C