IEICE Technical Committee Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
... (for ESS/CS/ES/ISS)
Tech. Rep. Archives
... (for ES/CS)
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2013-11-12 15:45
[Poster Presentation] Exaluation of Revised IP-OLDF with S-SVM, LDF and logistic regression by K-fold cross-validation
Shuichi Shinmura (Seikei Univ.)
Abstract (in Japanese) (See Japanese page) 
(in English) In this paper, Revised IP-OLDF based on MNM criterion is proposed using a mixed integer programming. The new discriminant function is compared with existed discriminant functions such as Fisher’s linear discriminant function (LDF), logistic regression and soft margin SVM (S-SVM), and it is requested to show the discriminant result is superior to other methods. Four real data such as Fisher’s iris data, Swiss bank note data, CPD data and
student data are used for the development of Revised IP-OLDF. The sample sizes of those are 100, 200, 240 and 40 cases, respectively. Several new facts are found by these data.
And, K-fold cross-validation for small samples is proposed. One hundreds resampling samples are generated from real data by bootstrap method. And Revised IPLP-OLDF is compared with other methods by 100-fold cross-validation. We compare 135 different
discriminant models, and the means of error rates of Revised IP-OLDF are less than others.
In the application research, we focus on the pass/fail determination using four testlets scores as independent variables. Four discriminant models are linear separable among 11 models from 4- to 2-variables. Minimum means of error rates of Revised IP-OLDF, LDF, logistic regression and S-SVM in the validation samples are 0, 9.91, 0.77 and 0.81, respectively. The result of LDF is worst. Next, we compare the difference of the means of error rates of three methods and Revised IP-OLDF in the validation samples. The ranges of LDF, logistic regression and S-SVM with Revised IP-OLDF are [6.23,10.55], [0.39,1.62], and [0.4,1.19]. The worst model of LDF is 10.55 % higher than Revised IP-OLDF, nevertheless logistic regression and S-SVM are 1.62 and 1.19 higher than Revised IP-OLDF. Revised IP-OLDF overestimate in the training samples, but its generalization is the best in the validation samples. LDF’s generalization is the worst.
Keyword (in Japanese) (See Japanese page) 
(in English) Optimal Linear Discriminant Function / SVM / Logistic regression / Linear Discriminant Function / Minimum Number of Misclassifications / Validation of small sample / K-fold cross validation / Linear Separable  
Reference Info. IEICE Tech. Rep., vol. 113, no. 286, IBISML2013-44, pp. 61-68, Nov. 2013.
Paper # IBISML2013-44 
Date of Issue 2013-11-05 (IBISML) 
ISSN Print edition: ISSN 0913-5685  Online edition: ISSN 2432-6380

Conference Information
Committee IBISML  
Conference Date 2013-11-10 - 2013-11-13 
Place (in Japanese) (See Japanese page) 
Place (in English) Tokyo Institute of Technology, Kuramae-Kaikan 
Topics (in Japanese) (See Japanese page) 
Topics (in English) The 16th IBIS Workshop & The 2nd IBIS Tutorial 
Paper Information
Registration To IBISML 
Conference Code 2013-11-IBISML 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Exaluation of Revised IP-OLDF with S-SVM, LDF and logistic regression by K-fold cross-validation 
Sub Title (in English)  
Keyword(1) Optimal Linear Discriminant Function  
Keyword(2) SVM  
Keyword(3) Logistic regression  
Keyword(4) Linear Discriminant Function  
Keyword(5) Minimum Number of Misclassifications  
Keyword(6) Validation of small sample  
Keyword(7) K-fold cross validation  
Keyword(8) Linear Separable  
1st Author's Name Shuichi Shinmura  
1st Author's Affiliation Seikei University (Seikei Univ.)
2nd Author's Name  
2nd Author's Affiliation ()
3rd Author's Name  
3rd Author's Affiliation ()
4th Author's Name  
4th Author's Affiliation ()
5th Author's Name  
5th Author's Affiliation ()
6th Author's Name  
6th Author's Affiliation ()
7th Author's Name  
7th Author's Affiliation ()
8th Author's Name  
8th Author's Affiliation ()
9th Author's Name  
9th Author's Affiliation ()
10th Author's Name  
10th Author's Affiliation ()
11th Author's Name  
11th Author's Affiliation ()
12th Author's Name  
12th Author's Affiliation ()
13th Author's Name  
13th Author's Affiliation ()
14th Author's Name  
14th Author's Affiliation ()
15th Author's Name  
15th Author's Affiliation ()
16th Author's Name  
16th Author's Affiliation ()
17th Author's Name  
17th Author's Affiliation ()
18th Author's Name  
18th Author's Affiliation ()
19th Author's Name  
19th Author's Affiliation ()
20th Author's Name  
20th Author's Affiliation ()
Date Time 2013-11-12 15:45:00 
Presentation Time 180 
Registration for IBISML 
Paper # IEICE-IBISML2013-44 
Volume (vol) IEICE-113 
Number (no) no.286 
Page pp.61-68 
#Pages IEICE-8 
Date of Issue IEICE-IBISML-2013-11-05 

[Return to Top Page]

[Return to IEICE Web Page]

The Institute of Electronics, Information and Communication Engineers (IEICE), Japan